CA1036385A - Biological cell analyzing system - Google Patents

Abstract

ABSTRACT

A biological cell analyzing system which is capable of automatically categorizing unstained biological cells as normal or non-normal. The cells are made to flow through a transparent tube in single file and are scanned with a mixture of ultra-violet and visible light. The cytoplasm and nucleus of each cell absorb ultra-violet radiation to different degrees, and the emergent light signal, as modulated by the scanned cells, is detected, amplified, and extended to a data processor which logically analyzes the signal from each cell on a real-time basis. The visible light signal is subtracted from the ultra-violet light signal to improve the signal/noise ratio of the latter, and to automatically cancel out non-biological debris.
A number of acceptance tests are electronically performed on each cell, and if any of the tests is failed the cell is cate-gorized as non-normal. Ambiguous conditions, resulting for example from the clumping of cells, are identified and sepa-rately counted. The system can process up to several thousand cells from a single sample during a one-minute run.

Description

10363~3S
Thi~ application i9 a division of Canadian Application 153,963 filed October 16, 1972.
~his invention relato3 to ~stems for i automatically analyzing the chemical and morphological characteristics of biological cells, and more particularly to a system which is capable of analyzing unstained cells at a rate up to several thousand per Minute.
; . :
For many years biological cells have .. . . . .
been analyzed by first staining and then visually inspect-;' ing them under a microscope. The cytoplasm and nucleus .. . . . .
-of a cell appear differently when properly stained and i 10 a variety of objective criteria have been formulated to , determine whether a cell is abnormal. The major dis- ' advantages of the conventional approach to the examination 'l ,, of biological cells are that the cells must first be stained ~', and isolated, and then visually examined by a cytotechnologist , or a pathologist on a coll-by-cell basis. Since the vast ' majority of randomly obtained cell samples are normal, the.. . . .
~ task of ma~s screening is time con~uming and costly.
.. . .
: ~ - . . . . - . . .
-~ In an effort to automate the anal~sis of ' biological cells, various schemes have been proposed. Some:: -. . . .
of these schemes entail the scanning of cells by irradiating them with ultra,-violet light. Ultra~violet light is absorbed , to a much greater extent by the D~ and R~ in the nucleus , '' ... .
of a cell than by the cytoplasm of a cell. The ~ltra-violet ' light a~sorption in the two regions of the cell can ~e used , to characteriæe the cell in accordance with various criteria ; as is known to those s~illed in the art.
, : -2-... .
~, ,, ,.. ; - ... . . _ .. ..... _ . . _ . . .. . . .... . .
~: -10363~5 For example, the total absorption is a measure of the size of the nucleus, an enlarged nucleus often being indicative of a cancerous cell. Examples of such prior art systems are disclosed in Ramentsky Patent Nos. 3,327,117 and 3,327,119. In general, a shortcoming of the prior art automated scanning sys-~ tems is that cells must still be handled individually, e.g., h" they must be placed on slides prior to examination. Also,the staining of cells is subject to non-uniformities in both the stain itself and the uptake per cell, and the overlapping of ; 10 cells makes it very difficult to anaylze them on an indivi- ;
.
dual basis.
It is a general object of our invention to provide ;., 1!
a system which analyzes unstained biological cells on a cell-by-cell basis at a very high rate and which does not require ~',, s ~;, the placing of cells on slides prior to scanning, thereby ; making mass screening feasible.
Accordingly, the present invention provides a system for analyzing biological cells comprising a scanning station, means for controlling the movement of biological cells in essentially single file past said scanning station, means for scanning said cells at the scanning station with line scans : ~.
~' of light, means for detecting the instantaneous intensity of the light scanning said cells at the scanning station for deriving a signal proportional thereto, means for distinguish-ing between nuclear and cytoplasmic cellular material along any line scan through a cell in accordance with the derived signal, and means responsive to said distinguishing means for ; categorizing a cell in one of at least two groups.
~: O
As suggested in the prior art, the present invention is predicated on the facts that the major portion of the DNA
~: .
` contained in a cell is in its nucleus and that DNA (when un-stained1 absorbs ultra-violet light to a much greater extent than do the other constituents of the cell. Individual cells . ~
; in the system of a described embodiment of the invention are r .. :. ~ ~ , .

`` ` ~0363~35 scanned on a line-by-line basis, in a manner similar to the generation of a television signal.' One of the problems which may ~e encountered with the scanning of a cell with ultra-violet light, however, i5 that conventional sources of ultra-violet radiation do not emit radiation with a con5tant intesity - -- there is generally a considerable amount of noise in the ultra-violet light output of the source. ~he logic circuits which anal~ze the ultra-vioIet signal after it is modulated b~ ¦
a G~ll may erroneously treat noise in the origina~ source out- I' put as cellular information. Similarly, debris s~ch as food particles, blood clots, etc. associated with the cells may moclulate th'e ultra-violet signal although not representinq true' celLular information.
: , ~ , - -. ,, :~.
It is an object of a described embodiment of our invQntion to improve the signal/noise ratio of a signal which is derived by scanning cells with~ultra-violet light and to `' ' prevent non-cellular debris from being considered as cellular r:
material.
~lthough thQ nucleUs and cytoplasm o the cell 2a. a~or~'ultra-vi~let light to different degrees and this has been us~d in th~ prior art to distinguish between nuclear and cytoplasmic material, in some cases the relative absorption - '-characteristics are not in themselves sufficiently different -.
to allo~ differentiation between the two regions of a cell.
.
- - Another object of a described embodiment of our invention is to provide a mechanism for more clearl~ di~ferent-iating between the nucleus and cytoplasm of a cell with the use of ultra-violet light.
, It is also important in may cases to determine 3Q ~hether the cells being examined are deep lung cells. Lung cells can ~e obtained from a patient by having him cough i .

~' `"` ~0363~5 .
deeply into a collection kit, such as the Cytec (Trademark) sputwm collection kit marketed by the assignee of this application. It is to be expected, however, that some persons will simply spit into the collection kit without firs~ coughing deeply; in such a case, the cells collected will not be lung cells. It is very often desirable to verify that lung cells are =2 being examined, which is usually done by veriying that a spec-if ic type of cell -- the histiocyte -- is contained in ~he specimen.
laIt is another object o a described embodiment of ou~ invention to provide a system which is capable of automat-ically characteri~ing a cell a~ a histiocyte.
In any system designed to examine automaticall.y ~
cells on an individual basis, it is necessary to detect "ambig- -uous" conditions. An ambiguous condition often arises when cells cling together in a clump. The various tes~s per~ormed on the cells may erroneously categorize a clump of cells or -debris as a non-normal cell when instead the scanned biologiaal material should not be categori~ed as either noxmal or non-normal since a single cell has not been ~xamined.t It is another object of a described e~bodiment of our invention to detect ambiguous conditions ~cell clumps, }
organic debris, etc.) and prevent their categorization as normal or non-normal cells.
In general, it will be appreciated that the tech-niques of automated cytology to be described herein can be used in many other applications, e.g., to distinguish between the various types of white cells in blood, to examine cervical cell samples, e~c.
3~Before considering the illustra~ive embodimen~

_5_ : ~

' ' .....

r~ 36 3 85 - of the present invention, it is necessary to understand the use of the terms "normal" and "non-normal" as used in the specification. Cells are usually categorized ~y a pathologist as normal or abnormal. However, there is a "gray" area separating the two regions defined by these antithetical terms.
It is exceedingly di~ficult in some aases even for a trained pathologist to properly categorize a cell when it falls in the gray area. It is contemplated that the greatest advantage of an automated system of the type ~
~Q considered herein is that cell sam~les from vast numbers of persons can be accurately screened at relatively little cost. The system is designed to categorize a cell 'I

.. : . - .,...... . ~. 1 `, ~ . - ' . / ' ' i - - ~- / :: - .

' ' ' '' ' :' / ' ' ' '' -''' ', / " , . ~ .

. . ~ , . ~ . ,.
~, . . ~.
: . . , !
: .

'. ' - ' " i ' . ~ . / , - . : , '' ' / '', '`, ,` '' ' '. / ' ' ' ''' ' '' ' / ' 3~, /

j,.
' -- ., . . , I
.

1036;3 85 as "normal" if it is indeed clearly normal. Any cell falling into the gray region, that is, any cell which i8 not clearly normal, is categorized as "non-normal". Thus, abnormal cells are cate-gorized as "non-normal" as are cells which may be normal but are not clearly so. This latter class may be loosely referred to as "suspicious" or , "aeypical", In order to examine cells on an individual basis, it is necessary to 1solate them.
Rather than placing the cells on slides, in accor-dance with the principles of our invention, they .. , . . . ~
are made to f}ow in an essentially single file through a capillary tube ha~ing an inner diameter in the order of only 90 microns. Typical normal lung cell diameters are 40-50 microns, and by causing the cells to flou ~in a carrier solution) through a small-bore capillary tube, it is possi-ble to control an essentially single file flow past an optical scanning station. (In general, the inner diameter o~ the capillary tube is approximately twice as large as the diameter of a normal cell of the type being examined.) -. .~ ..

~o36a~s ~ :
The sample obtained from a patient i8iconverted into a homogeneous dispersion of cells in an alcohol-base carrier solution. In the case of the Cytec (Trademark) sputum collection kit, the cells are received .~. s from the patient in the form of a mucal clump. They can -;
be broken up mechanicaIly (for example, with the use of a blender) and then filtered. The carrier solutuion is 50 water and 50% alcohol. Techniques for obtaining a filtered . : .:
homogeneous dispersion of cells from such samples are well ~ ~
: ;: ; .
known to those skilled in the art.

As the cells flow in single file through , ,, . : .
~ ~ the capillary tube, successive transverse lines are scanned .:
~ - ~ with a mixture of ultra-violet and visible light derived :, ~ ~ . ,.
from a common source. The modulated light signal~is spectrally separated and extended to~respective photo~
multipIier tubes which produce electrical signals corres~
ponding to the degree of modulation of the respective light wavelengths during each line scan. The ultra-violet sig-; -nal is the prlmary source of cell information in that it .
20 ~ ~provides a measure of the bounds of nucleus and cytoplasm along~each llne ~scan~.~ However,~the vlslble light signal s~electrically subtracted from the ultra-violet signal to~ lmprove the slgnal/nolse ratio.~ The correlation of ~ noise in the two common-source signals is great and con-s~ sequèntly by subtracting one from the other, it is possible to minimize the resulting noise in the combined slgnal.

,:
:, :
: ..
~ ~.
.;~, slubtra The ction proces~ also filters out non-cellular debris because the ultra-violet light and visible light are absorbed to approximately the same extent, resulting in a net cancellàtion. The subtraction of the vi~ible light signal from the ultra-violet signal serves still a third function. Although the nuclear and cyto-plasmic regions of a cell can be distingu1shed by their different ultra-violet light absorption characteristics, in some case~ the diference may be too small to allow the differentiation. By subtracting the visible light signal from the ultra-violet light signal, a much greater difference is aiscernible in the portions of the resulting signal - ~
. ~ :
corresponding to nucleus and cytoplasm. This is due to the fact that the visible light absrption is only slightly ~
greater in the nucleus than it is in the cytoplasm, 80 that the subtraction of an approximately constant signal from the ultra-v~olet light signal results in an enhinced ~ . .
;~ nucleus-to-cytoplasm discernibility. '~ -.
, . ~
The visible light signal is also used to - determine whether a particular cell is a histiocyte.

Histiocytes contain a considerable amount of carbon and, : . , . .
therefore, absorb visible light much more than other types of cells. Depending on how much of the visible light is ` absorbed, a cell can be characterized as a histiocyte.

The signal resulting from ea d line scan through a cell is processed in real-time. After a cell has been completely scanned, the results of the various tests performed on the individual line scans are used to .
categorize the cell.
: _g_ . -~; . ~ .. , . . .

The final decisions are tallied in counter banks, and may be displayed on television-type monitors as well as recorded for subsequent study.

Although the system of the invention is designed to cause cells to flow in a single filé through a transparent capill~ry tube, it haQ been found that it is not always possible to prevent the clumping of groups of cells together. In the case of a cell clump, it is desirable to inhibit the categoriæation of the biological material as either a normal or non-normal cell. For this reason, in the illustrative embodiment of the invention, a separate count is maintained of "ambiguous" conditions, along with the counts of "normal" and "non-normal" cells. The ambiguity logic is predicated on too many line scans detecting cellular material in succession.

If too many line scan~ in succession all result in the detection of cellular material, it is an indication that a group of cells may be clumped together, In the normal case where individual cells flow through the capillary tube, many line scans take place between successive cells since the cells do not immediately follow each other.
The line scans in succession which result in the detection of cellular material are counted, and if the number exceeds a predetermined value it is an indication that a clump of cells may be present. At the same time, however, it is possible that what has been encountered is an abnormally large cell, which cell also results in a very large number of successive line scans verifying the presence of cellu-lar material.

. .. _ .. .. . . . . . . . , , . . _ ........ .. , .. . . . ~ . ... . . . .. . .. . . .. .. .. .
.. . .

10363~5 For this reason, the total "ambiguous" count at the end o the scanning o a sample may be indicative of an exces~ive number of abnormally large cells, or of inadequate aell processing which resulted in clumping. At the end of the ~ scanning of each sample, the total "non-normal" and "ambiguous"
- counts are recorded, together with the total number of indi-vidual cells detected (where a cell clump i8 counted as an indi~idual cell)~ If the total number of non-normal cells or the total number of ambiguous cells is too high for the particular total cell count, it is an indication that the sample requires further study.

The line scan signal, in addition to being operated upon by a data processor to determine cell characteristics, is also used to form a display on a TV monitor, the display actually showing the silhouette of cells flowing through the capillary tube. The display allow~ various adjustments to be made by the operator, e.g., the focusing of the light raster through the center of the capillary tube, as well as visual verification of air, intermittent flow, or clogging of the tube whenever such conditions are present.

The signal can also be recorded on video tape. The tape can subsequently be played back and the signal can be fed to the data processor in lieu of an ` actual cell scanning signal. In this manner, previously recorded cells can be observed at the same time that the data processor evaluates them again.

.. .. ... . . .. . . . . . . ..
. .
~'' - ~ ' ' . ' ' ~ ' ' ;:

` 1036385 Various threshold levels in the data processor can be adjusted at this time to change subsequent categorizations, or repeatability tests can be performed to see if the same input data produces the same category counts.

An "annotation" signal is developed which affords a visual presentation of the categorization of each cell in accordance with a pre-set code. The annota-tion signal is made to appear on the TV monitor display following each cell a~ it flows through the capillary tube.
The anpotation signal may also be recorded on the video tape 80 that a convenient record may be made not only of the silhouette of every cell, but of the decisions made by the data processor with respect to it.

An electronic cell simulator is pro-vided for generating an artiicial cell scanning signal.
The ~imulated signal is u~eful, for example, in adjusting various threshold levels in the data processor and for tracing malfunctionc in the system.

In the illustrative embodiment of~the invention, five distinct tests can be performed on each cell. These tests relate to the size of the diameter of the cell nucleus, the size of the cytoplasm shoulders on - either side of the nucleus, the ratio of the sizes af the nucleus and the overall cell, the symmetry of the cytoplasm shoulde~s on either side of the nu deus, and the product (integral) of the size and density of the nucleus.

. . - -: . - - . . . .
. : . .
, ::
, --:

$, -- 1036~85 - Each of these tests may be ~elected to be operative or not, as the operator chooses. When active, each test is applicable to every line in every raster. -An embodiment of the invention will now be des-cribed, by way of example, with referenae to the accompanying drawings in which:
Fig. 1 depicts, in block diagram form, an illustrative embodiment of the invention;
Figs. 2~and 3 are curves representing relative celluLar absorption as a function of wavelength and relative scanning lig~t intensity as a function of wavelength Figs. 4 - 25 depict in detail the system of Fig. l; -,.
Fig. 26 shows the arrangement of Figs.
. - .
4 - 25;
- Fig. 27 depicts a circuit which can be used in - the illustrative em~odiment of the invention in lieu of :
some o the circuitry shown on Fig. 15; and -Fig. 28 depicts the form of the TV display of flo~Lng ae~ls which can be derived in the illustrative em5odiment o the invention. ~ !
~he detailed description of the illustrative l -embodiment of the invention will be presented in accordance . With the following index:
' ', , ' . ' , .

', .

~, ~
.

Block Diagram of the System -- FIG. l............... Page 14 Scanning System -- FIGS. 2 - 7...................... 19 Video Conditioner -- FIG. 8......................... 43 Electronic Cell Simulator -- FIG. 22................ 48 Recorder Monitor Display -- FIG. 28................. 57 Data Processor -- FIGS. 9 - 21...................... 62 Recorder and Playback System -- FIG. 23............. 122 Start/Stop Control, Counter Bank and Data Acquisition System - FIG. 24.................. 126 Alternate Non-~ormal Logic -- FIG. 27............... 130 Automatic Sample Handling System -- FIG. 25......... 132 Block Diaqram of the SYStem -- FIG. 1 The individual blocks of equipment in the system will be described in detail below. FIG. 1 will facilitate an understanding of the way the individual blocks i of eguipment relate to each other. There are seven ma~or ; blocks of equipment on FIG. 1, all interconnected by various cables. Each block of equipment is shown in detail on other figures. For example, video conditioner 70 is shown on FIG. 8, while the data processor is shown on FIGS. 9 - 21.
The same numerals whlch are used for identifying the inter-connecting cables of FIG. 1 are used in the remaining figuFes Automatic sample handling system 30functions to mechanically convey prepared samples of different persons and to automatically cause cells from successive samples to flow through guartz capillary tube 46.

~: .

; ~ ~' ~ ' ' ' ''- ' .. . .

10363~5 Each sample requires a total time of two minutes for pro-cessing. Start/stop control 37 in block 86 (start/stop control 37, counter bank 40, and data acquisition system 42 are shown within one block because they are all shown in detail below on a single figure, FIG. 24~ applies pulses over cable 61 at two-minute intervals to the auto-matic cell handling system. The cell handling system reguires 30 seconds for changing samples be~ore cells from the new sample start to flow through the capillary tube.
Once cells start to flow, 30 seconds are allowed so that the cells from the previous sample are swept or cleaned out of the capillary tube. During the next 60 seconds of the run, the cells flowing through the tube are scanned, evaluated, and the resultant data is stored in counter bank 40. At the end of the 60-second data collection run, the next pulse on cable 61 causes the automatic cell-handling system to change samples. ~t the same time, the pulse controls data acquisition system 42 to permanently record the pertinent data concerning the sample just run.

20 - Scanning and monitoring system 80 includes .
flying spot scanner 31. This scanner generates a raster which is a mixture o~ ultra-violet and visible light, shown by dotted arrow 48. The light passes through quartz optical system 32 and quartz capillary tube 46 to two photo-multiplier tubes included in light box assembly 34. The photomultiplier tubes develop two signals on conductors 93 and 94.

, . . . . . .

~036385 One signal is proportional to the intensity of the ultra-violet light after modulation by the flowing cells. The other signal is prop~rtional to the intensity of the visiblé
light after modulation by the flowing cells. The cells themselves are drawn through the quartz capillary tube by the vacuum created by pump 35. The vacuum can be measured by gauge 97, and the cell flow rate adjusted accordingly.
After analysis, the cells are collected in waste cell reservoir 54 which is emptied periodically.

The two signals on conductors 93 and 94 are extended to video conditioner 70. The video conditioner processes the two signals to develop a "processor video"
signai which is extended to the data processor over con~
ductor 52. This signal is a combination of the ultra-violet and visible light signals, combined in such a way as to reduce the effects of noise. Amplified and normalized versions of the ultra-violet and visible signaln are trans-mitted respectively over conductors 53 and 50 to the data processor because certain operations are performed on these signals alone. The data processor extends a signal over conductor 51 to the video conditioner which gates the sig-nal on conductor 52 off at all times, except during the time that a line scan is within prescribed limits (72 microns) within the bore (90-micron diameter) of capillary tube 46.

The data processor analyzes each cell based on the signal transmitted over conductor 52.

.
.' ` ' ' , ... .. . . . . . . . . . .. ... .... .. .. . .. .. .. . . ~ . . . . .. . . .. . . .. .

~Oa63t~5 But in addition to having the data processor process a signal derived from the actual scanning of cells, it is possible to simulate the signals obtained from cells.
Electronic cell simulator 36 includes circuitry for deriving a signal on conductor 72 to simulate a cell for the data pro-cessor to analyze. Cell simulation is advantageous, for example, in setting various decision threshold levels in the data processor. ~ cell can be simulated to have certain characteristics which should control a certain logic decision by the data processor, and the appropriate threshold levels can then be set as the data processor operates on this 8ig-nal. The timing of the-electronic cell simulator and the data processor are keyed to the sy w signals appearing on cable 84.

The data processor transmits signals over cable 81 to the recorder and playback system 41~ The video tape recorder allows cell signals to be played back in the event it is desired to examine a particular cell in detail. -.
The recorder and playback system is also capable of trans-, mitting a recorded video signal over cable 88 back to the data processor input. It thus becomes possible for the data processor to categorize a cell whose characteristics have previously been recorded. This is advantageous for a number - of reasons, for example, it facilitates the tracing of sig-nal flow and malfunctions within the data processor while operating on system-derived signals.

:
:

-17~

. .
. ' .

1036~ 5 The ultra-violet signal transmitted over cable 81 to the recorder and playback system i8 also trans-mitted to microscope monitor 700 in the scanning and moni-toring system. This allows the operator to observe the cells flowing through the capillary tube.

The scanning and monitorlng system also includes a system oscilloscope 714. The processor video signal on conductor 52 is extended to the signal input of the oscilloscope. The X and Y sync pulses are extended from the data processor to the sync input of the oscilloscope on conductor 132. The processor video signal is the princi-pal input to the data processor: it is, therefore, desirable to provide an oscilloscope for the purpose of monitoring it.

As each cell i8 categorized by the data processor, appropriate pulses are transmitted over cable 69 to counter bank 40 and data ac~uisition system 42. The counter bank consists of a set of counters ~hich are incremented depending upon how each cell is categorized. Cable 75 is extended from the counter bank back to the data processor to stop further pulsing of the conductors in cable 69 in the event any counter in the bank has reached its maximum count before the one-minute run of the sample under test has been , completed.

The counter bank data is extended over cable 64 to data acquisition system 42.

:' .

..

Similarly, certain informatlon from the data proces~or i8 extended directly over cable 69 to the data acquisition sys-tem. When start/stop control 37 pulses cable 61 in the appro-priate manner at the end of each run, the data acquisition sy~tem records all of the data pertinent to the sample whose run has just been completed. Thereafter, the start~stop con-trol resets the counter bank prior to the data processor operating on the next sample. Conductor 87 is provided to allow the counter bank to inform the start/stop control mechanism in the event a run has been completed and an insufficient number of cells have been counted. In such a case, a lamp circuit iq energized to inform the operator that either the capillary tube has become clogged, or the sample . .
is too sparse of cells.

Scanninq SYStem -- FIGS. 2 - 7 The timing ~ignals which control the scanning of the cells flowing through quartz capillary tube 46 are generated by the circuit on FIG. 4. As will be described below, it is possible to record an image of the flowing cells on a video tape, and to thereafter play back the tape and feed into the data processor the video signal rather than .. . .
a signal which is derived during actual scanning of the ~
... .
- - cells. It is also possible to electronically simulate the flowing of cells through the capillary tube, for example, in order to adjust various threshold levels in the data pro-:
cessor. During playback, switch 90a is connected to play-; back terminal 90c.

. .
.

. .

:, ~ ~ , ' . ' ' .
. , ~.0363~5 During the scanning of actual cells, at which time the video signal is recorded and during the electronic simulation of flowing cells, the switch is connected to operate terminal 90b as shown in FIG. 4.

The various gates shown in the drawing are of the type in which a logical "1" is represented by a positive potential, and a logical "0" is represented by ground potential. However, if any input to a gate i~ left floating, the gate operates as though a positive potentiai ~logical 1) were applied to this input. It should also be noted that various inputs to individual gates or other cir-cuits are provided with a little circle, and that similarly some of the gate outputs are provided with such a circle.
This conventional notatlon i8 used to represent an inverting operation.

' With switch 90a in the position shown in FIG. 4, ground potential appears on terminal 90b. This termi-nal is connected through an inverter to one input of AND
gate 196. Thus, one input to the gate is energized. Free-running multivibrator 190 generates CLOCK pulses on conductor191. As shown in FIG. 4, the multivibrator cycle is adjusted to a period of 63.5 microseconds. The width of each positive CLOCK pulse is not important insofar as the derivation of the timing signals is concerned. Each positive pulse on conductor 191 is transmitted through AND gate 196 to one input of OR gate 197.

~` ~ ` "'' ' :

The output of the OR gate is extended to the input of one-shot multivibrator 198. This multivibrator is triggered by a negative step applied to its inverting input; therefore, at the termination of each CLOCK pulse, the multivibrator qenerates a pulse at its output. Each pulse on conductor 113 is 5 microseconds in width and is identified throughout the drawing as an X-SAMPLE pulse.

Each X-S~MPLE pulse is applied to the invert-ing input of one-shot multivibrator 104. This multivibrator is ~imilarly triggered by a negative step and generates another 5-microsecond pulse. This pul~e, as shown directly below the CLOCK and X-SAMP~E pulse waveforms in FIG. 4, is referred to as the ~-RESET pulse.

' The X-SAMPLE pulse~ are applied to one input of NOR gate 106, and the X~RESET pulses are applied to the other input of the NOR gate. The output of the NOR gate i8 an inverted "OR" function, and thu~ X-SWEEP GATE conductor 107 is ordinarily high when the two inputs to gate 106 are - -low. However, for the 10 microseconds that an X-SAMPLE
pulse i9 generated followed by an X-RESET pulse, one of the two inputs to the NOR gate is high and thus conductor 107 is low. The resulting waveform on X-SWEEP GATE conductor 107 is as shown in FIG. 4. The conductor ordinarily is high in potential, but goes low for 10 microseconds in every 63.5~
microsecond cycle.

~21-: ' '; '` ' " ' ~ ' ' :' -' -- , , ~ .. - . . , , .: .

The input to integrator 108 is a constant potential provided by source 110. The output of the inte-grator on conductor 111 is the integral of the input, that is, a ramp. However, the integrator functions to derive a ramp on X-SWEEP conductor 111 only when X-SWEEP GATE conductor 107 is high in potential, When the conductor is low in potential, the output of the inte~grator drops to ground potential. Thus ; as shown in the X-SWEEP waveform on FIG. 4, during the 10 microseconds that X-SWEEP GATE conductor 107 is low, the integrator output iB at ground potential. As soon as X-SWEEP
GATE conductor 107 goes high, the voltage on X-SWEEP conductor 111 rises in the form of a ramp. At the end of any cycle, when the X-SWEEP GATE conductor goes low, the voltage on conductor 111 drops to ground and remain~ at this level for the 10 microsecondo that the X-SWEEP GATE signal is low.

The three signals on conductors 111, 113, and 114 are used throughout the system for controlling many functions. One of the primary functions of the X-SWEEP sig-nal is to control scanning of the cells in the X direction (across the capillary tube). A complete line scan takes place during the generation of each ramp signal. The X-SAMPLE and X-RESET pulses which alternate with each line scan are used to control various logic operations.

In order to scan an area, it is also desirable to generate a Y-SWEEP signal, as in a TV-type raster. The Y sweeps take place at a rate of 60 per second.

' -~2-.. . . . . .. . . .... . . .

10363~35 As shown in FIG. 5, line source 120 i5 connected to ~ne input of comparator 121, the other input being grounded. The comparator operates to cause i1:s output to go high whenever the line voltage is positive. Thus, the output o~ the comparator is a square wave whose ~requency i5 the same a~
that of the line source. The square-wave signal is applied to operate terminal 129b, and when switch 129a is in the position shown, the signal is applied to the set input of flip-~lop 122. (Switches 90a, 129a, and 288a, respectively 10 on FIGS. 4, 5 and 8, can be ganged together for convenience.) Each positive step in the signal causes flip-flop 122 to be placed in the 1 state with output Q going high and output Q going low. With the flip-flop in the 1 state, conductor 128, connected to one input of AI~D gate 123, goes high.
; Conductor 113 is connected to the other input of the ANI) gate, and thus whenever flip-flop 122 is in the 1 state, the X-SAMPLE pulses on conductor 113 appear at the output of A~D
gate 123. -.
Eight-stage counter 124 is initially in the 20 0 state. Each X-SAMPLE pulse applied to its count input causes the count to be incremented. The counter cycles from 0 to a maximum count of 28-1, or 255, and then back to 0. The eight outputs of the counter are extended to the inputs of digital/analog converter 126. The converter develops an analog signal on Y-SWEEP conductor 127 which is proportional to the binary count represented in counter 124. Thus, the Y-SWEEP signal is a staircase with 256 steps. It is this signal which is used to control the generation of the raster in the Y direction (along the capillary tube~.

~. .
. .. ... ~ . ...... . . .. . ...... ....

:: :
:,: : ..
-. :

1036~35 The eight outputs of counter 124 are also extended over cable 133 (included within cable 84) to other circuits which require the encoded counter states for identi-fication of the particular raster line which is being generated at any given time.

The eight outputs of the counter are also coupled to the eight inputs of decode "0" circuit 125. This circuit detect~ a transition from the maximum count (255) in the counter to the minimum count of 0, and when the counter 10 represents 0, the output of circuit 125 goes high. The output is extended to the reset input of flip-flop 122 and causes the flip-flop to be reset in the 0 state. At this time, conductor 128 goes low and no further X-SAMPLE pulses are transmitted to the count input of counter 12~. Counter 124, fli-flop 122, and gate 123 all remain in the 0 state until the next high level output from comparator 121.

Flip-flop 122 is set in the 1 state by comparator 121 sixty times each second. Since X-SAMPLE
pulses are generated at 63.5-microsecond intervals, in any 20 period of 1/60 seconds the number of X-SAMPL~: pulses which are generated is (1/60)/(63.5) (10 6) Thus, during any period of 1/60 seconds, 262 X-SAMPLE pulses are generated.
Only 256 of these pulses are required to cause counter 124 to start incrementing from an initial count o 0 through the maximum count, and then back to 0. After these 256 pulses have been generated, the decode "0" circuit 125 operates to reset flip-flop 122.

:-... ~ .. . ....... .. ~ ............. ...
..

.. .. ~ . . .= ... . ~

`" 10363E~S
After flip-flop 122 i~ reset, six X-SAMPLE pulses are generated which have no effect on counter 124. Similarly, for ~he first 256 X-SAMPLE pulses which are generated in any group of 262, the signal on Y-SWEEP conductor 127 ri~es in 256 successive steps. ~s soon as the counter switches to a 0 count, the output of converter 126 drops to ground, and Y-SWEEP
conductor 127 remains at ground potential until flip-flop 122 is once again sat in the 1 state.

The X-SAMPLE conductor 113 is extended 10 to one input of OR gate 131, and the Q output of flip-flop 122 is connected to the other input of this gate. During the Y-SWEEP flyback time, that is, while conductor 127 is low in potential, the Q output of flip-flop 122 is high, -and thus conductor 132 is high in potential. During the generation of the Y-SWEEP signal, the Q output of the flip-flop is low in potential, and one input of OR gate 131 is de-energized. However, each time an X-SAMP~E pulse i8 generated, the other input of the OR gate is energized and conductor 132 goea high in potential. Thus, the sigr~al on 20 conductor 132 is a composite synchronizing signal. In each .; .. . .
1/60-second cycle, 256 X-SAMPLE pulses appear on conductor 132, followed by a much wider pulse which represents the Y
flyback. Conductor 132 is extended over cable 84 to the data processor.

Actually, in the analysis above, it has been assumed that in each perioa of 1/60 seconds, 262 X-SAMPLE pulses are generated.

10363~5 This nuniber was determined by dividing the Y-SWEEP period of 1/60 seconds by 63.5 microseconds. The division results in a number slightly in excess of 262. It will be apparent that in some cases 263 X-SAMPLE pulses are generated rather than 262 during one complete cycling of flip-flop 122.
Whether 262 or 263 pulses are generated depends on how soon the first X-SAMPLE pulse occurs after flip-flop 122 switches to the 1 state. It makes no difference whether 262 or 263 pulses are generated. As will become apparent below, what 10 i~ important is that the data processor be able to acknowledge the completion of any raster scan. The test to be described below is that at least five X-SAMPLE pulses be generated without the aetection of any cellular material flowing through the quartz capillary tube. At the end of each Y-SWEEP, the scanning ceases for the Y flyback period ànd there is obviously no detection of cellular material during this interval. Since the scanning terminates at the end of 256 X-SAMP~E pulses in any period of 1/60 seconds, and there are either another six or seven X-SAMPLE pulses 20 generated before the Y-SWEEP staircase signal once again .
begins, the data processor is capable of recognizing the completion of a complete raster scan.

Thus far the X-SAMPLE, X-RESET, X-SWEEP
and Y-SWEEP signals have been described as being generated unaer control of free-running multivibrator 100. This is the case when cells are being scanned or electronically simulated.

.

.
, . . . ;
.. ..
... :
: , , .: .
., ~ .

But when the video tape i9 being played back and the system is used to analyze the cells represented by the video sig-nal, it is necessary to key the system operation to the recorded video sync signals. The composite video sync information tin the form of inverted X and Y sync pulses) which is initially recorded is developed on conductor 132, and to properly key the system operation to the video sync information, it is necessary to use this pulse train to control the generation of the X-SAMPLE, X-RESET, X-SWEEP
and Y-SWEEP signals.

During playback of the video signal, switch 90a in FIG. 4 is connected to terminal 90c. Terminal 90b ; is left floating and while ordinarily a floàting input is ` -treated by a gate as a 1, because terminal 90b is connected through an inverter to an input of gate 196, the effective input is a 0 and the gate does not opérate. Instead, the ground potential on terminal 90c is extended through tvo inverters to respective inputs of each of gates 194,195.
The video playback signal, as will be described beIow~ is extenaed over conductor 88 to the input of X-sync stripper 192 and Y-sync stripper 193. Each of these units extracts the respective sync pulses. The X-sync information, namely, the originally recorded X-SAMPLE pulses, is extended to the other input of A~D gate 194 whose output is coupled to the -second input of OR gate 197. Thus, each inverted X-SAMPLE
pulse recorded on the video tape triggers one-shot multi-vibrator 198 just as does each CLOCK pulse when switch 90a is in the OPERATE position.
.

... . . .. . .. .. . . . . .. . . .. . .. . ... .. . . . . ............ . .. . . . . .. . . ..

, .

10363~95 Each recovered inverted X-SAMPLE pulse causes a new 5-micro-second X-SAMPLE pulse to be generated, followed ~y a 5-micro-second X-RESET pulse.

The Y-sync pulses at the output of Y-sync,stripper 193 are coupled to the second input of gate 195, and inverted Y-sync pulses thus appear on conductor 115.
This conductor is extended to playback terminal 129c, to which switch 129a is connected in the playback mode. It is necessary to sync the Y-SWEEP signal on conductor 127 to the Y-sync information recorded on the video tape. At the termination of each Y-sync pulse, the positive step which appears at the set input of flip-flop 122 causes the flip-flop to switch to the 1 state. Gàte 123 is once again enabled and the X-SAMPLE pulses on conductor 113 are trans-mitted through the gate to cycle the counter. Once again, when the counter cycles back to 0, decode "0 aircuit 125 detects thi~ condition and reset~ ~lip-flop 122 80 that the counter remains in the O state and the Y-SWEEP signal iB held at ground potential.

The X-SWEEP and Y-SWEEP signals on respectiYe conductors 111 and 127 are extended to flying spot control circuit 96 on FIG. 6. This circuit develops the necessary signals to control the operation of flying spot source 82.
The flying spot source incorporates a conventional flying spot scanner 83 which includes a cathode ray tube coated with a special phosphor which emits a substantial amount of light in the far ultra-violet (W) region. Some visible light is also produced along with the ultra-violet light by this phosphor.

.. . . . . . . . . . . . . . . . . . . .. . ... . .. . . . . .
... , . ~ , ~
. ~
, 1036;~5 A number of different phosphors can be used, among them, aluminum oxide, beryllium silicate and calcium borate-lead.
In the illustrative embodiment of the invention, the phos-phor which is coated on the face of the cathode ray tube i5 a beryllium silicate. The light emission pectrum of this phosphor is of the form shown in FIG. 3.

The curves of FIG. 2 show the relative absorption of D~ and RN~ a~ a function o~ wavelength. As i8 known, most of the D~A of a cell is in its nucleus, while the RN~ is distributed throughout the cell. The wavelengths are divided into two groups, " W band" and "visible band", as shown in FIGS. 2 and 3, with the dividing line chosen at a wavelength of 300 millimicrons. Absorption by nuclear -material (D~ and RN~) increases markedly at wavelengths below 300 millimicrons. Although the visible wavelengths of light are generally considered to begin at 400 milli-microns, throughout this specification the "visible band"
Will be considered to begin at 300 millimicrons. The terms " W band" and "visible band" are used herein merely for convenience in defining the two wavelength bànds of interest.
The important characteristics of the two spectral bands - insofar as the present invention is concerned is that the wavelengths in the W band are absorbed to a very great extent by DX~ and R~ in a cell nucleus; they are absorbed to a much lesser extent by the cytoplasm of a cell. Energy in the visible band is also absorbed more by the nucleus than the cytoplasm; but the difference in the degree of - absorption is not as great, primarily because both cell regions absorb very little of the impinging visible light.

(The excep~ _n to this general statement -- histiocytes --will be discussed in detail below.) .
The combined visible and ultra-violet light leaves the quartz ~ace plate 84 of the cathode ray tube and is focused by quartz objective lens 85. The path of the light is shown by dotted line 48; it is to be under-stood that instead of a single ray of light a raster is generated. Quartz is used rather than glass throughout the optical system because it offers much less attenuation to ultra-violet light. The objective lens 85 images a micro-version of the cathode ray tube raster in the quartz capillary tube 46 at the level of maximum cell flow. As the cells flow by in the capillary tube, they absorb some of the focusea spot of ultra-violet and visi~le light. The focused , spot of light has a diameter in the oFder of .5 microns.

The light which emerges below the capillary tube is gathered by guartz condenser lens 86 and .
is transmitted through guartz window 87 to light box 34. In this light box there is an ultra-violet interference filter 89 which reflects the ultra-violet light (wavelengths below 300 millimicrons) toward photomultiplier tube 91, and allows the visible light (wavelengths above 300 millimicrons) to - be transmitted through it to photomultiplier tube 92. An additional filter 90 is provided to insure that any ultra-violet light which is transmitted through filtér 89 is blocked from photomultiplier tube 92.

.
~ -30-. . . .

1~)36385 The photomultiplier tubes provide electrical output signals which are proportional to the magnitudes of the respective incoming light signals. The electrical signal on W-VIDEO
conductor 94 is an indication of the amount of ultra-violet light absorbed by cells flowing through the capillary tube.
The magnitude of ~he signal on VISIBLE-VIDEO conductor 93 i9 a measure of the amount of visible light absorbed by the cells.

In the case of both the W-VIDE0 and the VISIBLE-VIDEO signals, in the absence of any light input the output of the respective photomultiplier tube is at a near-ground potential known as the "dark current" PMT level.
The output of either photomultiplier tube goes negative when light at the respective input i8 détected. The greater the light intensity, the more negative the output signal.

It should be noted that with cells moving through quartz capillary tube 46, it is possible to scan the cells simply by generating successive X sweeps. In order to scan the cells, it is not necessary to scan in the Y
direction since the moving cells in effect provide a scan in this direction. However, if successive X scans are made along the same line on the phosphor coating of the cathode ray tube, the phosphor coating may be seriously damaged. In order to distribute the heat generated by the scanning, sweeps are generated in the Y direction as well as the X
direction.
' ' 10;163~5 The cells are drawn in a carrier solution from the automatic cell handling ~ystem through capillary tube 46 in FIG. 6. They then flow through transition piece 98 and are collected in waste cell re~ervoir 54, which reservoir can be emptied periodically. ~he solution is drawn through the capillary tube by the operation of vacuum pump 35, and vacuum gauge 97 is provided to facilitate adjust-ment of the degree of vacuum inside the flow system so as to control the flow ratc. The quartz cap~llary tube in the illustrative embodiment of the invention has an inner dia-meter in the order of 90 microns and an outer diameter some-what close to 115 mîcrons. The transition pLece for coupling ,the capillary tube to the waste cell reservoir and the vacuum pump can be a hypodermic needle into which an end of the quartz capillary tube is inserted. At the other end, of the hypodermic needle,,a rubber hose can be attached for connecting the needle to the waste aell reservoir and the vacuum pump.

The flying spot control circuit 96 is adjusted such that the size of the imaged raster'through'the center of the quartz capillary tube is 110 microns in the X
direction (the X dir,ection being perpendicular to the plane of the drawing) and 82.5 microns in the Y direction (the Y
direction being along the capillary tube). The ratio of the two dimensions is 110:82.5 or 4:3, the aspect ratio in an ordinary television receiver. The cells move from left to Fight through capillary tube 46.

...... . ..... . . . . . . . . . ... .. . ... . . .. . . ... . .. . .

, . ~ .
. .

_ ~e raster is ~imilar ~ ~ ~m3 ~ln thi~ direction with the first X scan in each ~ield being at the left-most sides of the objective and condenser lenses. A normal cell has a diameter of approximately 45 micron~. The use of a capillary tube with an inner diameter of 90 microns insures that almost all of the cells will ~low through the capillary tube in single file. It is this technique which permits the examina-tion of individual cells. (CLumping of cells can still occur, ~however, and various circuits are provided to detect this condition and to prevent the aells in any clump from being categorized as either normal or non-normal.) The cells are contained in a carrier solution which is half water and half alcohol. The concentration af the cells in the carrier solution and the flow~rate are such that any-where from a few hundred to several thousand cells pass through the scanned region in the 60 seconds alloted to data gathering.

~ he velocity profile along a diameter of the capillary tube is parabolic. In accordance with well-known pri~ciples, the velocity v of a particle in a planeperpendicular to the axis of the tube (starting a short distance along the tube from the intake end) is determined from the following equation: v = ~P/4Kl) (R2-r2), where R is the inner radius of the tube (45 microns, or 4.5xlO 3cm, in the illustrative embodiment of the invention), r is the distance of the particle from the axis of the tube, K is the dynamic viscosity of the carrier solution, 1 is the length of the tube (12", or 30.48cm, in the illustrative embodiment of the invention), and ~P is tke pressure difference between the ends of the tube.

; . ~ .. .. . .. .. ~ . ... ............. .... . .. .. ... .. .... .... . ..... ... . ..... .... .. .. . .

10363t35 The parabolic flow tends to cause cellular particles to flow along the tube' 5 central axis.

The maximum flow rate is along the axis where r = 0. In the case of a perfect vacuum at the right side of the tube, aP = 76cm Hg, or ~P = 106dynes/cm2. The dynamic viscosity for water is .010 gm/cm-sec. for alcohol it i8 . 018 gm/cm-sec. For a mixture of the two in equal proportions, X - .015 gm/cm-sec. approximates the correct value. The maximum velocity is thus ~106/4(.015)(30.48?) ~4.5xlO 3)2 or 11.2cm/sec.

.
A measure of the stability of any laminar flow is given by the value of the Reynold's number. The smaller the Reynold's number, the more stable is the flow and the more easily and quickly is any perturbation ~moothed out. This i8 of particular importance in the illustrative embodiment of the invention where cells of diameters ranging from 10 to 70 microns may flow in the fluid through the tube.
The extent to which the fluid flow is disturbed by such parti-cles is determined by the Reynold's number. The maxlmum Reynold's number, RmaX~ is given by the following equation:

max = dVmax~k~ where Vmax is the maximum velocity of the flow, _ is the diameter of the tube, and k is the kinematic viscosity. For a mixture of water and alcohol k can be taken as .015cm2/sec. (same mumerical value as K). Since vmax was calculated above to be 11.2 cm/sec., the maximum Reynold's number is (9xlO 3)(11.2)/(.015), or approximately 7.

~, .

.
:: , -- . . . . . .
~ - ` ' :' , :.. ; ~ : '~ `

Reynold's numbers under 1000 have been known to result in stable flows. Certainly a very conservative stability cri-terion could be taken to be a Reynold's number under 100.
Since the maximum Reynold's number for the flow under con-sideration is 7, great stability and continued laminar flow in the presence of the cells is to be expected.

As will be described below, the capillary tube extends directly into each sample which is examined.
To insure the stability of the flow, it has been found de~irable to prevent the slowing down of the flow at any point prior to the scanning station. Such a slowing down would take place wherever the diameter of the capillary bore (or any other type of tube) increased. Thus, it is highly advantageous to extend the capillary tube directly lnto each sample -- the constancy of the inner diameter assures a stable flow. After the cells pass the scanning station, they do slow down as they enter the hypodermic needle and rubber hose which have larger inner diameters. At this point, the cells may encounter turbulence and clump together, but this is of no concern for the cells simply enter the waste -~ - cell reservoir.

~ he size of the light raster is 110 microns in the X direction and 82.5 microns in the Y direction (a 4:3 aspect ratio). The X sweep velocity is 1 line/63.5 microseconds. If 10 sweeps are desired through a typical ; normal cell, the total scan time required is 635 microseconds ; per cell.

.

, . . . . . . .. .. . . . .. . . .. . ...... ..... .. ... ..... . ... . . . ........ . . .. . . . ... . .
. . . ..

`- 10363~5 If the Y displacement o the sweeps i8 neglected (i.e., only the cell -- and not the scan lines --are assumed to move in the Y direction), and if the typical normal cell has a diameter of 45 microns, the cell must move 45 microns within 635 microseconds. This is equivalent to a velocity of 7.lcm/sec., less than the maximum velocity of 11.2cm/sec. and obtainable by having only a partial vacuum at the exit end o~ the capillary tube. Since 256 line scan~
take place within 82.5 microns in the Y direction, the raster moves 82.5/256 or .322 microns per line in the Y direction.
The cell moves at a velocity of 4.5 microns per line. The ratio of the cell velocity in the Y direction to the raster velocity in the same direction is 4.5/.322 or approximately 14.4:1; thus, the original assumption that the Y displace-ment of the sweeps can be ignored in computing the required , cell velocity for 10 line scans is verified. As far as each cell is concerned, it ~ee~ line scàns which are approximately perpendicular to the direction of its movement. (The raster is moved in the Y direction only to avoid burns on the phosphor coating on the face of the cathode ray tube.) The concentration of the cells in the carrier solution determines the number scanned per second.
- Actually, not all of the cells flowing by the scanning sta-tion are scanned. Many flow by above or below the plane of optimum focus and still others may be "hidden" by noise in the scan signal. As will be described below, the scan of such cells does not result in the collection of any signi-ficant data.

~036385 Assuming that only one of every three cells is properly scanned, and that 40 cells are properly scanned every second ~2400 per minute), the cell flow rate must ~e 120 per second.
The cumulative length of these cells is 120(45) or 5400 microns. Thus, .54cm of cellular material flows by the scanning station each second, while 7.lcm of the carrier solution flows by in the same time period. The spacing between cells is thus (7.1-.54)/(.54) or approximately 12 timeq greater than the cell dimension.

There are 60 fields (rasters) generated each second in the illustrative embodiment of the invention.-If 40 cells are scanned each second, on the average there will be two cells scanned during every three fields.

It is apparent that it is necessary to support quartz capillary tube 46 in view of the fact that its outer diameter is only 115 miaron9. It is also important that there be minimum attenuation o the ultra-violet light and that there be minimum distortion of the raster between the objeative and condenser lenses. The mounting for the quartz capillary tube, shown in FIG. 7, achieves all of these objectives.

Plate 135 is provided at its center with a recessed area in the middle of which is a hole 140. On top of the recessed area, there is placed a guartz cover ; slip 138, as shown in FIGS. 7~ and 7B.

:~-::: : :

-- 1~6385 Along the length of plate 135, on both sides o the recessed area, i9 a groove 136 into whic~ capillary tube 46 is placed (see ~IG. 7C). As shown in FIG. 7B, the capillary tube extends from the right side of plate 135, over the ~uartz cover slip, and around the curved end at the left side of the plate in the downward direction for immersion into a sample.

A syrinye 98 is placed in groove 136 as ~hown in FIGS. 7A and 7B with its beveled tip facing upward.
The syringe needle is fixed to plate 135 by silver solder 142 applied to it. The syringe serves as transition piece 98 shown in FIG. 6. Rubber tubing 137 can be coupled to the wide end of the syringe for extending the capillary bore to the vacuum system. After the syringe is mounted on the plate, the capillary tube is placed in groove 136 and its end is forced into the tip of the syringe. ~iquid wax is then allowed to drip at the junction o~ the syringe needle and the capillary tube, as shown at 141, in order to form a good vacuum seal, and in order to secure the capillary tube to --supporting plate 135. Wax is similarly applied on the otherside of the quartz cover slip and on the bend of the support-ing plate in order to secure the capillary tube in place.

A droplet of glycerine oil 139 is then placed on top of the capillary tube and quartz cover slip prior to the lowering of quartz objective lens system 85.
The glycerine oil has an index of refraction close to that of the quartz cover slip and the quartz capillary tube.

.

..... ~ , . . . . . . . ... . ..
~.~ - . .
- ' - ~ 036.385 In this manner, there is a minimum of distortion of the ultra-violet light raster as it is transmitted to the center of the capillary tube. Quartz objective lens system 86 is then raised into hole 140 in the mounting plate.

The glycerine oil is quite viscous and, therefore, adheres to the quartz cover slip and the objective lens system without a need for providing a well for containing it. The oil can be repleni3hed at weekly intervals if necessary .

.
The use of quartz throughout the optical assembly insures that there is minimum attenuation of the ultra-violet light. The use of the glycerine oil reduces the number of abrupt optical interfaces, thereby reducing light loss due to scattering and reflection, and also mini-mizes the distortion of the imaged raster. The overall a~embly also provides a convenient mount for the quartz capillary tube as well as a transition piece for extension to the vacuum pump and waste cell reservoir.

The two lens systems can be moved relative to each other until the light raster is properly focused in the vicinity of the center of the quartz capillary tube, at the - - ~ plane of maximum cell flow (the level at which the greatest numbee of fl~wing cells are in focus).

. .

.... . . .. .. .. . . . .... . .. .. . .... ... .. .. .. . . . . . . .. .. . . .. . ...... ... . .
;. . . :.: ~ . i , .,, ~ :, .
:: .

--As will be described b ~ ~ ~ells ~lowing through the capillary tube can be actually observed on a monitor di~play to fac~litate the focusing.

What emerges from the bottom of the quartz capillary tube is a moving beam of light whose ultra-violet ana visible component intensities are dynamically modulated by the absorbing cells flowing through the capillary tube, and whose instantaneous X-Y location in the capillary tube i9 synchronously related to the X-Y coordinates o~ the flying spot of light on the face of the cathode ray tube. One of the main reasons for separating the visible and ultra-violet wavelengths is that the signal/noise ratio can be improved to a considerable extent. There are two sources of noise whose effect can be reduced. The first, called "phosphor noise", arises from variations in the intensity of the out-put ~rom the cathode ray tube. If the original light inten-sity varies, the resulting signals on conductors 93 and 94 Will also vary and might otherwise be treated erroncously as representing cellular information. Although originating from the same spot of phosphor, the variations in the visible ; and ultra-violet light intensities as a result of the cathode ray tube "phosphor noise" are not identical. In other words, the ratio of the ultra-violet light intensity to the visible light intensity is not constant at all times. However, there is a close degree of correlation between the two. For this reason, as will be described with reference to FIG. 8, the two signals on conductors 93 and 94 are normalized such that their noise amplitudes (measured without cells, and with - only the carrier solution, flowing through the quartz capillary tube) are the same.

.. . . .. . .. ... . . .. . ...... ... . .. . . .. .. . .... . .. . . . . . . . .
' ~: -, -. ~ , . .
... :

By thus normalizing the two signals, it i9 possible to sub-tract the VISIBLE-VIDEO signal from the W-VIDE0 signal with the result that the two common sourced noise components, which are closely correlated with each other, to a great degree cancel out.

The second source of noise consists of non-cellular particles flowing through the quartz capillary tube. These particles generally absorb both ultra-violet and visible light. By normalizing the two signals, when the VISIBLE-VIDE0 signal is subtracted from the W-VIDEo signal, the effects of any non-cellular particles on the system operation are minimized.

.
This video conditioning minimizes tbe effects of the two types of noisé described above, and also enhances the nucleus-to-cytoplasm distinguishability. ~ow-ever, this technique itself introduces another source of noise. The visible light and ultra-violet light are detected by respective photomultiplier tubes. Each tube is in itself a source of noise (dark current noise, partition noise, shot noise, etc.). The noise components in the two photomulti-plier tube signals are not correlated with each other, and subtracting one signal from the other does not diminish noise of this type. However, it has been found that the benefits ; of canceling the other sources of noise outweigh the dis-` advantage of having this additional noise source.

.. .

10363~3S
As will be described below, the signal which is operated on by the data processor consists of the W-VIDE0 signal on conductor 94, after the VISIBLE-VIDE0 signal on conductor 93 has been subtracted from it. This signal is called the PROCE5SOR VIDEO signal. But it is also possible, as will be described with reference to FIG.
22, to electronically simulate a cell, that is, to generate an artificial W-VIDE0 signal which represents a cell with predetermined characteristics. The simulated W-VIDE0 signal appears on conductor 72. Referring to switches 28a, 29a in FIG. 6, in the operate mode switch 29a is connected to terminal 29b and the W -VIDEO signal which is transmitted over conductor 100 to the circuit of FIG. 8 is simply the output of photomultiplier tube 91. Similarly, with switch 28a connected to terminal 28b in the operate mode, the VISI-BLE-VIDE0 signal on conductor 95 is extended over conductor 101 to the circuit of FIG. 8. However, in the simulate mode, the two ganged switches are connected respectively to termi-nals 28c, 29c. In this case, the simulated signal on con-ductor 72 is transmitted over conductor 100 in lieu of the W-VIDE0 signal. ~he simulated signal has no noise in it since it does not arise from the actual scanning of cells.
There is thus no need to subtract a VISIBLE-VIDE0 signal from it to minimize the effect of noise; switch 28a is tied to a negative D.C. potential through terminal 28c in the simulate mode so that the signal on conductor 101 will simu-late a perfectly quiet PMT output corresponding to a con-stant brlght field level.

, . . . . . . . . ,.. . . : - .
. . . . . ..
--; .: . ~ .. : ~

As will be described below, the simulated signal on conductor 72 is used to adjust various threshold levels in the data processor. It would be possible to introduce the simulated signal to the data processor after the subtraction circuit of FIG. 8 rather than before it as shown. However, the advantage of introducing the signal before the subtraction circuit is that t~e signal can be used to chec~ the operation of this circuit as well as that of the data processor.

During playback of the recorded video in~ormation, the playback signal is used in lieu of the signal at the output of the subtraction circuit of FIG. 8 as controlled by switch 288a.
It therefore makes no difference in this mode which signals are fed to the subtraction crcuit and switches 28a and 29a can be placed in either position.

~ideo Conditioner -- ~IG. 8 The W-~IDEO signal on conductor 100 (or the simula-ted cell ~ignal on the same conductor) feeds into AGC amplifier 150. The output of the amplifier on conductor 162, for any line scan through the center of a typical normal cell, has a waveshape as shown adjacent to conductor 162. Reca`lling that i, the signal on conductor 94 increases in the negative direction .
with increasing ultra-violet light input to photomultiplier tube 91, in the absence of any cellular material in the quartz ; capillary tube the output of amplifier 150 is at a maximum (the , . ~
negative bright-field level). On the bright-field level there ;~ ..

.
~ -43-., .
.

~ ~ ~ r . _ _ ~ _ _ , _ '-' " ` `

1o36~s -is superimposed a noise signal, as seen most clearly to the left of the leftmost rectangular shaped pulse and to the right of the rightmost square-shaped pulse. The two pulses themselves represent the result of CAT blanking during X flybacks, During X flyback, the output from the flying spot source is at a minimum (it is not zero due to PMT dark current and afterglow of the phosphor), and the output of the ultra-violet photomultiplier tube is correspondingly at a minimum.

.. ..
' .

-43a-; . .

.... . .. . . . . . .... .. . ... . .. ...
.~:
.- . .. : , .

To the right of the leftmost X flyback pulse and to the left of the rightmost X flyback pulse are two small humps.
These represent signals due to the walls of the quartz capillary tube; the ultra-violet light is absorbed and scattered to some extent by the walls of the capillary tube. At the center of the waveform is the output of the photomultiplier tube signal as the cell is scanned. As the waveform shown is for a reference line through the center of the cell, the signal riees slightly as the cytoplasm is first scanned due to the slight absorption by the cytoplasm of ultra-violet wavelengths. As soon as the light scan enters the nucleus, the absorption increases markedly.
As the line scan proceeds past the nucleus and once again enters the cytoplasm, the signal again decreases in magnitude.

The output signal from amplifier 150 is fed into D.C.
filter 151. The filter output i8 proportional to the average value of the W-VIDE0 signal on conductor 162, which is almost exclusively dependent on the bright-field level of the signal.
The output of the filter is extended to one input of dif~erence amplifier 152, the other input of which is connected to potent-iometer tap 164. Depending on the position of thi~s tap, theoutput of the difference amplifier is a signal whose magnitude is an indication of the deviation of the amplitude of the signal on conductor 162 from a desired amplitude. ~he output of the difference amplifier is fed back along conductor 153 to the ~GC control input of amplifier 150. ~his type of AGC
control loop will be familiar to those skilled in the art.

.

10363~3S
The function of the loop is to insure that the average signal amplitude at the output of amplifier 150 remains essentially constant at a level determined by the setting of potentiometer tap 164 and the loop gain.

The VISIBLE-VIDEO signal is extended over conductor 101 to the input of AGC amplifier 154. The output signal from this amplifier is of the form shown adjacent to conductor 163.

.. . .

`` .

; ' .'' ~ .
.: .
. -44a- .

.. . ..

During X flyback, the visible light is at a minimum level and this results in the two rectangular-shaped pulses at the two ends of the line scan. Similarly, two small humps can be ob~erved as a result of the absorption of the visible light by the walls of the capillary tube, as in the case of the W-VIDE0 signal.
When the cell itself is being scanned, the visible light i9 absorbed by both the cytoplasm and the nucleus of the cell, although the absorption difference between the two cell regions iq far less than in the case of ultra-violet light. Also the VISIBLE-VIDE0 signal as the cell i~ being scanned is of lesser magnitude than the W-VIDE0 signal (even when the bright-~ield signals are normalized) because very little of the visible light is absorbed by nuclear and cytoplasmic material.

. .
The output of amplifier 154 is fed into D.C. filtar -155 to derive a D.C. signal which is dependent on the average amplitude of the VISIBLE-VIDE0 signal. The filtered D.C. s~gnal : i9 fed into one input of difference amplifier 156, the other input o~ which has a voltage which can be adjusted by the setting o~ the potentiometer tap 167. The output of the difference amplifier is fed over conductor 157 to the AGC control terminal of am-plifier 154 to maintain the amplitude of the signaL on conductor 163 essentially at a predetermined level. --In the initial set-up of the system, the signals on conductors 162 and 163 can be observed on an oscilloscope while only carrier solution flows through the capillary tube. The two potentiometer taps 164, 167 are adjusted until the two noise levels are at the same level. It is this normalizing adjustment which insures that when the two signals are snb-tracted from each other the noise components cancel to a significant extent.

. .
.. . .

,~ .

10363~S
The signal on conductor 162 i~ extended to video filter 158 which serves to attenuate the relatively high frequencies in the noise superimposed on the information of interest in the W-VIDE0 signal. Similarly, the signal on conductor 163 is extended to video filter 159. Video filters are used because the cathode ray tube phosphor noise is experienced as high-frequency random modulation. Each filter has a 12 db/octave roll-off with a 3 db corner at 450 khz.
The outputs of the two video filters are extended to the two inputs of video subtractor 160. In this unit, the two signals are subtracted from each other. The new base line is zero and the resulting signal ideally is simply the filtered ultra-violet signal which results from the scan of a cell alone tless the much smaller visible signal which also results from the same scan of the cell.) As will be described below, the W-VIDE0 signal on conductor 53 is extended to a TV type system monitor so that the operator of the machine can observe the flow of cells through the quartz capillary tube. The distinguishable pair of pulses in the signal on conductors 162 and 53 represent the signal due ~' to the walls of the capillary tube, and consequently the operator can see the walls of the tube on the TV monitor and can observe the cells flowing between them. The presentation, for example, gives the operator visual assurance that the cathode ray tube .~ . .
raster image is centered within the capillary tube, that the cells are flowing and that the focus is optimized. However, ~ -46-.... .... . . . . . . ~ . . .. . .. . . .. . .. . , .. .. . , . . . . .. _ .. . . .
. . .
.... . . .

- ~036385 there exist steep opkical gradients at the walls of the capillary tube which make them abqorb and/or reflect more visible light than ultra-violet light at some points, and more ultra-violet light than visible light at other points. To eliminate thi~
boundary interface problem electronic switch 161 is provided to gate out the portions of the subtracted signal corresponding , '.: ' ' , ' ' ~. ' - ' :
, ., .

-46a-~ . . .. . .. . . . . . .. . . .. . .. . . .... .. ...
: - :
: ~ . .. .
. .

-` 1036385 to the furthermost portions of the video presentation prior to entering the data proce~sor on conductor 52. As will be described below in the description of the data processor, conductor 51 is ordinarily low in potential, but rises in potential as the central portion of the capillary tube is being scanned. The width of the video window pulse on conductor 51 is 35 micro-~econds as compared to an X sweep period of 53.5 microseconds.
Since the dimension of the X light scan through the capillary tube is 110 microns, the physical dimension of the video window i8 ~35/53.5) (110) or 72 microns as compared to the capillary inner diameter of 90 microns. Electronic switch 161 extend~ the output of video subtractor 160 to conductor 52 (with switch 288a in the position shown) only when the VIDEO WINDOW pulse is generated. Consequently, the PROCESSOR VIDEO signal, as ~hown adjacent to conductor 52, does not contain any information outside the boundaries of the video window on any scan line.
A~ ~or noise within the video window it is attenuated to an acceptable level as a result o~ the video conditioning described above.

- 20 During both actual cell scanning and cell simulation, switch 288a is connected to operate terminal 288b since both of the scanning and simulate signals are derived at the output of subtractor 160. But during playback of a previously recorded PROCESSOR VIDEO signal, the playback signal appears on conductor 88. Accordingly, switch 288a is connected to playback terminal 288c. The playback signal includes sync information (and ., ~ -47-. , ..... .. .. . . . , .: .. . . .. ., . . _.. . .. . . . ... ....... . ... _ . . . .
~ - - . :
. .
-; :

10363~35 annotation information as will be described below), as well as cell scanning information. This sync information i~ eliminated from the PROCESSOR VIDEO signal by the operation of electronic switch 161 so that the signal on conductor 52 i8. of the same form as that obtained during the actual scanning or simulation of cells.

:. :

` ' ' .

~ , - .

-47a-.

. . ..... ... ...... , . . ... .. _ . . . . .... . . . . .. .. . . .......... ... . _ .. , . _ . ..

~; , `' ' " " ' -:
. .

-" ~03638S
Electronic Cell_Simulator -- FIG. 22 Be~ore proceeding with a deRcription of the data processor, it will be convenient to analyze the operation o~
the electronic cell simulator. The purpose of the simulator is to allow the operator to feed into the data processor repetitive information pertaining to selected cell shapes.
This information allows various potentiometers in the data processor to be adiusted to control the triggering of associ-ated circuits at diferent level~ of input. The operator can ~imulate a particular type of cell and insure that the data processor will thereafter categorize it in the desired manner by adjusting the level-setting potentiometers in the data pro-cessor which determine the categorization of the cell. Of course, the electronic cell simulation also enables the machine operation to be checked: it is easier to trace many malfunctions with known repetitive input signals than it is to trace them with the non-periodic signals resulting from actual cells.

When viewing the TV-type monitor, it will appear that the simulated cell is rectangular in shape -- both the border of the cytoplasm around the cell and the border of the nucleus inside the cell are rectangles. The left and right sides of the cell, and the left and right sides of the nucleus can be adjusted to any position within the range of the video window. Similarly, the top and bottom edges of the nucleus, and the bottom edge of the cytoplasm, can be adjusted in the Y direction. (The top of the cytoplasm cannot be ad-justed -- it always appears on line 50 of each frame. But since only the relative positions of the eight edges to each other :, ., , .. . , , . , ,.. , . ,, . . . . , ~ , . . ... . ....... . .... . ..
., , ~ .
. .
,, ~
':, . ', 10363~95 affect the categorization of a simulated cell, there is no need to vary one of the edges7 the vertical location of a cell within the video window is of no importance.) The signal amplitude (density) can be adjusted both within the simulated cytoplasm and the simulated nucleus. This is a desirable control inasmuch as the data processor includes threshold adjustments ~hich determine when the PROCESSOR VIDEO signal ha~ an amplitude suficient to characterize a cytoplasm region and a higher amplitude level suficient to characterize a nucleus region.

Referring back to FIG. 4, the X-SWEEP signal appears on conductor 111 and this signal is extended to one input of each of comparators 536, 541, 545 and 549 on FIG. 22. The setting o potentiometer tap 540 controls the leftmost boundary of the cytoplasm in the simulated cell. As long as the poten-tial on tap 540 is greater than the potential on X-SWEEP conduc-tor 111, the output of the comparator is low. ~Iowever, when the X-SWEEP signal exceeds the potential at the minus input of the comparator, the output of the comparator goes high. Thus, depending on the setting of potentiometer tap 540 (the "CYT0 , .
LEFT" control), at the same selected point along each line scan the output of comparator 536 goes high to indicate the leftmost border of the cell.

The setting of the potentiometer 542 (the "CYT0 RIGHT"
control) determines the rightmost border of the cell. Comparator 541 functions such that its output ia high as long as the potential at the plus input exceeds the potential at the minus input. It can be seen therefore that the output of comparator : . , . .
.

1~6385 536 is high from the time the X-SWEEP voltage exceeds the poten-tial on tap 540 until the end of the sweep, and the output of comparator 541 is high from the start of the sweep until the X-SWEEP voltage exceeds the voltage on potentiometer tap 542.
The outputs o~ the two comparators are extended to the two inputs of A~D gate 537. The output of the AND gate is ordinarily low but goes high during each X-SWEEP while the X-SWEEP signal is within the bounds determined by the "CYT0 LEFT" and "CYT0 RIGHT"
thresholds. The output of gate 537, the CY~0 L/R GATE, thus represents the width of the cell during each X sweep. (The width of the pulse at the output of gate 537 is the same for all 256 X sweeps in each field.) Comparators 545 and 549 serve in a capacity identical to that of comparators 536 and 541, except that they determine the width of the nucleus in the simulated cell. The output of comparator 545 is ordinarily low, but goes high when the X-SWEEP voltage exceeds the voltage on potentiometer tap 546, this voltage being the "NUC LEFT" threshold. The Gom-parator output remains high until the end of the sweep. The output of comparator 549 is initially high -- the comparator . ~..... .
output is energized whenever the potential on potent ometer tap 550 is greater than that on X-SWEEP conductor 111. The ~- potential is determined by the "NUC RIGHT" threshold adjus-; ment. When the X-SWEEP voltage exceeds the potential corres-. .
ponding to the rightmost border of the nucleus, the output of the comparator goes low.

- The outputs of the two comparators are extended to two of the four inputs of AND gate 553. The gate can thus operate only when the line scan is within the bounds determined by the ; . , . , . .. . . .. . ........ . .. ... . _ .. .. ........... . . .

potentiometer ~ettings for the left and right boundaries of the nucleus. The CYTO L/R GATE signal is extended to A third input of AND gate 553. Since, for a properly simulated cell, the left and right bounds of the nucleus must be within the left and right bounds of the cytoplasm, the output of gate 537 will be high whenever the outputs of both comparators 545 and 549 are high.
The output of gate 537 is extended to the input of gate 553 to negate the effect of erroneous settings of the threshold potentiometers by the operàtor. As a result o~ gate 553, it 0 i8 not possible to simulate a cell in which either the left or right edges of the nucleus extend outside the left and right edges of the cytoplasm.

The eight outputs of counter 124 on FIG. 5 are extended over cable 133 (a part of cable 84) to FIG. 22 where they are applied to the input of decode "50" circuit 520.
This circuit energizes its output when the binary encoded information on the eight conductors in the cable represents the decimal nuniber 50. In the simulation of a cell, the top border of the cell always begins on line 50 in each frame.

At the beginning of the scan of linq 50 in each raster, the set input of flip-flop 521 is energized and the flip-flop output switches to the 1 state energizing one input of gate 522. The X-SAMPLE pulses on conductor 113 in FIG. 4 are extended to the other input of gate 522. Once flip-flop 521 is set in the 1 state at the beginning of the scan of line 50, successive X-SAMPLE pulses are extended through gate 522 to the count input of 5-stage counter 523. In the simulation of a cell, a maximum of 31 (2 -1) lines are generated. ~If .. .. . . ....... . . . . . . . .. . . . . . .. . .. ..
,. . ~ .
' '" : ' ' .

- 10363~35 more lines are desired a larger counter can be used.) The five output conductors of the counter are extended over cable 524 to the inputs of the thr~e decoders 525, 527 and 528. Each o~ these decoders can be manually adjusted 90 that its output is energized when the count represented in counter 523 exceeds a pre-set level ~31 or less).

Decoder 528 is set so that its output i5 energized when the count in counter 523 represents the top of the nucleus.
For example, if only 5 lines are desired to represent the height of the cytoplasm above the nucleus, decoder 528 can be set to energize its output when the count in counter 523 is five. Thi~ would, of course, correspond to line number 55 of the raQter. Decoder 527 is set to energize its output when -the count in auxiliary counter S23 represents the bottom of the nucleus. For example, lf the size from top to bottom desired for the nucleus represents ten lines, and decoder 528 is set for 5, then decoder 527 would be adjusted to energize its output when the count i~ at 15. Finally, decoder 525 i8 . .
set to energize its output when the coun~ in counter 523 represents the number of lines of cytoplasm in the simulated cell. If the cell is to have a total top-to-bottom dimension equal to the spacing between 30 lines, for example, the decoder would be set to energize its output when the counter represents a count of 30. In such a case, the top-to-bottom dimension of cytoplasm below the nucleus would be 15 lines.

The output of decoder 525 is extended directly to the reset input of counter 523 and to the reset input of flip-flop 521. As soon as the line being generated corresponds to .. . . . ...... .. . .. ...... .. . . .... . . ..... .
` ~' ` ,' ' :
' ,~.... .

~036385 the bottom edge of the simulated cell, the counter is reset (in preparation for the next field) and flip-flop 521 iq switched to the 0 state. The Q output of the flip-flop goes low so that gate 522 is no longer enabled.

The Q output of flip-flop 521 is extended to the second input of gate 531. The CYTO L/R GATE input of this gate is high during all 256 line scans in every field between the outermost horizontal limits of the simulated cell. The Q output of flip-flop 521 i8 high only when the lines being generated are those within the top-to-bottom bounds of the cell. Consequently, the output of gate 531 is high only during the scan of those lines representing the cell (the sequence always begins with line 50 and in the illustrative example ends with line 80); but even during the scan of the selected lines the output is low outside the left and right bounds of the cell. Thus, the output of gate 531 is high whenever any part of the desired cell is being simulated -- this includes both the nucleus and the more inclusive cytoplasm.

The output of gate 531 is extended to the input of adjustable clipper 532. Depending on the setting of potentio-meter tap 533, the "CYTO DEWSITY" control, the output of the clipper can be adjusted to any level corresponding to the desired intensity of the PROCESSOR VIDEO signal during the simulation of the cytoplasm portion of the cell. The potentio-meter input to the clipper determines the simulated density of the cytoplasm.

Flip-f lop 530 is set in the 1 state when the output of decoder 528 goes high, that is, when the line being scanned .. . . . . .. . . . . . . . ..... . . ........ . . . . .. . . .. . . . . . ..

corresponds to the selected top of the nucleus. When the output of decoder 527 goes high, that is, when the line representing the bottom of the nucleus is being scanned, a pulse is extended through OR gate 529 to the reset input of flip-flop 530. Thu~, the flip-flop is in the l state only when the lines being scanned correspond to those desired for the simulated nucleus (lines 5-15 in the illustrative example ). The OR gate 529 is provided so that the operation of decoder 525 can also reset flip-flop 530. The bottom of the nucleus should not extend past the bottom of the cytoplasm in a simulated cell. When decoder 525 operates, it is an indication that the bottom of the cell has been reached - in the simulation, and even if the operator sets decoder 527 so that the bottom of the nucleus extends past the bottom of the cytoplasm, the pulse transmitted from decoder 525 through OR gate 529 to the reset input of flip-flop 530 prevents the simulated nucleus from having a bottom edge below the bottom edge of the cytoplasm.

While flip-flop 530 is in the 1 staté, its Q output is high. This output is coupled to an input of gate 553.
As described above, the other three inputs to the gate are high during every line scan during that portion of the swëep corresponding to the width of the desired nucleus. Although the other three inputs to gate 553 go high during each of the 256 line sweeps in each frame, the fourth input to gate 553 is high only duxing the period that the generated lines corres-pond to the desired height of the nucleus. Thus the output of gate 553 is high only when the X-SWEEP signal corresponds to the nucleus portion of the simulated cell. - -.. . .. . . . . . .. . .. . . . . ... . . . ... . . . ..
'" .

iO3f~385 The output of gate 553 is extended to the input of adjustable clipper 554. The setting of potentiometer tap 555, the "NUC DENSITY" control, adjusts the level of the output of the clipper. The output of the clipper is high only when the X-SWEEP signal corresponds to a portion of the nucleus, and the setting of potentiometer tap 555 i8 adjusted so that the output of clipper 554 represents the difference between the desired densities of the nucleus and the cytoplasm.

The output of clipper 532 is extended through resistor 560 to one input of operational amplifier 561 and the output of clipper 554 is extended through resistor 559 to the same input. ~he plus input of the operational amplifier is grounded and resistor 558 provides negative feedback from the output to the ungrounded input. As will be apparent to those skilled in the art, operational amplifier 561 functions as a summer.
The output potential to resistor 563 is the slmulated cell signal which is inverted by operational amplifier 562 and extended to terminal 29C on FIG. 6. During the first 49 line scans in each frame, neither of gates 531 and 553 operate and the potential on resistor 563 is zero. Starting`with line 50 . ~ , . . .
and until the line corresponding to the setting of decoder 528 i& reached, only g~te 531 operates. The signal across resistor 563 represents the desired cytoplasm density: during each line scan a pulse is generated whose width corresponds to the left and right bounds of the cell and whose amplitude corresponds to the simulated cytoplasm density.

:

.. . . ,. .. .. .. . . ~ ...... , . . . . . . . , . . .. . ~ . . . . . . . . ... .... .. . . . ...
.- . . .. .. . : . .. .. : .. . . :. .. . .

. ".. : : . . . .
~- .

10363~5 Beginning with the line which corresponds to the top of the nucleus as determined by the setting of decoder 528, both of gates 531 and 553 are enabled. At the start of the generation of each X sweep, neither of gates 531 and 553 operates. Gate 531 first operates when the left-most edge of the simulated cell is reached and the voltage at the output of clipper 532 rises, causing the voltage across resistor 563 to drop. As soon as gate 553 operates, corresponding to the left edge of the nucleus, the output of 554 rises, causing the voltage across resistor 563 to fall below the previous level. As soon as gate 553 turns off, at the rightmost nucleus edge in each line scan, thé
signal across resistor 563 returns to the level correspond-ing to that due to the output of clipper 532. When gate 531 turns off, corresponding to the rightmost edge of the cell, the signal across resistor 563 returns to ground potential.

The outputs of clippers 532 and 554 are positive, but the output of summer 561 is negative because of the inversion which is inherent in the summer operation.
It is necessary to invert the signal at the output of the summer in order that the signal extended over conductor i2 can simulate the signal shown adjacent to conductor 162 on FIG. 8. It is also necessary to shift the base level of the simulated cell signal to correspond to the bright-field level of the signal on conductor 162. Both o these functions are accomplished by operational amplifier 562.
The output of summer 561 is coupled through resistor 563 to the minus input of the amplifier. This configuration, ., ~ I

_ogether with the inclusion of feedback resistor 564, cause~
the signal on conductor 72 to be the inversion o~ the input signal. The plus input of the operational amplifier is held at a negative potential to shift the base level of the output to that level corresponding to the bright field.

After decoder 527 operates, gate 553 remains off. Thereafter, only gate 531 operates during each line scan and a 9 ingle-level pulse i~ generated on conductor 72 during each line scan corresponding to the cytoplasm at the bottom of the cell. After decoder 525 operate~, corresponding to the bottom of the cell, gate 531 remains off for the remainder of the frame since the cell simulation has been completed.

It will be apparent that considerable -flexibility is provided in the electronic cell simulator.
The size of each of the cell edges can be adjusted.
Similarly, the size of each of the nucleus edges can be ad~usted. Finally, the den~ities of both the cytoplasm and the nucleu~ can be varied independently.

Recorder Monitor Display -- FIG. 28 , The PROCESSOR VIDEO signal on conductor 52 is extended via video recorder summer to the video tape recorder. In the summer, the signal is combined with composite sync pulses. The signal can be recorded for subsequent play-back. Depending on the setting of a switch to be described below, the signal which is extended to the video recorder can be digitized. The PROCESSOR VIDEO signal is an analog ".. ~

signal in that the instantaneous magnitude of the signal is dependent upon the density of the particular part of the cell being scanned. It is sometimes more advantageous to display the following cells with only three levels o~ contrast (apart from the white background on the TV display) the cytoplasm can be shown as gray, a typical nucleus can be shown as gray-black, and the center of a nucleus, if it is very dense, can be shown a~ black. The data processor is capahle of digitizing the PROCESSOR VIDEO signal, and if it is desired to do so the digitized PROCESSOR VIDEO signal can be recorded rather than the analog signal itself.

.~
The data processor, in addition to categorizing each cell, also genarates various signals which are combined with either the analog or digitized PROCESSOR
VIDEO signal so that each cell which appears on the recorder monitor display during playback is accompanied by indications of the decisions concerning it made by the data processor.
~y providing a permanent record of the decision made with respect to each cell, a pathologist can play back the video tape and determine whether the data processor categ~rized the cell properly. ~ -The recorder monitor display is shown inFIG. 28. There are two data columns to the left of the video window and there are five data columns to the ri~ht of the video window. Four cells are shown in the video window.
If played back at normal speed, the cells would appear to move up rapidly on the TV display. However, the video tape can be played back on commercially available machines such .. .. . .. . . .. ... . . .. . . . . . . . . . ..

that the TV display can be slowed down or stopped so that a particular cell or pattern can be studied.

The display shown in FIG. 28 is of the type produced when the digitized PROCESSOR VIDEO signal is extended to the monitoring equipment. Cell 1 is a normal cell. The cytoplasm is gray and this is shown by vertical hatching. The nucleus is darker (gray-black) and this is indicated by the cross-hatching.

If two (or more) cells flow through the capillary tube in the form of a clump, they will be displayed as shown on FIG. 28 for cells 2 and 3. The two cells shown are normal, but as will be descr~bed below they are categor-izea as neither normal nor non-normal due to the fact that cells scanned in a clump may result in erroneous categorization.

Cell 4 i8 non-normal in a number of respects.
For one thing, at the center o~ the nucleus ~he density is too high, and this results in a black spot at the center of the nucleus on the TV display. For another, the shape of the cytoplasm is clearly irregular as compared with the other cells.
' As will be described in detail below, the data processor is arranged to perform up to five tests on each cell. Switches are available so that selected tests can be activated to allow the categorlzation of a cell as non-normal. The five columns to the right of the TV dis-59- ' ' . .

.
... . . .. . . ...... . ......... . .... .. ... . .. .
... . . . . . .

- . .

play represent the output decisions of these ~ive tests.
In the upper righthand corner of the display, a dark band appears at the top of those columns whose respective selector switches are operated so that failure of any of the activated tests will result in the categorization of a cell as non-normal. Each of the dark bands is ackually formed by blanking gated portions of six successive lines in the TV scan as the lines pass through the respective column, at the start of each ~ield. In the example shown in the drawing, the data pro-cessor has been set so that the "diameter" test is not active.
The other four tests are made on every cell. By this coding of the five rightmost columns, anyone examining the video tape will know which tests were active during the displayed frame.

The two leftmost columns are labelled "histiocyte" and "normal". Blanking of six lines in the histiocyte column below any cell is an indication that the cell has been recognized as a histiocyte. The test for histiocytes is dependent upon the VISIBLE-VIDEO signal amplitude and will be described below.
.
Each line scan on the TV display requires 53.5 microseconds -- the same time required for each ~ sweep across a cell flowing through the capillary tube. The video window ~orresponds to 35 microseconds of the total as described above. Each of the five columns on the right side of the display corresponds to two microseconds of each sweep, and each of the left-most columns correspond to 4.25-micro-seconds of the sweep. (Since the raster has an aspect ratio of 4:3, it will be apparent that the display of FIG. 28 is .. . . . ... , . . . . . .. . . .. . . . . ~ .
. . .
: j . .. - . ~
: ~ :
.:

not drawn to scale. The actual display is wider than it i5 high by a ratio of 4:3.) Cell 1 is a normal cell and accordingly 8iX
successive lines of the scan are blanked in the "inonmal"
column below cell 1. In the case of a cell clump or any other condition which is "ambiguous", a decision is not made as to whether the scanned cell ~or cells) is normal or non-normal. An ambiguou~ condition is represented by blanking the "normal" column and all five "non-normal" columns, as shown below cells 2 and 3.

If a cell is categorized as non-normal, as is cell 4, the particular tests which the cell failed and which resulted in categorizing the cell as non-normal are indicated by blanking six lines in the respective "non- :~
normal" column below the cell. In the illustrative example of cell 4, the "shoulder", "shape" and "area" tests were those failed. (The "diameter" test might also have been failed by cell 4 had the diameter test been in operation,-but since the absence of blanking at.the top of the "diameter"
; 2Q column indicates that the test was not in opèration,. no information can be obtained from the display in this regard.). As shown, cell 4 was also analyzed by the data processor as being a histiocyte. In such a case, six lines in the "histiocyte" column are blanked below the cell.

Many variations are possible, and the particular examples shown in the drawing have been selected : for illustrative purposes only. The form of the.display will .

, ..... . . . . ......................... . . .. . .. . . . .
', ~ ,. ' ' - 10363~5 be better understood following the discussion below of the five tests which are performed on each cell, as well as the method of generating the display.

Data Processor -- FIGS. 9-21 In the upper right-hand corner of PIG. 9, there is shown an idealized cell (circular, with a circular nucleus) within the video window. Five line scans are shown across the firnt half of the cell. Basic to the operation of the data processor is the initial determination as to whether the region o cellular material being examined at any particular time iQ cytoplasm or nucleus. The PROCESSOR VIDEO signal is of the form shown in the waveform adjacent to conductor 52 in . .
FIG. 8. The signal is zero (except for any noise) when a cell does not flow between the objective and condenser lenses in the optical system. As the cell cytoplasm is bcing scanned, the PROCESSOR VIDEO signal increases. And as the nucleus of the cell i8 being scanned the PROCESSOR VIDEO
signal increases till further.

In FIG. 9, directly below the idealized ~; . .
cell, there are shown the PROCESSOR VIDEO signal for~line 5 of the scan, and three waveforms which are produced as a result of the scan. In the upper waveform, the PROCESSOR
VIDEO signal is shown having a ground base voltage. Two thresholds, VcytO and Vnuc, are shown superimposed on the PROCESSOR VIDEO signal. If the signal exceeds the Vcyto threshold but is below the ~nuc threshold, it is an indication that cytoplasm is being scanned. The Vcyto threshold is .. . .. . . .. . . .. . . .. ... . ... .. . . .. . . .. . .. . . ..... . .. .. ..

~ . .

- ~036385 adjusted such that it is not exceedea by the quiescent back-ground noise which i9 present in the PROCESSOR VIDEO signal.
The Vnuc threshold is adjusted such that it is exceeded by the PROCESSOR VIDEO signal when the signal is greater than the minimum ordinarily obtainea when a nucleus is scanned.
The PROCESSOR VIDEO signal is a maximum while the center of the cell is being scanned partly because of the greater absorp~ion of ultra-violet wavelengths by the nuclear material, and partly because of the generally spherical geometry of the cell flowing through the capillary tube.
Because of the spherical shape of the cell, the ultra-violet light must pa s through more cellular material in the middle region of each cell.

..
The first waveform which is derived from the PROCESSOR VIDEO signal i8 that on conductor 204, identïfied throughout the drawing as CYTO-CHORD. This signal is ordinarily low, and goe~ high whenever the ;VcytO threshold is exceeded by the PROCESSOR VIDEO signal.
:~ .
The PROCESSOR VIDEO conductor 52 is connected to one input of comparator 200. The other input, the VcytO threshold, is connected to potentiometer tap 203 and the threshold can be varied by adjusting the position of the tap. The output of the comparator goes high whenever the PROCESSOR
- VIDEO signal exceeds the Vcyto threshold. As shown in FIG. 9, the signal is high not only while the cytoplasm is being scanned, but also while the nucleus is being scanned since at this time the VCyto threshold is certainly exceeded.

.
The second waveform which is derived is that on conductor 209, identified throughout the drawing as , . . . . . .. . . . . . . . . . . .. . .. . . . . . . . . .. . . . . .. .. .. . . .. .. . .

. . .. .

NUC-CHORD. Thi~ signal is ordinarly low, but goes high whenever the PROCESSOR VIDEO signal exceeds the Vnuc threshold. The PROCESSOR VIDEO conductor 52 is extended to one input of comparator 208. The other input, the Vnuc threshold, i9 connected to potentiometer tap 206 and as the position of this tap is varied the Vnuc threshold can be adjusted. The output of comparator 208 is ordinarily low, but goes high whenever the PROCESSOR VIDEO signal exceeds the Vnuc threshold.

The third waveform which is generated is that on conductor 211, identified throughout the drawing as C-N-CHORD.
The signal is ordinarily low, and goes high only when the CYTO-CHORD signal is high and the NUC-CHORD signal is low. The CYTO-CHORD signal on conductor 204 is connected to one input of gate 210 and thus the output of gate 210, the C-N-CHORD signal, can be high only when the CYTO-CHORD signal is high. The other input to gate 210 i8 the NUC-CHORD signal, but the signal is inverted at the input. Conseguently, the output of gate 210 can go high only in the absence of the NUC-CHORD signal.
For a scan through the center of a cell, the C-N-CH~RD signal consists of two pulses representative of the two regions of - .. . .
the cytoplasm on either side of the nucleus. These regions are referred to herein as "cytoplasm shoulders".

.
For a line scan above or below the nucleus, the CYTO-CHORD signal would be similar to that shown in FIG. 9 but would have a smaller width, the ~UC-CHORD signal would not be generated, and the C N-CHORD signal would be identical to the CYTO-CH~RD signal. The three signals are extended throughout the data processor to control various tests performed ~ . .

1ith respect to each cell. These digitized ~yes-no) signals are used in those tests where cell dimension~ (as opposed to cell densities) are the controlling criteria.

As each line scan proceeds, many digitized decisions regarding normalcy are made in real-time. However, these decisions are stored in "scratch-pad" type memories until the entire cell has been scanned. Since the digital decisions are made on a line-by-line basis, it is apparent that the X-SAMPLE and X-RESET pulses generated by the circuit of F$G. 4 can be used for the proper timing of the line-by-line decision logic circuits. However, additional signals are required to indicate the start of the scan of a cell and the end of the scan of the cell. The Y-SWEEP signal on conductor }27 (FIG. 5) which is used to control the Y scan is not sufficient for thiJ
purpose because a cell may have been scanned completely before the raster ha~ been completea. Furthermore, the scan of a cell does not necessarily begin at the ~tart of ~ach ~ sweep. In fact, during normal cell flow, it will be found that for many complete Y sweeps no cells are detected within the video window.

The circuitry on FIG. 10 is used to determine the start and end of the scanning of any cell. Each X-RESET
pulse on conductor 114 resets flip-flop 220-1 in the 0 state.
Each CYTO-CHORD pulse on conductor 204 sets the flip-flop in - the 1 state. An X-SAMPLE pulse is generated at the start of each line scan. The X-SAMPLE pulse on conductor 113 is coupled to the C (clock) input of each stage of the five-stage shift register consisting of J-K flip-flops 220-2 through 220-6.
The two outputs of each of the flip-flops are connected to the two inputs of the succeeding flip-flop in the chain.

.. . . .. .. .. .. . . . ..
~ . . .. . .

lO~;~bS
It is apparent that the six flip~flops comprise a shift register, with each X-SAMPLE pulse shifting the data down the register, and each X-RESEr pulse making a new CYTO-CHORD pulse a requisite for an entry of a 1 into the register.

The state of flip-flop 220-1 is determined by the presence of an X-RESET pulse on conductor 113, connected to the reset input, or the presence of a CYTO-CHORD pulse on conductor 204, connected to the set input. In the absence of CYTO-CHORD pulses for a number of line scans, 0's are continuously stored in flip-flop 220-1 and shifted down the chain. However, if successive CYTO-CHORD pulses are generated, l's are continuously shifted down the chain. The net result of the arrangement is that the states of the six f1ip-flops represent the recent past history insofar as the generation of the CYTO-CHORD pulses during prior successive line scans is concerned.

The Q outputs of flip-flops 220-2, 220-3 and 220-4 are coupled to three inputs of gate 221. This gate can therefore operate only if the three flip-flops are all in the 1 state. This in turn requires that three CYTO-CHORD pulses were generated during three prior successive line scans. It is this criterion that is used to indicate the start of the scan of a cell; One or even two CYTO-CHORD pulses during successive line scans is not considered to be a sufficient indication of the start of a cell. For example, an isolated CYTO-CHORD pulse may sometimes be generated when the background noise exceeds the VC~tO threshold level. With the operation of gate 221, flip-flop 225 is set in the 1 state. As long as this flip-flop is . ~

; -66-.. . .... ... . . . ... _ . , _ .... . . .. . . . . .

10363~3S

in the 1 state, it is an indication that the scan of a cell is in progress.

The end of the scan of a cell is determined by verifying the absence of the generation of the CYTO-CHORD
pulse during five successive line scans. It is only after five line scans in succession have taken place without the generation of a single CYTO-CHORD pulse that an end-of-cell condition i8 indicated. To implement this criterion, the Q
outputs of flip~flop~ 220-2 through 220-6 are each extended to a re3pective one of the five inputs of A~D gate 232. As soon as all five flip-10ps are in the O state, all Q levels will be high, and the gate operates. The signal on conductor 233 triggers monostable multivibrator 227. A positive 2-micro-second pulse is generated on CELL SAMPLE conductor 229. The CEL~ SAMPLE pulse is extended to one input of AND gate 226.
The other input to the gate is connected to the Q output of flip-flop 225. With the generation of the CELL SAMPLE pulse, gate 226 operates to generate a pulse on conductor 241. This pulse is called the 3 CON-CHORDS pulse throughout the drawing and is an indication that three consecutivei~YTO-CHORD pulses were detected to indicate the acknowledgment of the scan of a cell, and that the end of the scan of the cell has now taken place.
" :
Although ~he detection of only three CYTO-CHORD pulses during three consecutive scans is the criterion for determining the start of the scan of a cell, the absence of five such pulses in consecutive scans is required before , .

.
'' - - - --~- - - . .
~- , : . . . . :
.. ...
. - . . . .
. ~,. .. , . . :. ' - , .
. ~ .

-~ 1036385 an end-of-cell condition is established. Thi~ is due to the fact that even during the scan of a cell a number of C~TO-CHORD
pulses may be absent. As the cells flow through the capillary tube they may wander out of the focal plane of the light raster in which case several CYTO-CHORD pulses may not be generated.
Or a cell may be folded, for example, in the shape of a U, in which case CYTO-C~ORD pulses may not be generated while the mlddle part of the cell is being scanned. Furthermore, low-frequency noise of polarity opposite to that of the positive PROCESSOR
VIDEO signal, may prevent CYTO-CHORD pulses from being gener-ated during a number of successive line scans.

At the trailing edge of the CELL SAMPLE pulse monostable multivibrator 228 is triggered. A 2-microsecond positive CELL RESET pulse is generated on conductor 230. The pulse is applied to the reset input of flip-flop 225 to place the flip-flop in the O state, in which state it remains until three CYTO-CHORD pulses are detected during three succesYive line scans o~ a new cell.
'-~ ~ It is possible for a raster to end before a cell ;
has been completely scanned. In such a case, the tail part ofthe cell cannot be scanned. However, it is still desirable to make a logic decision as to normality at the end of each ; raster even if a cell has not been scanned completely. In ; general, it has been found that making logic decisions based on partial scans is better than dumping whatever data may be stored in the PROCESSOR.

.. . ... . . . ... . . ... . .. . . . ... ... .. . . . . . . .. . .. . . . .... .
. - ~ . ~ . - ~ .
., . - . . . .. . . .
: :.~ . . . .

. .

As discussed above with reference to the circuit of FIG. 5, there are at least six X-SAMPLE pulses during the flyback blanking time of the Y-SWEEP signal. During the flyback time, the PROC~SSOR VIDEO signal is at ground and the CYTO-CHORD
pulses cannot be generated. But the X-SAMPLE and X-RESET pulses are still generated at 63.5-microsecond intervals. Thus the six or seven X-RESET pulses during each Y flyback period are applied to the reset input of flip-flop 220-1, and the X-SAMPLE pulses in the succeeding cycles ~hift the O's stored in flip-flop 220-1 down the chain. Consequently, gate 232 must necessarily oper-ate at the end of each Y flyback period. In theevent flip-flop 225 has been aet earlier by the generation of three CYTO-CHORD pulses in three successive cycles, the 3 CON-CNORDS
pulse is generated to indicate that the scanning of a cell has been completed. In fact, scanning of the cell may not have been completed~because the raster may have terminated within the cell. However, whatever information has already been stored in the remainder of the data processor i9 used at this time to generate a decision as to the morphology of the cell.

There are three possible categories`into which ~-any cell may fall -- normal, non-normal or ambiguous. The -non-normal criteria will be discussed below. a cell (or -clump of cells) is considered to be ambiguous if data is - produced which is so atypical that it cannot be considered to have been derived from a single cell at all. An ambiguous condition is detected by counting too many CYTO-CHORD or ~UC-CHORD pulses during the scan of any supposedly single cell.
: .

.

~ ':, '- ' . .

Each CELL RES~T pulse is applied to the reset terminal of the five flip-flops 250-1 through 250-5 and the six flip-flops 252-1 through 252-6, shown on FIG. 11. Initially, all of the flip-flops are placed in the 0 state and the Q output terminals are low. The NUC-CHORD pulses are applied to the T (toggle) input of flip-flop 250-1. It is apparent that the five flip-flops 250-1 through 250-5 serve as a binary counter for counting the total number of NUC-CHORD pulses which are generated following the generation of a CELh RESET pulse.

The CYT0-CHORD pulses are applied to the toggle input of flip-flop 252-1. The six flip-flops in the second level form a binary counter for counting the total number of CYT0-CHORD pulses which are generated following the generation of a previous CELL RESET pulse.
..
The Q outputs of the NUC-CHORD counter, conductors 251-1 through 251-5, are extended along cable 265 to decoder 254. This decoder can be manually adjusted and serves to energizé output conductor 263 when the total count represented by flip-flops 250-1 through 250-5 exceeds the pre-set NUC-CHORD pulse count figure set in the decoder.
When conductor 263 goes high it is an indication that too many ~UC-CHORD pulses have been counted following a CELL
RESE~ pulse for the particle being examined to be considered a cell at all. Instead, the cell (or cell clump) being examined should be categorized as ambiguous.

-- . .
~.

.

.. ~ . . . , , . ... . . . . ., .. ... _ .. _ . . . . , , . , ., .. ... . . , _ _ .. . .. ..... .. . .....
. .. . . .. .. .

,: .

~ 10363~5 Similarly, conductors 253-1 through 253-6 extended from the U outputs of flip-flops 252-1 through 252-6 are connected to the inputs of decoder 255. A pre-set maximum CYTO-CHORD pulse count can be set in this decoder, and output conductor 262 is energized when the nuniber o CYTO-CHORD pulses which are counted is too large for the decision logic to categorize the cell (or cell clump) as normal or non-normal.
The pre-set count for decoder 255 i~3 larger than that for decoder 254 since more cyro-cHoRD pulses than NUC-CHORD
10 pulses arq generated for any sample. For the same reason, while only five flip-flops are required to coun-t NUC-CHORD
pulse~ (up to a maximum of 31) six flip-flops are used to count CYTO-CHORD pulses (up to a maximum of 63). In the illustrative example above in which a normal cell is scanned by ten lines, decoder 255 is preferably pre-set with a count of 15 and decoder 254 is pre-set with a count of 8.

The CELL RESET pulse on conductor 230 is applied to the reset input of flip-flop 257. With the flip-~lop in the 0 state, the Q output is low indicating that an 20 a~iguous condition has not been detected. However, if following the generation of the CELL RESET pulse, either of decoders 254 or 255 operates, the resulting pulse on conductor 262 or 263 is transmitted through OR gate 256 to t~e set input - of flip-flop 257. The flip-flop switches to the 1 state, and the AMBIGUOUS STATE conductor 260 goes high to indicate that when the CELL SAMPLE and CELL RESET pulses are eventually generated following the end of the scan of the cell (or cell clump) the al[ibiguous count should be incremented, not the non-normal count in the event the cell failed any of the 30 various tests performed on it.

. ' ' .
..
:

The VIDEO WINDOW signal is used to gate electronic switch 16] on FIG. 8 in order to eliminate noise in the PROCESSOR VIDEO signal outside the ficanning area of interest.
The window actually appears on the TV monitor presentation as two vertical lines, shown in FIG. 28 as occurring at times Tb and Tc in each line scan. (FIG. 28 depicts the form of the recorder monitor display. The microscope monitor display, which will be described below, also shows the walls of the capillary tube, and thus the video window can be accurately adjusted within the capillary tube.) The VIDEO WrNDOW signal is derived by the circuitry on FIG. 12.

The X-SWEEP waveform on conductor 111 is applied to the plus input of comparator 282 and the minus input of comparator 281. Depending on the setting of potentiometer tap 279, the output of comparator 282 is low until the X-SWEEP voltage exceeds the potential on the tap.
Thereafter, the outpu~ o~ comparator 282 remain~ high through the end of the sweep. The output of comparator 281 is high whenever the potential on tap 280 is greater than the magnitude of the X-SWEEP voltage. At the beginning of each sweep the output of the comparator is high, but as soon as the X-SWEEP
voltage exceeds the potential on tap 280 the output of the comparator goes low.

The outputs of the two comparators are ... . .
connected to two of the inputs of A~D gate 283. The third input to the gate is coupled via conductor 128 to the Q
output of flip~flop 122 on FIG. 5. The flip-flop is in the 1 state with its Q output high during the 256 line scan which ~.' .
... .. .. . . .. . . . . .. . . ..
.. . . ..... : . . ...
,, ~ .. ....
:, - : ; .:. : .
~. , ~ :. , .

~ 103638~

occur in every complete raster. The Q output is low during the Y blanking time. Thus gate 283 is enabled only during the raster time. This insures that PROCESSOR VIDEO conductor 52 i~ held at ground potential by the action of electronic switch 161 between rasters. At the beginning of each X sweep only the output of comparator 281 is high, and at the end of each X
sweep only the output of comparator 282 is high. But in the middle of each sweep both outputs go high and during active raster time the VIDEO WINDOW conductor 51 will accordingly be gated high. This signal is fed to the gate input of electronic switch 161 as described above.

It is also necessary to derive two short (.2-microsecond) pulses at both ends of the VIDE~ WINDOW.
For example, these pulses are used to control the generation of the video window limit lines on the display of FIG. 28.
The VIDEO WINDOW signal on conductor 51 is extended to the input of one-~hot multivibrator 286. The po~itive ~tep on the VIDEO WINDOW signal triggers the multivibrator which then energizes its output WI~DOW START conductor 292 for .2 microseconds. Similarly, the VIDEO WINDOW signal is inverted by inverter 284 so that the signal on conductor 285 consists of an inverted VIDEO WINDOW pulse during each line - scan. At the trailing edge of this pulse, the positive step triggers one-shot multivibrator 287. A .2-microsecond ` pulse is generated on WINDOW STOP conductor 293. The WINDOW
ST~RT and WINDOW STOP pulses are identical and occur at the start and end limits of the VIDEO WINDOW portion of each X sweep.

.
~ ... . .... . .. . . . . .. ... . . . ... . . .... . .. .... .... ...

1()36~8S
Two qummers are provided on FIG. 12 -- micro-scope monitor summer 234 and video recorder summer 294. These summers are used to combine various signalq in the data processor - for deriving the composite video signals necessary to operate the microscope monitor 700 and the video tape recorder 701 (FIG. 23). Microscope monitor summer 234 has three inputs.
The first is conductor 132 which is extended from FIG. S and has on it the composite X and Y sync information necessary to produce a TV display synced to the operation of the data processor. The second input to summer 234 is connected to ANNOTATION conductor 600. As will be described below, this conductor is connected to the output of gate 447 on FIG 20 and it is the signal on this conductor which controls the generation of the vertical lines in the display of FIG. 28 as well as the horizontal blanking segments. At this point,~
it should be noted that if switch 648 on FIG. 20 is closed, gate 447 cannot operate and there i~ no A~NOTATIO~ signal on conductor 600. The third input to summer 234 is connected to conductor 53 on which the UV-VIDE0 signal appears. This signal contains information on capillary location, cell flow, - ................................ . .
optical focus, raster uniformity, etc. (Alternatively, referring back to FIGS. 6 and 7, depending on the setting of switch 29a, the sim~lated cell signal may be selected to appear on the third input to summer 234.) --The output of the microscope monitor summer on conductor 81a is extended to a TV-type microscope monitor 700 on FIG. 6. The microscope monitor is provided so that the operator can continuously monitor conditions in the critical : ~ - - . .. . . .. . . .

10363~5 raster scan area. The following different ~V pre~entations can appear on the microscope monitor:
(1) With switch 29a in FIG. 6 in the operate position, the W -VIDE0 signal controls the display. The operator can observe the walls of the capillary tube together with the cells flowing through the tube. He can also determine optimum focus and clogged capillary conditions as they occur. If switch 648 on FIG. 20 is closed, the ANNOTATION signal is not pre~ented on the monitor. The operator sees just the walls of the capillary tube and the stream of flowing cells.

(2) If switch 29a is in the operate position but switch 648 is open, the ANNOTATION signal information is also extended to system monitor 700 through microscope monltor summer 234.
The ANNOTATION data (the information in the histiocyte, normal and vie non-normal~ columns of FIG. 28) appears on the display superimposed on the presentation of the walls of the capillary tube. The walls of the capillary, however, do not ~how up ac very dark or well ~ocused vertical columns and thus it is possible to extract all of the information from the display.
More important, the video window appears on the microscope monitor within the walls of the capillary tube and the operator can adjust the video window dimension by changing the setting of the potentiometer taps 279 and 280 on FIG. 12, or, if necessary, he can re-position the capillary tube.

(3) If switch 29a is in the simulate position, the simulated cell appears as a pair of stationary rectangles on the microscope monitor. The walls of the capillary tube are not displayed in this mode since they are not included in the simulated cell signal. However, when a cell is being simulated tbere is .

.. . . .. . .. . . . . ... . . . . . . .
: . - .

~`` 10363~35 no need to see or simulate the capillary walls since they are gated out of the data processor by electronic switch 161 in any case. If switch 648 on FIG. 20 i9 closed, the simulated cell - would appear on the microscope monitor with no annotation. If the switch i9 open, the ANNOTATION signal is allowed to pass through gate 447 and the data processor decisions appear on the microscope monitor to the left and right of the simulated cell. The characteristics of the cell can be adjusted by the operator and various threshold levels in the data processor can be changed so that desired decisions will thereafter be made for cells having pre-set characteristics. These threshold - adjustments will be described below.

(4) If switche~ 90a (FIG. 4), 129a (FIG. 5) and 288a (FIG. 8) are set for video playback, the microscope monitor display ~either cells flowing through the capillary tube or simulated cells depending on the setting of switch 29a, (FIG. 6) would have no correspondence to the playback information being fed into the data processor. The microscope monitor display should be turned off or ignored in this mbde. A display corresponding to the PROCESSOR VIDEO signal appears on recorder monitor 709. ;~ -Video recorder summer 294 also has three inputs. Two of these in~uts are the same as two of the inputs ; to the microscope monitor summer. Conductor L32 is extended to the recorder summer to extend the composite sync information to the video tape recorder. The AN~OTATIO~ signal on conductor 600 is also extended to an input of summer 294.

.

, ' . , - . . . ... . .. . . . .... .. . .. . .. .

r 1036385 The third input to the video recorder summer depends on the setting of switch 297a. With the switch connected to terminal 297c, conductor 320 is connected to the third input of the summer. The signal on conductor 320, as will be described below, is the PROCESSOR VIDEO signal, but in a digitized form so as to produce a digitized display such as that shown in FIG. 28 it is accordingly referred to as digitized video. On the other hand, if switch 297a is connected to terminal 297b, conductor 52 is connected to the third input of the video recorder summer, and the analog form of the PROCESSOR VIDEO signal is available for recording.
As for the PROCESSOR VIDEO signal itself, either one of two signals appears on conductor 52 depending on the setting of switch 288a in FIG. 8. If the switch is connected to OPERATE
terminal 288b, the PROCESSOR VIDEO signal is either an actual scanning signal or a simulated cell signal (depending upon the settings of switch 29a on FIG. 6).

on the other hand, if switch 288a is connected to playback terminal 288c, conductor 88 is connected ..
to the third input of the video recorder summer. As described above, when the video recorder is played back the signal -extended to conductor 52 is the PROCESSOR VIDEO signal previously recorded. This signal is extended through the video recorder summer back to the video tape equipment. However, the signal so obtained is not ordinarily used for recording since it is already recorded and is in fact being played back at that time.
The primary reason for extending the playback signal on conductor 88 to the conductor 52 when switch 288a is connected to playback .. , . . ~ - ~

-` 1036;~85 terminal 288c is to allow the previously recorded video signal to be applied to the data processor so that system operation can be studied while decisions are being made on the cells represented by the playback signal. The video tape signal includes ANNOTATIO~ information so that when it is used ~o ~orm the recorder monitor display (FIG. 28), the operator can view the previously scanned cell together with the previously determined ANNOTATION at the same time that the cell signal is fed to the data processor. The video tape recorder is o~ the type in which the same field can be played continuously: therefore, a stationary display can be made to appear on the recorder monitor and the same cell scan information will be fed to the data processor for many repeated fields. In theevent the data proces~or has malfunctioned, it is possible to check individual circuits in the data proaessor to determine the source of the malfunction.

As described above, with switch 297a connected to terminal 297c rather than to terminal 297b, a digitized PROCESSOR VIDEO signal is applied to the input of video recorder summer 294. This digitized signal corresponds to ; .
the PROCESSOR VIDEO signal, but at any instant in time has a discrete amplitude of only one of three possible values (in addition to the quiescent background value).

Three threshold levels are utilized to derive the digitized PROCESSOR VIDEO signal. Referring to FIG. 9.
two of these levels, Vcyto and Vnuc, have already been described. The third level, Vmia, is somewhere between the other two levels. The PROCESSOR VIDEO signal on conductor 52 is applied to the plus input of comparator 316 on FIG. 16.

..... . . . . . . .............................. . .... .. .. . .
. ~ . .
` ' ' . ~

-- 10363~3s The minus input of the comparator i9 connected to p~tentiometer tap 315 whose potential i9 held at Vmid. The output o the com-parator goes high to energize the MID-CHORD input o~ converter 319 whenever the PROCESSOR VIDEO signal exceeds the Vmid threshold.

The other two inputs to the converter are CYTO-CHORD conductor 204 and NUC-CHORD conductor 209. The digital-to-analog converter energizes its output conductor 320 with one of four po~sible potentials. A ground potential corresponds to the VCyto level not being exceeded, i.e., the absence of a cell in the quartz capillary tube. The first non-zero level corresponds to only the Vcyto level being exceeded and only the CYTO-CHORD conductor being energized.
This level results in a gray display on the recorder monitor and is depicted by the vertical hatching on FIG. 28. The second level corresponds to the Vmid threshold being exceeded a~ well as the Vcyto threshold, and is generated by the converter when both the CYTO-C~ORD and MID-CHORD inputs are high. This level results in a gray-black display and is symbolized on FIG. 28 by the cross-hatching. Finally, if all three thresholds are exceeded, all three inputs to the digital-to-analog converter will be energized, the converter ou~put potential will be at a maximum and will cause a dark display on the monitor, as ~hown in the center of the nucleus of cell 4 on FIG. 28.

The data processor performs five tests for non-normality on each cell. The first of these is "nuclear diameter".

., , "

. .
.: ,, .. ., . _ ..... . . .

10363~5 A well-known morphological criterion for cellular non-normality is the examination for enlarged nuclei. Each NUC-CHORD pulse on conductor 209 is extended to the input of integrator 270 on FIG. 12. The inverted VIDBO WINDOW pulse on conductor 285 is extended to the "dump" input of the integrator. The positi~e step at the end of each inverted VIDEO WINDOW pulse resets the integrator and its output returns to ground. The output remains at ground level until the start of the next inverted VIDEO WINDOW pulse. At this time the dump input of the inte-grator is disabled, and the integrator output is allowed torise linearly for as long as the NUC-CHORD pulse appears at the integrator input.

The NUC-CHORD pulse, referring back to FIG. 9, has a width dependent upon the size of the nucleus in each line scan, that is, the NUC-CHORD pulse is high during each line scan for as long as the PROCESSOR VIDEO signal exceeds the Vnuc threshold. The output of the integrator is connected to the pulse input of comparator 271. The minus input of the comparator is connected to potentiometer tap 272 whose setting determines the non-normal nuclear diameter threshold. The output of the comparator is low as long as the potential at the plus input does not exceea the potential on potentiometer tap 272O
During most line sweeps the output of the comparator remains low.
However, if there is at least one ~UC-CHORD pulse during the scan of a cell which causes the output of comparator 271 to go high, flip-flop 274 is set in the l state. Initially, the flip-flop is reset by the CELL RESET pulse on conductor 230 at the end of each cell scan. During the scan of the cell, the . ~ . . . . . . . . ... .. . . ~ . . . . . . .. . . . .

--` 103638S
flip-flop can be set in the l state only i~ the nuclear diameter test is failed during any line scan. The Q output of flip-flop 274 is extended to one input of AND gate 275. The second input of the gate is connected to C~L~ SAMPLE conductor 229 which is energized at the end of the cell scan. Thus if any cell fails the nuclear diameter test, NON-NORMAL NUCLEAR DIAMETER conductor 276 goes high with the generation of the CELL SAMPLE pulse at the end of the cell scan.

A second indicator of possible cell non-normality lies in the fact that malignant cells often have lesscytoplasm than normal cells. This shrinkage of cytoplasm is usually accompanied by an enlargement of the nucleus, but inasmuch as it is a distinct phenomenon it is deemed desirable to electronically detect this condition.

The widths of the two pulses in each C N-CHORD
waveform (see FIG. 9) are a measure of the dimensions of the cytoplasm region~ ("shoulders") on either side of the nucleus.
The C-N-CHORD signal is applied to the input of integrator 337 on FIÇ. 13. The X-RESET pulse on conductor 114 grounds the output of the integrator prior to the generation of each X-SWEEP
waveform. The plus input of comparator 343 is connected to potentiometer tap 339. During each line scan, whenever the C-N-C~ORD conductor 211 is high, the integrator output increases in the form of a ramp. (During the first pulse in each C.N-CHORD
pair, the integrator output increases in the form of a ramp, the -output voltage then remains constant while the nucleus of the cell is being scanned, and then the output continues to increase in the form of a ramp during the second pulse in the C.N-CHORD

... .. . . . ... . .. . .. . ... . . . . . . . . . . . ... . . . .

~ir.) IAS long as the plus input o~ the comparator has a ~ tential applied to it which i~ greater than the potential at the output oE the.integrator, the output of the comparator is high. It is only if there is sufficient cytoplasm on the two sides of the nucleus that the output of the integrator rises high enough to the point where the comparator output is driven low.

There a~e two important considerations that :. .
. muqt be borne in mind ~ith regard to the "insufficient cyto shoulders" test. In the first place, it is the measurement . . .: . , .
of the cytoplasm thickness in the vicinity of the nucleus . .
that is of interest, not the cytoplasm above or ~elow the ~ . , ., , , I
nucleus. On a scan which intercepts c~toplasmic material but no nuclear material, it is meaningless to speak of cytoplasm . ..
, . . . . . . .................................... : . .
"s~houlaers". Therefore, although integrations and comparisons : ~re made on every line scanr the result of any comparison is inhibited from registering a non-normal cytoplasm shoulderæ

. condition unless a ~UC-CHORD pulse is generated during the ,, . . : : , ., . . same line scan, indicating that the line scan is through the ..20 .nucleus. Each X-RESET pulse on conductor 114 resets ~lip-flop .
.. . . . . .
.

. 336. The Q output of the ~lip-flop goes low, and disables . . .
gate 340, It is the operation of this gate which results..

in an indication of insufficient cytoplasm shoulders. Thus , . ordinarily the condition is not indicated even if the output . of comparator 343 remains high (to indicate that the inte~rator , output aia not grow sufficiently with the application of the C-~-CHORD waveform) when the test result is sampled. Bu~ if the line scan is through the nucleus of a cell, the ~UC-CHORD
pulse on conductor 209 sets flip-flop 336 in.~he 1 state. In.
3~ ~his case the Q output of the flip-~lop goes high to enable . the test via A~D gate 340. 1~W8S
.. -82-, : .. - . . .

: '.

_ ~he second consideration of importance in the shoulders test is that the output of comparator 343 is or-dinarily high. The output of the comparator is initially high because with the generation of each X-RESET pulse the output of integrator 337 goes low. Thus at the very beginning of every line scan the output of comparator 343 is high. The output of the integrator increases during the line scan as the C-N-CHORD waveform is applied to its input. The cell is normal, insofar as this test is concerned, if the integrator output exceeds the "insufficient cyto shoulders threshold"
potential on potentiometer tap 339 by the time the C-~-CHORD
pulse pair has terminated. ~Thus it is not until towards the end of the scan that the output of comparator 343 goes low even for a normal cell. For this reason the output of the comparator should not be sampled until the end of each scan~
line.

The output of the comparator is sampled at the end of each scan line by utilizing the X-SAMPLE pulse which occurs at the beginning of the next scan. This pulse on conductor 113 is applied to the third input of A~D gate 340. It is generated before the X RESET pulse which is applied to the "dump" input of integrator 337. Thus at the beginning of each line scan, with the generation of the X-SAMPLE pulse, the output level of integrator 337 will still be at its maximum level and will be proportional to the combined width of the cytoplasmic material on either~side of the nucleus during any line scan, and correspondingly the output of comparator 343 will still be high whenever the integrator output does not exceed the insufficient cytoplasm shoulders criterion. With the application of the X-SAMPLE
` -83- ~036385 .. . . .... . . . . . ...

~0363~5 pulse to one input of AND gate 340, together with the energization of the gate input connected to the Q output of flip-flop 336 verifying that a NUC-CHORD pulse was generated on that scan, the ~ut put of the gate goes high during X-SAMPLE pulse time, indicating that the cytoplasm shoulders test was failed by the cell.

At the end of the examination of each cell, the CELL RESET pulse on conductor 230 resets flip-flop 341.
If after any line scan of the next cell, the output of AND
gate 340 goes high, flip-flop 341 i8 set in the 1 state and the Q output goes high to energize one input of AND gate 342.
At the end of the scan of the cell, the CELL SAMPLE pulse i5 generated on conductor 229 and is applied to the second input of gate 342. If at least one line scan through the cell re~
sulted in a C-N-CHORD double-pulse with a total width less than the pre-set threshold level and a nucleus was detected on that line, NON-NORMAL CYTO SHOULDERS conductor 334 goes high during CELL S~MPLE time to indicate this particular non-normal characteristic of the cell.

As mentioned above, two indicators of cell , non-normality lie in the fact that malignant cells manifest both an enlarged nucleus and shrunken cytoplasm. But it i9 possible that the nuclear diameter of a particular non-normal cell is just less than that which would result in the gener-ation of a NON-~ORMAL NUChEAR DIAMETER pulse on conductor 276 and that the width of the cytoplasmic materiaI on either side of the nucleus is just large enough to prevent the . generation of a NON-NORMAL C~TO SHOULDERS pulse on conductor ' .. . . . . ... . . . . . . . . . . . . . . .. . . . . .
.
-~34. For this reason an additional test is performed in which the ratio of the volume of the nucleus to the volume of the over-all cell is formed and compared to a pre-set threshold.
If this ratio exceeds the established threshold, it i9 taken as an indication that the cell is non-normal. The test is referred to as the N/C (nucleus/cell) ratio test, and the circuitry for performing it is shown on FIG. 14. (It is also possible to use a nucleus/cytoplasm ratio in this kind of test, the cytoplasm volume being the overall volume less the nucleus volume.) Assuming sphericity, the volume of the nucleus is proportlonal to the cube of its diameter and can thus be determined by cubing the width of the longest NUC-CHORD pulse, the pulse arising from a line scan through the center of the nucleus. Similarly, the volume of the cell is proportional to the cube of the longest CYT0-CHORD
pul~e which is generated during the same line scan. The logarithm of the volume ratio is approximately equal to three times the logarithm of the largest nucleus diameter minus three times the logarithm of the largest cell diameter.
A voltage proportional to this difference of logarithms is formed by the data processor and compared to a threshold level. The threshold level corresponds to the logarithm of the maximum acceptable nucleus/cell volume ratio for a normal cell.

.
The X-RESET pulse on conductor 114 is applied to the "dump" input of each of integrators 381 and 382, and the outputs of these integrators go to ground at the beginning of each line scan. The CYT0-CHO~D pulse is applied to the .... . . . . ............. . . . .. .
, . . .

~ 036~85 input of in~egrator 381 and thus during each line ~can the output of integrator 381 increases in the fo~m of a ramp with the maximum potential reached during each scan being proportional to the width of the CYTO-CHORD pulse. Similar-ly, the NUC-CHORD pulse is applied to the input of integra-tor 382 whose output also increases in the form of a ramp;
the maximum potential of the output of integrator 382 is a measure of the NUC-C~ORD pulse width.

The output of integrator 381 is fed to the input of logarithmic amplifier 383. The output of logarithmic amplifier 383 is proportional to the logarith of the out-put of integrator 381. Similarly, the output of integrator 382 is fed to the input of logarithmic amplifier 384, whose output is proportional to the logarithm of the integrator signal. The two logarithmic amplifier outputs are extended to the two inputs of subtraction circuit 385, whose output i9 proportional to the voltage difference between the output of amplifier 383 and the output of amplifier 384. The output of subtraction circuit 385 is thus proportional to the logarithm of the ratio of the integrator outputs, which ratio is in turn the desired volume ratio.

The output of subtraction circuit 385 is connected to the plus input of comparator 386. The minus input of the comparator is connected to potentiometer tap 388 whose potential is set to the non-normal N/C ratio threshold. The output of the comparator is ordinarily low, and goes high only if the N/C ratio exceeds the pre-set threshold.

~03638S
A~ in the case of the insufficient cytoplasm shoulders logic, concern must be given to the possibility of having a non-normal indication somewhere along the measure-ment, but not necessarily at the end of it. Consider a sin-gle line scan through the center of the cell. At the start of the scan only the CYT0-CHORD pulse is generated and con-sequently the output of subtraction circuit 385 will be below the threshold level. The output of integrator 382 rises in the form of a ramp. When the scan enters the 10 nucleus and the ~UC-CHORD pulse is generated, the output of integrator 381 ætarts to rise rapidly in the form of a Yoltage ramp, as the output of integrator 382 continues to rise in the same fashion. At the end of the scan of the nucleus, even in the case of a normal cell, the output of -subtraction circuit 385 will very likely be greater than the threshold voltage because the second cytoplasm shoulder has not yet been scanned and the output of integrator 381 ha~ not yet risen to its maximum value even though the output of integrator 382 has reached its final peak ualue.
20 It is only at the end of the line scan of any cell that the ratio test should be performed, because it is only at the end of the line scan that the desired ratio ha`s -meaning. For this reason, the ratio test is not evaluated until the end of each line scan.

The output of comparator 386 is ~ed to one input of Al~D gate 389. The X-SAMPLE on conductor 113 is connected to a second input of the gate. The X-SAMPLE pulse is generated at the end of each line scan or, more precisely, at the beginning of the next line scan. The integrator outputs . . . -... . . ~ . . . .. . . . .

.. : .
: ~ :

10363~35 _ lre still high at this point because the in~ormation the~
represent i9 not "dump~d" until the generation of the X-RESET
pulse which occurs after the X-SAMPLE pulse. ~hus, at the be-ginning of each line scan with the generation of the X-SAMPLE
pu}se, AND gate 389 energizes its output if the output of comparator 386 is high, that is, if the N/C ratio analog voltage at the plus input of comparator 386 excaeds the pre-set threshold.

At the end of the scan of every cell, the pulse on CELL RESET conductor 230, connected to the reset input of flip-flop 391, sets the flip-flop in the O state. If during any line scan of the next cell AND gate 389 operates, the flip-flop switches to the l state and the Q output goes high. The Q output is coupled to one input of A~D gate 392, the other input to which is connected to CELL SAMoeLE conductor 229. At the end of the complete cell scan, with the generation o~
the CEL~ SAMPLE pulse, NON-NORMAL N/C RATIO conductor 380 goes high to indicate that the cell just scanned has a nucleus-cell volume ratio which exceeds the maximum ratio for a normal cell. As in the case of the first two tests described above, the non-normal condition is indicated during CELL SAMPLE pulse time if at least one line scan - causes a non-normal condition to be registered. - In the case of the ratio test, as in the diameter and shoulder tests, only scans through the nucleus can result in the registering of a non-normal condition; if a CYTO-CHORD
pulse is generated without a NUC-CHORD pulse, the output o1subtraction circuit 385 cannot possibly exceed the pre-set threshold (the volume ratio is O in this case).-88-:~ ,. . ... .... .. . .. ..... .....
.. , ;. : - -' ; ' ' - ~ ' ~ 103638S
The fourth test which is perEormed concerns the shape of a cell. Consider any line scan through the nucleus of the idealized cell shown on FIG. 9. the line results in the generation of a C.N-CHORD signal which, if the nucleus is exactly centered in the cell, has two equal-width pulses. In an actual cell, the nucleus is rarely centered and one of the two pulses in the CN-CHORD waveform of even a normal cell will be wider than the other. What is not normal, however, is for one of the pulses to be wider than the other during one line scan, and the opposite condition to be true for another line scan. In suc~ a case the cell has an irregular shape: this condition generally arises in the case of a non-normal cell whose outer contour is not smooth.

In the shape test, the widths of the two pulses in the C-N-CHORD waveform generated during each line scan are designated A and B. Re~erring to FlG. 9, the A
pulse corresponds to the cytoplasm shoulder to the left of the nucleus and the B pulse corresponds to the cytoplasm shoulder to the right of the nucleus. The cell is considered normal, insofar as the shape test is concerned, if;for every line scan passing through the nucleus the A pulse is wider than the B pulse, or the B pulse is wider than the A pulse.
However, if for at least one line scan through the nucleus of a cell, A is greater than B, and for at least one line scan through the nucleus of the same cell, B is greater than A, the cell is considered to have failed the shape test.
Actually, some margin must be allowed; even in a normal cell there may be slight variations and in some line scans A may ,.. .. . . . ... ... . ...... .... .. . .... . .. .

C: ~036385 slightly greater than B and in others B may be slightly greater than A. In the implementation o~ the shape test, a record is made if B>A+~ on at least one line scan, an~
another record is made if A~-~ on at least one other line scan. The "margin" factor ~ is sufficient to prevent the registering of both inequalities in a normal cell. In a non-normal cel} whose shape if far from symmetrical, however, the test results in the registering of both inequalities during scans o the nuclear region, and the registering of both inequalities within the same cell results in the energization o ~0~-~ORM~L SH~PE conductor 480 (FIG. 17) ' ' . - .' , .- ,.~ .
during CEL~ SAMPLE pulse time. ~ - --.. . ~, . ......................... , ~ .. . .
-- , . . - .
At the beginning of each line scan, the .
~-RESET pulse on conduc~or 114 resets flip-flops 482 and .
485 on FIG. 17, and causes the outputs of integrators 486 and 487 to go to ground. In the usual case, integrator I . . . . .
487 is a5sociated with the le~tmost ~A) shoulder pulse of each pair in the C N-CHOR3 wavefonm, and integrator 486 --- is associated with the right-most (B) shoulder pulse in each pair. With 1ip-flop 482 in the O state, the Q
output is high and enables one input of gate 484. The .
second input of the gate is connected to the output of gate 481. One input to gate 481 is connected to 3-CHORD
I~IBIT conductor 488, which as will be described below is ordinarily high. The C-N-CHORD signal on conductor 211 is applied to the second input of gate 481 and the output of the gate thus ordinarily follows the C-~-C~ORD
signal.

- -- -.
.' ' ~ , .

. . : . .

la363~s As pulse A in each C-N-CHORD waveform is being transmitted through gates 481 and 484, it is applied to the input of integrator 487. The output of the integrator increases in the form of a ramp and the peak output is achieved at the termination of the A pulse. The peak output is thus proportional to the width of the first cytoplasm shoulder (the A pulse).

At the termination of the pulse, the negative step at the output of AND gate 481, coupled to the toggle input of flip-flop 482, causes the flip-flop . .
to change state. The Q output goes low and gate 484 is no longer enabled. Instead, the Q output goes high to enable one input of AND gate 483. When the B pulse in the C ~-CHORD waveform is transmitted through AND gate 481, it is now passed through gate 483 to the input of inte~rator 486. The output of this integrator now increases in the ~orm of a ramp with the peak output being achieved at the termination of the second cytoplasm shoulder, or B pulse.

Potentiometer tap 497 is adjusted until its potential is equivalent to the margin factor referred to above. This potential is applied to one input of each of -summers 498 and 499. The second input to summer 498 is con-nected to the output of integrator 486, and consequently the output of summer 498 represents the quantity B+~ . The signal is applied to the minus input of comparator 4gO. The plus input of this com-parator is connected to the output of integrator 487. The output of comparator 490 goes high only if A>B+a . The ,`- 10363~S
utput of the comparator is coupied to one input o~ A~D gate 492. One of the other inputs is connected to X-SAMPLE
conductor 113. Neglecting the other two inputs to the gate fox a moment, it is seen that the output of gate 492 goe~ -high at the en~ of each line scan (more precisely at the beginning of the next line scan with the generation of the X-SAMPLE pulse) i~ the inequality A>B~ is satisfied at this.
time.

The CEL~ RESET conductor ~30 is coupled to :
.
10 the reset input of flip-flop 494. At the end of the scan of . : .

any cell the flip-flop is reset in the O state and the Q . ---.-.- output goes low. If during an~ line scan of the next cell~- .

. : the inequality A~B~a is satisfied, the flip-flop is set in. : .
.
the l state and output Q goes high. . .
.
The output of intégrator 487 is similarly-.
extended *o the second input of summer 499. Conseguently, , the output of sum~er 499 i5 a measure of the quantity.A~ 4 ~ -The output of summer 499 is connected to the minus input ; .. .

` of comparator 489, the positi~e input of the comparatox. ..-being connected to the output of integrator 486. The output : -. of the comparator goes high only if the potential o~ the plus input exceeds the potential of the minus input, that is, i the nequality B~ A~ is satisfied. The output o comparator 489 is coupled to one input of gate 491 and X-SAMPLE pulse :

conauctor 113 is connected to a second input of the gate~ . -Neglecting the other two inputs for the moment, the output of gate 491 goes high wi~h the generation of each X-SA~LE

- .pulse following a line scan if the width of the B pulse in any ' ; -92-.
- .
~ . . . .

- _ 1()363~S
C-N~CHORD waveform exceeds the width of the A pulse by the margin factor ~. The CELL RESET pulse on conductor 230 at the end of the scan of the previous cell resets flip-flop 493 in the 0 state. If, at the end of any line scan of the present cell, AND gate 491 operates, flip-flop 493 is set in the 1 state with the Q output going high.

The Q outputs of both flip-flops 493 and 494 are coupled to two inputs of AND gate 495. The third input i to the gate is connected to CELL SAMPLE conductor 229. When the CELL SAMPLE pulse is generated at the end of the scan of any cell, AND gate 495 operates only if both inequalities have been satisfied during the course of the cell scan. The NON-~ORMAL SHAPE conductor 480 is thus energized during this time to indicate that the shape test has been failed by the cell. Immediately after the generation of the CELL SAMPLE
pulse, the CELL RESET pulse is generated to reset flip-flops 493 and 494 in the 0 states preparatory to the scan of the next cell.

The other two inputs to each of gates 491 and 492 are provided to inhibit the operations of the gates in the event either of two conditions is detected, which conditions should not be registered as representative of either inequality --.--.
FIG. 17a shows two cells 501 and 503 following each other closely through the capillary tube. Line 505 represents a line scan.
This line scan actually results in the generation of a C-~-OEIORD
signal of the type shown in FIG. 9. Even though the scan does .

.. . . . . ... . ., .. . .. . . . _ . ~ _ .. . .. . . .. .. . . .... . .. . . .. .. . . . .. ... . . . .. ..
., .

1036:~85 not pass through the nucleus of either cell, a CYT0-CHORD pulse is generated as the scan proceeds through cell 501, the pulse then terminates as the region between the two cells is scanned, and another CYT0-CHORD pulse is generate~ as cell 503 is scanned. Even though a NUC-CHORD pulse is not generated, the resulting C-N-CHORD signal on conductor 211 is indistinguishable from that which would ordinarily resùlt from a scan through the nucleus of a single cell.
It is also apparent that with the cells moving in a direction toward the top of FIG. 17a, the widths of the CYTo-CHORD pulses at the tail end of cell 501 decrease in successive line scans, while the widths of the CYTo-CHORD pulses generated at the leading edge of cell .
503 increase in successive lin0 scans. Thus while both cells are being.scanned together, during the first few scans the A pulses may be wider than the B pulses while in the last few scans the B pulses will be wider than the A pulses. ~his would result in the setting of both of flip-flops 493 and 494 in the l state with the generation of a NON-NORMAL SHAPE pulse on conductor 480, even though such a decision is not called for in:. ... .
view of the fact that the line scans which control the registrations of both inequalities do not even pass through the nucleus of the cells (which is the only time the "shape" test has any significance). -... . . . ... . .. . . .. .. . . . . . . . . . .
:..,: . . . .
. ~,. , ~ , . ., '. . .
. ~ .. . . ..

For this reason, one input of each of gate~
491 and 492 is extended along conductor 345 to the Q
output of flip-flop 336 on FIG. 13. It will be recalled that this flip-~lop is set in the 0 state at the beginninq of each line scan with the generation of the X-RESET pulse.
The flip-flop is set in the 1 state with the Q output going high only if a NUC-CHORD pulse is generated. Con-sequently, at the end of each line scan, when the X-SAMPLE
pulse is generated and applied to one input of each of gates 491 and 492, one of the gates may operate depending on which of comparators 489 and 490 has a high output --but only if conductor 345 is high at the same time. And this is possible only if the line scan included the nucleus of a cell because it is only in this case that the NUC-CHORD
pulse was generated and the Q output of flip-flop 336 is high.

The fourth input to each of gates 491 and 492 is provided to prevent the setting of either of flip-flops 493 and 494 in the 1 state in the event another anomalous ? condition occurs. This condition is also shown in FIG. 17A
by cellular material 502 flowing adjacent to cell 503.
This cellular material may be a fragment of a cell or even a white blood cell. When a line scan is taken through the nucleus of normal cell 503, but toward the upper part of the nucleus where it does not also pass through fragment 502, it is possible that pulse B will be sider than pulse ' ~ -95-A . .

.. ~ ... . . . .. ... .. . . . . . . .. . ..
.~.. . .
: .
. . . . .
.

10~S385 A, that is, there may be more cytoplasm to the right of the nucleus than to the left of the nucleus, and flip-flop - 493 would be set in the 1 state. When a line scan is next taken through the lower part of the nucleus, where the scan also passes through fragment 502, the C~-CHORD

.

~. : . , .

., .

. .

.
; ' , .
-95a-~ , .. . . . . .. .. ..
-.
.
.....

waveform actually consists of three pulses, one corre~ponding to fragment 502, and the second and third corresponding to the "true" C-N-CHORD signal for a scan through the cell.

Each negati~e step at the output of AND gate 481 toggles flip-flop 482. The first pulse in the C.N-CHORD waveform (corresponding to the fragment) is extend6d to the input of integrator 487, the second (A) pulse is extended to the input of integrator 486, and the third (B) pulse is extended to the input of in-tegrator 487 once again. (This is an exception to theusual case where integrators A and B are associated with ; the left and right cytoplasm shoulders respectively.) the output of integrator 487 at the end of line scan 506 ~ ~
is proportional to the width of the B pulse plus~the ~ ;
width of the fragment, while the output of integrator 486 i8 proportional to the width of the A pulse. Since in the example under consideration the B pulse i9 wider than the A pulse, flip-flop 494 would be set in the l state. The effect of fragment 502 is to switch the applications of the A and B pulses to the two integrators. This would result in the energization of NON-NORMAL SHAPE conductor 480 with the generation of the CELL SAMPLE pulse, and should be inhibited so that a non-normal condition lS not registered when it in fact does not exist.

Similarly, if fragment 502 is to the right of cell 503, even a normal cell might cause both of flip-flops .

~ ~ -96-, .

..... . ~ . . .... . .... . .... ..... . . . . . . . . . .. .. . . . .

10363~35 493 and 494 to be set. In this case, the A and ~ pulses are directed to the proper integrators, but the output of integrator 487 is boosted by the pulse due to the fragment. A line scan through the nucleus of cell 503 but above the fragment might set flip-flop 493, while a scan through both the nucleus and the fragment might :

' . ' ' .

-96a-' ' .

. ... . . . . . ~.. , .. , . . . .. :, ..... ..... .. ..
:~ . . . . .

~:. -. - -.

set flip-flop 494. 10363~S

For this reason, the second flip-10p 485 is incorporated in the design. If any three C-N pulses are generated during the course of a single line scan, the trailing edge of the first pulse toggles flip-flop 482 and the Q output goes high (a count of 1). The trailing edge of the second pulse again toggles flip-flop 482 and the Q output goes low. The negative step at the Q output of this flip-flop is applied to the toggle input of flip-flop 485 and consequently the Q output of flip-flop 485 goes high (a count of 2). The Q output of flip-flop 485 energizes one input of gate 496. Actuallyj this operation takes place even during the scan of a single cell without any associated cell fragments. However, when the Q output of flip-flop 485 goes high at the trailing edge of the second pulse in any two pulse C-N-CHORD waveform, the Q
output of flip-flop 482 goes low~ Since this output is connected to the second input of gate 496, the inver~ed output of the gate in the ordinary case remains high. This 2~ conductor is extended to one input of AND gate 481 which is thus ordinarily energized so that the C-N-CHORD signal ; can be extended through the gate. The 3-CHORD I~HIBIT
conductor is also extended to one input of each of gates 491 and 492 so that ordinarily these gates can operate when the X-SAMPLE pulse is generated.

However, if a third pulse occurs in the C-N-CHORD
signal during any line scan, the trailing edge of this pulse .

.. .

- . . .. . .. .. .
.

. --C 10363t~5 .
ggles flip-flop 482 for a third time. ~he Q output now goes high ~a count of 3). Flip-~lop 485 does not switch state because it does so only when the Q output of flip-flop 482 takes a negative step. Consequently, the Q outputs of both flip-flops 482 and 485 are high and both inputs of NAND gate 4~6 are thus energized. The 3-CHORD I~HIBIT sign~1 goes low to inh~bit gates 481, 491 and 492. The latter two gates are disabled.in order' . :
that neither of flip-flops 493 and 494 be set as.a result ,, , .. ; .
lO of the line scan. Gate 481. is inhibited from further : ~' ' -.
operation in order tbat both of flip-flops.482 and 485 .
.. ' remain "locked" in their l states until the start of the , . . : - . . , -' next X RESET pulse.' Were gate 481 allowed to'continue to ....... ' ' . ~ ' ,, . : . .- - .
. ' operate, it is possible that a noise spike or still another ~. - .- .
cell ~ragment might result in a fourth pulse which would . ~' ..
: toggle both flip-flops, cause the 3-CHORD INHIBIT conductor. ..
. .
' to go high once again, and permit the registering of''. ' ' .: . ; . .
information in one of flip-flops 493 or 494 which could ' :
' ' readily result in the subsequent erroneous gereration of -'. '.' ;
2'0 a pulse on NON-~OR~A~ SH~P~ conductor 480.

. It should be noted that a similar safeguard is ' ' ~ . not provided'for the three other tests described above '~nor for the fifth test to be described next). With ' . respect to the nuclear.diameter test,'the existence of . (~on-nuclear) cell fragments or white blood cells is ' ' .
. of no concern because the test is based only on the ~idth . ~ .' - , -.
, .
' ~ `98- ' : ' .
, ,; .
' ' ',' ~

1~363HS

of each NUC-CHORD signal~ With respect to the insufficient cytoplasm shoulders and ratio tests, the existence of cell fragments can result in erroneous logic decisions being made. However, this is unlikely because the cell f ragments are generally small enough so as not to increase the effective cytoplasm dimensions of a non-normal cell enough to cause it to pass the shoulder and , -98a-.. .. .. . . . . . ... . .
- . , , . :

. ' . ' : :

10363~3s `ratio tests. However, in the case of the shape test, a fragment, no matter how small, can cause the A and B
pulses to be directed to the wrong integrators and thus cause the wrong one of flip-flops 493 and 494 to be set in the l state. The decision inhibiting circuitry has been added for this reason.

These remarks apply not only to cell fragments, but to noise as well. Noise in the C.N-CHORD signal (ari~ing originally from noise in the CYT0-CHORD or NUC-CHORD pulses) might result in an extra toggling of flip-flop 482 and a change in the relative magnitudes of the outputs of comparators 489 and 490. The noise must be guarded against to a much greater extent in the case of the shape test than in the case of the other tests.

It should be noted that it would also be possible to generate a 3-pulse C-N-CHORD waveform if cell 503 in FIG. 17A is moved up slightly relative to cell 501 ~uch that line 505 passes through the nucleus of cell 503 rather than the upper part of the cytoplasm. In such a case, not only would the shape test be affected (without the provision of the 3-CHORD INHIBIT circuit), but the;dther : ~
tests could be affected as well. For example, even if cell 503 has insufficient cytoplasm shoulders, the shoulders test might not result in the setting of flip-flop 341 on FIG. 13 with the generation of the X-SAMPLE pulse if the third pulse in the C-N-CHORD waveform, arising from the scan of the tip of cell 501, causes the output of integrator 337 to increase beyond the shoulders threshold.

_99_ .. . ........ . . ..... .. ,. . ~ . ..... . ......... . .
`. . ' .'' ~

la363ss -~owever, it is extremely unlikely that two cells would overlap as they flow through the capillary tube to such an extent that the tip of the cytoplasm of one cell would be in line with the nucleus of the other.
The bore of the capillary is small enough to restrain the flow of cells to a 9 ingle file. In general, the worst case to be expected (with the exception of clumping--which is counted separately) is that in which two cells overlap as shown in FIG. 17A with a line extending through the tip of the cytoplasm of each cell, but not through the nucleus.

The fifth test which is performed by the data processor is based upon another morphological criterion frequently associated with non-normal cells and is a combination o two parameters. Cyto-technicians often characterize a cell as non-normal if it manifests both a somewhat darkened (when stained) nucleus and an enlarged nucleus. This combination of parameters quantitatively corresponds to the area of the PROCESSOR VID~O signal on FIG.
9 above the VnUc level. The larger the nu~leus, the wider and therefore larger the "peak nuclear area".- The more dense the nucleus, the greater the amplitude of the PROCESSOR
VIDEO signal and the larger the "peak nuclear area".

- The PROCESSOR VIDEO signal is extended to one input of D.C. level shifter 301 on FIG. 16. The other input is connected to potentiometer tap 306 on which appears the VnUc threshold potential. The level shifter serves to derive an output signal which is the same as the PROCESSOR VIDEO signal but with the signal level shifted .
--100-- .

... . : . . . . . ... . . .. .. ..... ... .. . .. . . .
.- :
;: . ~, .

10a6385 in the negative direction by Vnuc volts. In effect, the Vnuc line shown in FIG. 9 is shifted to ground potential, the Vcyto level becomes negative and the original bright field level is shifted to a maximum negative potential. The output of the level shifter is extended to the input of rectifier 302. The rectifier passe~ only the positive part of the signal to the in-put o integrator 303. This positive signal corresponds to the signal above the Vnuc threshold line in FIG. ~, with the VnUc line serving as the zero reference.

Referring back to FIG. 12, it will be -recalled that the NUC-CHORD pulse on conductor 209 is applied to an input of integrator 170 with the inverted VIDEO WINDOW pulse being appiied to the "dump" input of the integrator. Except during the VIDEO WI~DOW pulse, the integrator output is a~ ground potential; during the VIDEO WINDOW interval the output of integrator 270 is allowed to rise linearly with a NUC-CHORD input. A similar ar-rangement is provided for the peak nuclear area test, the inverted VIDEO WINDOW pulse on conductor 285 is extended to the "dump" input of integrator 303. The output of~the integr~tor is held at ground potential except during the - VIDEO WrNDOW pulse. However, in the case of integrator 303 the input is not a constant-level pulse. Consequently, the output does not rise linearly. Instead, the input is a signal which starts to rise from ground level, reaches a peak, and then returns to ground level. The output of the integrator is proportional to the area of the PROCESSOR VIDEO signal above the VnUc threshold.

--10~--.. . . . . . ... . . . ........ .. . . . . . . . .. . . . .. _ '' ' ' ~- The output of integrator 303 is coupled to the plus input of comparator 304. The minus input of the comparator is connected to potentiometer tap 313 whose potential is set to corréspond to the non-normal peak nuclear area threshold. The output of the comparator is low unless the potential at the output of the in-tegrator exceeds the threshold level.

With the generation of the CELL ~ESET
pulse on conductor 230 at the end of the scan of any cell, flip-flop 314 is reset. During each line scan on which there is a nucleus detected, the output of integrator 303 increases to a maximum potential proportional to the area of the signal above the Vnuc level. The output of the integrator then drops to ground prior to the next scan.
If during any line scan the output of comparator 304 goes high to indicate that the peak nuclear area threshold has been exceeded, flip-flop 314 i9 get in the l state. The Q
output goes high to energize one input of AND gate 309.
The second input, connected to CELL SAMPLE conductor 229, is energized at the end of the scan of the cell. At this time, if flip-flop 314 is in the 1 state, NON-NORMAL.:PEAK
NUC AREA conductor 310 goes high to indicate that the cell just scanned failed the flfth test. Immediately after the generation of the CELL S~MPLE pulse, the CELL RESET
pulse is generated to reset flip-flop 314 prior to the scan of the next cell.

There are thus five separate conductors on which pulses may be generated at the CELL SAMPLE time . , 103Y~385 to indicate respective non-normal characteristics of a cell. These conductors -- 276 (excessive ~UCLEAR DIAMETER), 334 (insufficient CYTO SHOULDERS), 380 (excessive N/C
RATIO), 480 (irregular S~PE), and 310 (excessive PEAK
NUC AREA) - - are extended to respective inputs of AND gates 410 through 414 on FIG. 15. As mentioned above, in some special cases it may not be desired to perform all five tests on the cells passing through the capillary tube.
Instead, it may be desired to perform only a selected number of the tests. The second input to each of the AND
gates is connected to one of switches 415 through 419.
If any switch is open, as shown in the drawing, the second .
input of the respective gate is enabled and the gate operates to transmit a pulse to its output when pulsed by the respective data input conductor at CELL SAMPLE time. The output of each of the AND gates is extended to one of the inputs of OR gate 418. Consequently, the output of the OR gate on conductor 427 is energized at CEL~ SAMPLE
time to indicate a non-normal cell if at least one of the operative tests has been failed by the cell.

Conductor 427 is connected to one input of A~D gate 428. A second input to this gate, conductor 261, is connected to the Q output of flip-flop 257 on FIG. ll. As described above, this flip-flop is placed in the O state by each CELL RESET pulse with the Q output going high. The flip-flop is switched to the 1 state only if an ambiguous condition has been detected. In the absence of the detection of an ambiguous condition, the cell can be categorized as either normal or non-normal.

.

. ~ .
. . , ., . , .. ,, , . . , ,.. , . . .. ~ , .. . ...

~0363~5 The high potential on conductor 261 energizes the second input of gate 428. The third input to gate 428 is connected to the output of AND gate 422. As will be described below, the output of this gate is normally energized during CELL SAMPLE pulse time for any cell which has met the 3 CON-CHORD criterion. Thus during CELL SAMPLE pulse time A~D gate 428 can operate to pulse conductor 69a and cause the non-nonmal cell count to be incremented whenever OR gate 418 operates.

On the other hand, if flip-flop 257 on FIG. 11 is in the 1 state, AMBIGUOUS STATE conductor 260 is high to indicate that the cell cannot be categorized as normal or non-normal. Conductor 260 is connected to one input of AND gate 420 on FIG. 15. The other input of this gate is connected to the output of AND gate 422, and with the generation of the CELL SAMPLE pulse AM~IGUOUS
conductor 69b will be pulsed to increment the "ambiguous"
count whenever an ambiguous condition has been detected.
It should be noted that at the same time the Q output of flip-flop 257 will be low, inhibiting a possible non-normal count to pass through gate 428.

The output of gate 422 is extended directly to CELL COUNT conductor 69c. The pulse on this conductor simply indicates that the counter which is keeping track of the total number of cells sampled should be incremented.
The total number of normal cells equals the total cell count, less the ambiguous and non-normal counts.

When AND gate 422 operates, it does so during CELL SAMPLE pulse time. One input to gate 422 is ,. .... .... . . , ............ .. .. . . ~,.. .. .. . . , . , ...... ~. . . .. .
.. : . : . .. .

connected to 3 CON-CHORDS conductor 241. Referring back to FIG. 10, it will be recalled that this conductor i~ pulsed when the CELL SAMPLE pulse is generated, provided that flip-flop 225 was previously set in the 1 state to indicate that three CYTO-CHORD pulses were detected on successive line scans during the scanning of the cell. The second input to gate 422 is conductor 61c which is connected to the output of one-minute run timer 754 on ~IG. 24. Samples from different persons are changed every two minutes; each sample is scanned and decisions are made and tallied during a one-minute run interval. Conductor 61c is held at a high potential to enable the second input of gate 422 for only one minute out of every two.

The third input to gate 422 is connected to the inverted output of OR gate 421. It is desired to stop the incrementing of the counters (FIG. 24) in the event the maximum count in any one of them has been reached before the one-minute run interval has been concluded.
Cables 75a, 75b and 75c are extended over cable 75 from non-normal counter 762, ambiguous counter 763 and cell counter 764. Each of these counters is incremented by a pulse on the respective one of conductors 69a, 69b and 69c.
The total count in each of the counters is extended over a respectlve one of the cables 75 to an input of the respective one of decoders 424, 425, and 426 on FIG. 15.
If the total number of cells represented in cell counter 764 equals the maximum permissible value as represented - by the setting of maximum total cells decoder 426, the 3~ output of the decoder is energized to enable one input of . . . .. .... .. . .. . .. . . . .. . .... . . . . .... . . . . .. . .
~, :

1036;~S
NOR gate 421. Similar remarks apply to decoders 424 and 425. Normally all three inputs of the gate are -disabled and the inverted output is high to energize gate 422. However, with the operation of decoder 424, 425 or ~26, the output of the r~OR gate goes low and gate 422 is disabled.

There are an additional two conductors extended over cable 69 to the circuits of FIG. 20 and FIG. 24 These conductors are HISTIOCY'rE conductor 69d 10 and LUNG SAMPLE conductor 69e. A pulse on conductor 69d indicates that the particular cell examined prior to the generation of the CE~L SAMPLE pulse is a histiocyte.
Conductor 69d can be pulsed only when a CELL SAMPLE pulse iB generated, just as conductors 69a, 69b and 69c can be pulsed only when a CELL SAMPLE pulse is generated. Con- -ductor 69e, although also relating to histiocytes, is not pulsed during the one-minute run. However, if a minimum nun~er of histiocytes are detected during the run -- a sufficient nu~er to indicate that the sample under test 20 can be characterized as a lung sample -- conductor 69e goes high at the end of the run and remains high until after the data acquisition system has recorded this fact.

The histiocyte detection circuit is shown on FIG. 18. The physiological function of histiocytes in the lungs is to trap the dirt in the air that gets through the nasal passages, and thus keep the lung sacs clear. As a result, histiocytes contain microscopic carbon particles in their cytoplasm. It is this feature which enables the data processor to detect the presence , :, , , ,, . . , . , . ..... . .~.. . . . .... .. . ... . . .. . . . . . -:: .

1~36;3~85 ~ of histiocytes in cell samples. The output of AGC
amplifier 154 on FIG. 8, the AGC-ed VISIBLE-VIDEO signal, is extended over conductor 50 to the plus input of comparator 440 on FIG. 18. The minus input of the comparator is connected to potentiometer tap 441, the potential on which represents the Vh (histiocyte) threshold. The VISIBLE-VIDEO signal is ordinarily relatively small in magnitude since the visible wave-lengths are absorbed only to a minor extent by cells flowing through the capillary tube. (The VISIBLE-VIDEO
signal is used, as described above, for noise cancellation purposes.) However, in the case of carbon-carrying histiocytes, the VISIBLE-VIDEO signal is large in magnitude and the plus input of comparator 440 exaeeds the minus input in potential. Under these circumstances, thé output of the comparator goes high to energize one input of AND gate 444.

The second input to the qate is connected to conductor 204 on which the CYTO-CHORD pulse appears.
The AND gate operates only if a CYTO-CH0RD pulse is generated while the output of comparator 440 is high. The reason for providing this second input to the gate is a safeguard --it is not enough that the VISIBLE-VIDEO signal exceed the Vh threshold; it is also necessary that the CYTO-CHORD
pulse be present to verify the presence of a cell before gate 444 operates.

Each CELL RESET pulse on conductor 230 resets flip-flop 446 in the 0 state. If AND gate 444 operates at any time during the scanning of the cell, the .. . . . .
. .. ~ , - ~ . ~ .

~ 1036;~5 flip-flop switches to the 1 state. The Q output of the flip-flop goes high and i9 extended over condu~ctor 449 to one input of gate 481 on FIG. 15. With the generation of the next CELL SAMPLE pulse on conductor 229, gate 481 operates to pulse HISTIOCYTE conductor 69d.

The Q output of flip-flop 446 is also extended to one input of gate 440. The other input to tbis gate is the inverted output of threshold histiocyte decoder 448. As long as the output of the decoder is low, the inverted input of gate 440 is high. Each time flip-flop 446 is set in the 1 state, the output of gate 440 goes high. Conductor 61c is connected to the reset input of counter 447. The positive step at the start of each one-minute run resets the counter. Thereafter, each time the output of gate 440 goes high, the counter is in-cremented. The count in the counter is extended over cable 451 to decoder 448. The decoder is pre-set with the minimum number of histiocytes required before the sample can be characterized as a lung sample. As soon as the total count in the counter equals the threshold count, the output of decoder 448 goes high. The inverted input of gate 440 goes low and consequently gate 440 no longer operates to increment counter 447. The count remains at the maximum value and thus maintains the output of decoder 448 energized.

.
; ~ -108-.. ... .. .. ........ . . . .... . .....
.
- : :

- 10363~5 The output of the decoder i9 also extended to one input of gate 450. Conductor 61c is connected to the other input of gate 450 but the signal is inverted.
The one-minute run pulse on conductor 61c i8 positive and thus the inverted input i9 low. Consequently, the output ' ~ ' - "" ' ;' ' ' ' ' , .

~' ' ' ' " ''.
' ' ---108a-.. .. .. .. . . . .. .... .. .. .....
... .. . . .. . . . . ... . . ~ . . . . .

., ~ ' ' ' - :
,: :.. :

-- 10363~5 of gate 450, conductor 61e, cannot be ~nergized during the run. ~owever, at the end of the run, the inverted input of gate 450 goes high. At this time, if the output of decoder 448 is high, both inputs of gate 450 are high and conductor 69e goes high to indicate to the data acquisition system of FIG. 24 that the sample can be characterized as a lung sample. Con-ductor 61c remains low in potential until the start of the next one-minute run. At this time the positive step on conductor 61c resets histiocyte counter 447 and disables gate 450.

After a decision is made on each scanned cell -- based upon the operation of the various gates on FIG. 15 -- signals are formed on ANNOTATION. conductor 600 which is connected to one input of each of microscope monitor summer 234 and video recorder summer 294 on FIG.
12. The purpose of the A~NOTATION signal is to produce the tabulated information in the display of FIG. 28.
The circuitry for deriving the ANNOTATION signal is shown in FIGS. 19, 20 and 21.

Referring to FIG. 28, it wi11 be seen that there are a number of vertical lines which are formed on the display. These lines are shown occurring at respective times To and Ta-Th. Each of time intervals To~Ta and Ta-Tb is 4.25 microseconds~ The VIDEO WINDOW corresponds to 35 microseconds of each line scan, and each of the five intervals between times Tc and Th lS 2 microseconds. The total active time required to form each horizontal line of the display is 53.3 micro-seconds. This is equal to the width of each X-SWEEP

... ... . . . . -- . .. ... . ~ . . .

ramp. The time T~, the start of each sweep on the dis-play, corresponds to the trailing edge of the X-RESET
pulse, the start of active sweep time.

- It will be recalled that the more positive the PROCESSOR VIDEO signal, the darker the corresponding display Thus, positive signals correspond to dark regions of the display. In order to form the vertical lines shown in FIG. 28, it is necessary to generate narrow positive pulses during each line scan at the discrete times To and Ta through Th.

The X-RESET pulse on conductor 114 is applied to the set input of flip-flop 626 and the tFigger input of one-shot multivibrator 625 on FIG. 19. Both the flip-flop and the multivibrator are triggered by positive steps but the X-RESET pulse is applied to each of their inputa through inverters. Consequently, it is at the termination of each X-RESET pulse that flip-flop 626 is switched to the 1 state and the O-A GATE conductor 615 goes high.

The positive step on conductor 615 is extended to one input o~ OR gate 622. The OR gate is - provided with a plurality of A.C.- coupled inputs, each having a short time constant. Thus, any positive step at an input of the gate results in a positive spike at the output; in effect, each input is differentiated before application to the OR gate. Each spike at the output of the OR gate triggers one-shot multivibrator 627.

.. . ...

.~ . . . . . .

10363~35 ~ he reason ~or providing A.C.- coupled inputs to the OR gate i5 that it is necessary for each new pulse on an input of the OR gate to trigger multivibrator 627, and this requires that the OR gate output go low between the energizations of the different input conductors. By differentiating each input pulse, only a short positive spike appears at the output of the OR
gate following each new pulse at one of its inputs.

The leading edge of the pulse on O-A
10 GATE conductor 615 results in a positive spike at the out-put of OR gate 622 which thus triggers one-shot multi-vibrator 625 at time To~ The period of multivibrator 627 iæ .2 microseconds, and thus at time To a .2 micro-second pulse appears on conductor 628. - The period of multivibrator 625 is 4.25 microseconds. At the end of this time interval the O-A GI~TE signal goes low. At this time both inputs of gate 624 become energized and the output of this gate, A-B G~TE conductor 616, goes high. The output of the gate is coupled to a second input of OR
20 gate 622, and the positive- step applied to an input of the OR gate triggers multivibrator 627 so that a second .2 microsecond pulse corresponding to line Ta in FIG. 28 is generated on conductor 628.

Referring back to FIG. 12, it will be recalled that the WI~IDOW START pulse on conductor 292 is ' ~ -111-.
- . ... . .. .

:. . ~ . .
.. ..

1(~363~5 generated at a time after the termination of the X-RESET
pulse determined by the setting of potentiometer tap 279.
The tap i9 adjusted to cause the pulse to be generated 8.50 microseconds after the X-RESET pulse terminates.
The WINDOW START pulse is applied to the reset input of flip-flop 626 on FIG. 19 to cause the Q output of the .. .
.:
. ~ .

-llla-.. : : . . . . . . .. .. ... .... . . . . .. . .
... . . .
'. ' ''' : , ' - flip-flop to go low. Gate 624 de-energize~, terminating the pulse on A~B GATE conductor 616. The vertical line in FIG. 28 at time Tb corresponds to the WINDOW START
pulse, and consequently this pulse on conductor 292 i9 applied to an input of OR gate 622 in order to directly trigger multivibrator 627 at the required time Tb.

The next pulse which must be generated on conductor 628 corresponds to the end of the VIDEO
WINDOW, the pulse occurring at time T as shown in FIG. 28.
The WINDOW STOP pulse on conductor 293 occurs at this time (as determined by the setting of potentiometer tap 280 on FIG. 12) and is applied directly to one input of OR gate 622. The WINDOW STOP pulse is also applied to the - ïnput of one-shot multivibrator 610. Each of multi-vibrators 610-614 is triggered by a negative step applied to its inverting input and has a fixed output period of two microseconds. The output of multivibrator 610, C-D
GA~E conductor 617 thus goes high for 2 microseconds, ~tarting at Tc. At the end of the 2-microsecond period, the negative trailing edge triggers multivibrator 611.
The output of this multiv~brator, D-E GATE conductor 618, goes high for 2 microseconds. The output conductor is connected to an input of OR gate 622, and the positive step at time Td triggers multivibrator 627 to generate a pulse on conductor 628. At the end of the 2-micro-second period, the D-E GATE conductor goes low and the trailing edge triggers multivibrator 612. Consequently - conductor 619 goea high, allowing another positive step to be applied to OR gate 622 to trigger multivibrator 627 at time Te.

s. ..... . .
" ., . . , . : .

~ , . . . .

In a similar manner multivibrator~ 613 and 614 generate F-G GATE and G-H GATE pulses on conductors 620 and 621, and the leading edge of each pulse causes a .2-microsecond pulse to be generated on conductor 628. It is still necessary to generate a pulse on conductor 628 corresponding to line Th on FIG.
28. At the trailing edge of the G-H GATE pulse, conductor 621 goes low. This conductor is applied to the inverting input of one-shot multivibrator 601 which is triggered by a negative step. The output of multi-vibrator 601 goes high and since it is connected to the last input of OR gate 622, the Th pulse is generated on conductor 628. The period of multivibrator 601 need not be two microseconds as in the case of multivibrators 610-614. Multivibrator 601 is used only to initiate the Th pulse after the G-H GATE pulse goes low.

The circuit just described serves to form hine .2-microsecond pulses during each line scan on conductor 628, which pulses, as will be described below, are inserted in the ANNOTATION signal on conductor 600 to control the generation of the vertical lines on the recorder and system monitor displays. Furthermore, various pulses derived b~ the circuit just descri~ed are used in connection with the circuits of FIGS. 20 and 21. These pulses are all positive and include the following: O-A GATE
(on conductor 615), A-B GATE (on conductor 616), C-D GATE

~ -113-.. . . . . .......... .. .. . . . . . ............... . .
:: .

1~363~35 (on conductor 617), D-E C~TE ~on conductor 618), E-F GATE
(on conductor 619), F-G GATE (on conductor 620) and G-H GATE (on conductor 621).

If switch 6~8, the annotation on-off switch, in FIG~ 20 is open, as shown, one input to gate -113a-.. . . ... . . .. . . . . . . . . .
. ^ . ~ . -. , : .
, ~ ~ . .~ . . .
~, . . . - .
.. . .
. .

~' 10363~3S
447 is energized. All signals tranqmitted through OR qate 641 are then transmitted through AND gate 447 to ANNOTATION signal conductor 600. Conductor 628, the output of one-shot multivibrator 627, is extended to a first input of OR gate 641. Consequently, all nine of the .2-microsecond pulses whose derivation has just been described appear on the ANNOTATION signal conductor, 600.

Referring back to FIG. 28, it will be noted that in the upper right hand corner selected columns are blanked during the first six line scans in each raster in order to display the tests which have been put into effect. It is therefore necessary to identify the first six lines of each raster. Cable 133 in FIG.

5 i8 connected to the eight outputs of counter 124 and the information transmitted over the cable represents a binary encoding of the number of each line scan.
Cable 133 is connected to the input of each of circuits 607 and 608 on FIG. 19. Circuit 607 functions to generate an output pulse whenever the line being scanned is line - -1, that is the first line in each raster. -The pulse at the output of circuit 607 sets flip-flop 609 in the 1 state and the Q output goes high. Circuit 608 functions to pulse its output and reset flip-flop 609 when the count represented by the in~ormation on cable 133 corresponds to the seventh line. Recalling that the count in counter 124 is incremented prior to the scan of each line (at the leading edge of X-SAMPLE time~, it will be seen that circuit 608 resets flip-flop 609 prior to the generation of the seventh X-SWEEP signal. Consequently, ' ' `

-` 10363~5 LINES 1-6 GATE conductor 606 is high during the first six line scans in each rast~r.

Referring to FIG. 15, conductors 430-434 are connected to one input of respective AND gates 410-414.
If the respective switch is open, the input to each gate is floating (a logical "1") as is the respective one of conductors 430-434. The conductors are extended along cable 435 to FIG. 21. Conductor 430 is connected to one input of AND gate 671, conductor 431 is connected to one input of AND gate 672, and similar remarks apply to con-ductors 432-434 and gates 673-675. Conductor 606 is connected to the second input of each of ~ND gates 671-675.
If, for example, the diameter test is.active, conduator ~30 enables gate 671. During the first.six lines of each raster, conductor 606 is high and energizes the second input of AND gate 671. However, the third input of the gate, connected to C-D GATE conductor 617, is high only between times T and Td f each scan. Consequently, gate ~71 operates only during the first six line scans while the first non-normal column of FIG. 28 is ~eing scanned --provided the diameter test is in effect. The resulting - :
pulse at the output of gate 671 is extended through OR
gate 676 to conductor 681. Th~ conductor is in turn :
connected to an input of OR gate 641. Consequently, if the diameter test is being performed, the top six lines of the diameter column in the display are darkened (not as shown in FIG. 28 since in the illustrative example the diameter test is not in effect).

Conductor 606, the LINES 1-6 GATE
conductor, is connected to one input of each of gates .: , . . . ~ . . ... . .,... . .... . : .... ... . .
,: -;, ~ .
'' ~ ' .
::

672-675. Another input to each of the gates is connected to a respective one of conductors 431-434 which enables the respective gate whenever the associated test is in effect. The third input to each gate is connected to one of conductors 618-621 to energize the gate only while the first six line scans of an active test are going through their respective column. The output of each of gates 572-675 is extended to one input of a respective one of OR gate~ 677-680. Respective output conductors 682-685 are extended to inputs of OR gate 641. ~t is apparent that this arrangement controls the coding at the top of each of the non-normal columns in accordance with the settings of switches 415-419 on FIG. 15, and the column assignments shown in the illustrative display of FIG. 28.

The rest of the information represented in the columns of the display also appears in bands having a height of six lines. The first line in any band can occur on any line of a raster after the first eight lines.
The end-of-cell condition cannot be detected until after the first eight line scans in a raster because thrëë
line scans are necessary to acknowledge the presence of a cell and five additional line scans are required to acknowledge the CYTO-CHORD pulses are no longer being generated. This insures that there can be no overlap of the active test bands at the top of the display and the decision bands below them in the same columns. It is necessary to count six lines after any decision is made in order to generate the respective 6-line decision ... . . . . . . . .. . . . , .. .. .. .. , ~ ~ . .......... . .. .
~ ' " ,, ~036385 ANNOTATION band or bands in the display. With the generation of the CELL S~MPLE pulse on conductor 229 at the end of the scan of any cell, gate 422 on FIG. 15 operates if uninhibited by the maximum count gate 421, and the one-minute run has not yet been terminated so that conductor 61c is still high. Under these con- -ditions, conductor 423 goes positive with each CELL
SAMPLE pulse for which the 3 CON-C~ORDS criterion has been met. This conductor is connected to the set input of flip-flop 630 on FIG. 20. The flip-flop is thus set in the 1 state at the acknowledged end of each cell, i.e., five lines after the last CYTO-CHORD pulse was detected for the cell. The Q output of flip-flop 630 is connected .
to one input of AND gate 631. The other input to the gate i8 connected to X-RESET conductor 114, At the start of each line scan following the setting of flip-flop 630 in the 1 state, the X-RESET pulse is extended through gate 631 to the increment input of 3-stage counter 632. The ~ount in the counter is thus incremented prior to each .
line scan. Just prior to the seventh line scan, the ~., .
count in the counter is a binary encoded 7. Circuit 633 is pre-set to energize its output for such a condition.
With the operation of circuit 633, one-shot multivibrator 634 is triggered to generate a DDR (DECISION DISPLAY
RESET) pulse on conductor 635. The pulse is e~tended to the reset input of flip-flop 630 to reset the flip-flop to the 0 state. A~D gate 631 is thereby disabled to ~ -117--- . .. ~.. . ..... . . ........
.. . .

. . .
:~ .

10363~35 inhibit subsequent X-RESE~ pulses from entering the input of counter 632. The DDR pulse also resets counter 632 to a count of zero, in preparation for the next cycle of operation.

' ~
. .
. ' :

--117a-. .

:
..

~0363~3S
Conductor 436a is connected to the output of gate 410 on FIG. 15. With the generation of the CELL SAMPLE pulse, the output of gate 410 goes high if the diameter teqt has been failed by the cell whose scan has just been completed. The pulse appears on con-ductor 436a at the same time that gate 422 operates and flip-flop 630 is set in the 1 state. Conductor 436a is extended to the set input o diameter ~lip-flop 661 on FIG. 21. The flip-flop i8 set in the 1 state to enable one input of AND
gate 666. The second input to the gate is connected to con-ductor 617 which is high during each scan between times T and Td. Consequently, the output of gate 666 goes high during each subse~uent line scan while the scan is passing through the diameter column of the display. Thë output of the gate is connected to one input of OR gate 676 so that the resulting pulse is extended to ANNOTATION conductor 600. Flip-flop 661 is set at the same time that flip-flop 630 is set, i.e., during CELL SAMPLE pulse time. The DDR pulse on conductor 635 is applied to the reset input of flip-flop 661. Consequently, flip-flop 661 if energized, remains in the 1 state only as long as flip-flop 630 remains in the l state, that is, during six line scans following the generation of a pulse on conductor - 436a. Thus, if the diameter test has been failed by a cell, æix lines in succession are blanked in the diameter column of the display.

In a similar manner, each of conductors 436b-436e is extended to the set input of a different one of ~ -118- , -, -.. . . . .. . .
.... , ~ ~

10;~S3~5 shoulder, ratio, shape and area flip-flops 662-665. The Q
output of each flip-flop is connected to one input of a respective one of gates 667-670, and the second input of each gate is connected to a respective one of the D-E GATE, E-F GATE, F-G G~TE and G-H GATE conductors 618-621. The output of each -118a-.. . . .... . ... .... .. . .... ... . .... .. . .............. _ .. . . . ~ .. .....

of gates 667-670 i9 connected to an input of a respective one of OR gates 677-680. Finally, the DDR conductor 635 is connected to the reset input of each of the flip-flops.
It is apparent that if any one of the active te~ts has been failed, the respective one of conductors 436a-436e is pulsed an~ a 6-line band appears below the level of the cell in the respective column of the display.

It will be noted that the WI~DOW STOP pulse on conductor 293 is applied to the set input of flip-flop 636 in FIG. 20. The WINDOW STOP pulse occurs at time Tc during each scan and consequently it is at this time that the Q output of flip-flop 636 goes high. The G-H GATE conductor 621 is connected to the inverted reset input of 1ip-flop 636. When the conductor goes low at the termination of the G-H GATE pulse, the negative step at the reset input of flip-flop 636 causes the Q output of the flip-flop to go low. Consequently C-H GATE
conductor 637 is ~eld high only between Tc and Th.

The C-H GATE conductor is extended to one input of OR gate 63R. The other input to the OR gate is the 20 A-B GATE conductor 616. Consequently, the output of OR gate 638 is energized during each X SWEEP, between times Ta and Tb, and times Tc and Th.

- If an ambiguous condition has been detected, conductor 69b in FIG. 15 is energized during CELL SAMPLE pulse time. This conductor is extended to the set input of flip-.. . . . . . . .. . . .. . .. .. .... . . . . .. . ... . . ..
~ .
..
- :

10363~35 flop 640 on FIG. 20. The Q output of the flip-flop goes high to energize the other input of gate 639. Consequently, gate 639 operates between times Ta and Tb, and between times Tc and Th, for as long as flip-flop 640 remains in the 1 state following an ambiguous decision. The output of gate 639 is extended to one ...

~ ':

-119a-.. . . . . . . . . . . .. .... . . . . . .. . .. . . . .. .... .... . .
' :: . ~: ' ' .
. ~. , ~ ' ,, - ~ , 10363~3S
input of OR gate 641 and controls the generation of the necessary bands on the display to indicate an ambiguous condition. Refer-ring to cells 2 and 3 of FIG. 28, it will be recalled that the ambiguous decision is represented on the display by a dar~
band in the normal column (from Ta to ~b) and across all of the non-normal columns (from Tc to Th). The band extenas over only six lines on the display because the DDR pulse on conductor 635 i9 sxtended to the reset input of flip-flop 640 to de-energize the Q output of the flip-flop and thus disable gate 639 six lines after CELL SAMPLE pulse time.

It should be noted that during the generation of the ambiguous bands on the display, certain ones of OR
gates 676-680 on FIG. 21 can still operate since the oper- - -ations of these gates i9 controlled by pulses on conductors 436a-436e and these pulses are generated when respective tests have been failed -- even in the presence of an ambiguous ~ituation. However, it i9 apparent that the outputs of gates 676-680 simply overlap the output of gate 639 in the non-normal columns and consequently there is no need to inhibit the operation of these gates.

If a cell has been categorized as normal~
NON-NORMAL conductor 69a is low in potential and AMBIGUOUS conductor 69b is similarly low in potential when CELL COUWT conductor 69c goes high.

.. . . . . ..

:

-10363~35 All three of the~e conductors are extended to respective inputs of gate 649 on FIG. 20, the signal on conductors 69a and 69b being inverted before they are applied to respective inputs of the gate. Consequently, in the case of a normal cell the output of gate 649 goes high to set flip-flop 644 in the 1 state for energizing one input of gate 645. The second input to gate 645 is connected to A-B GATE conductor 616. Consequently, the output of gate 645, connected to an input of OR gate 641, goes high during each of the line scans following the generation of the CELL SAMPLE pulse while the scan goes through the normal column of the display. After six line scans have taken place, the DDR pulse i9 applied to reset input of flip-flop 644 to diuable gate 645. Con-~equently, a band such as that shown for cell 1 in FIG. 28 is generated in the normal column.

The only remaining ANNOTATION band which must be generated is that in the histiocyte column ln the case of a lung cell. In the case of a hietiocyte, conductor 69d is pulsed when the CELL SAMPLE pulse is generated. This conductor is connected to one input of flip-flop 642 in FIG. 20 and the flip-flop is set in the 1 state to energize one input of AND gate 643. The other input to the gate is connected to O-A GATE conductor 615.
The output of gate 643 is extended to an input of OR gate 641. Consequently, following the generation of a pulse ~ -121-.. . . .. . . .. . . . . ........ . . . .
~. ~

10363~3S
on HISTIOCYTE conductor 69d, a band is generated in the histiocyte column such as that shown for cell 4 in FIG. 28.
The band extends over only six lines because only six lines are generated between the time the HISTIOCYTE
decision i5 made and the time the DDR pulse resets flip-flop 642.

, .

--12la-.

.-- :
' - 10363~5 Recorded and PlaYback System -- FIG. 23 The output signal of video recorder summer 294 on FIG. 12 is similar to ~he output signal of microscope monitor summer 234 except that instead of containing a W-VIDEO or simulated cell component it contains a PROCESSOR
VIDEO or simulated cell component in analog or digitized form. The output of the summer on conductor 81b is extended to the RECORD input of video tape recorder 701.

The signal recorded on the video tape is a composite signal including cell scan information, X and Y sync information and annotation information. When the data processor is switched to the playback mode, and the video tape is played baok, the composite playback signal appears on conductor 88.
Conductor 88 is extended to the clrcuit of FIG. 4, so tha~ the sync information recorded on the video tape can be used to sync the aetion of the data processor to the recorded signal. In the playback mode, switch 288a of FIG. 8 i~ eonnected to terminal 288e and the tape signal is extended through the synchronized electronic switch to PROCESSOR VIDEO conduct;or 52. The electronic switch eliminates the sync and annotation signals which were present in the composite playback signal so that the resulting PROCESSOR VID~O signal is of the same form as that fed to the aata processor in the operate mode.

The composite playback signal on conductor 88 is extended through video summer 715 on FIG. 23 to the input of .

~ -122-:. . .... ,, , . .. ., ... .. .... . , .,.. , .. .. , .. .. - .. . . ..
: - . . , . ,:;
; . :
~ : , .' ' ` :, ~

recorder monitor 709. The playback signal -- with its annota-tion -- appears on the recorder monitor. One of the main reasons for providing the recorder monitor is to allow a single cell to be examined in detail. This is easily accomplished with a conventional video tape recorder which -122a-, ... . . . ~ . . . . . . . . . , , ................ ~ .

. .

~0363~S
can be adjusted to play the same field over and over again.
In such a case the display on the recorder monitor will be stationary as shown in FIG. 23. It should be noted that in such a case the same cell would be analyzed over and over again by the data processor. In the event it is deter-mined that a decision previously made by the data processor is incorrect (by studying the cell on the display and noting the annotation below it on the recorder monitor), it is possible to trace the source of the mistake since the same cell information is being inserted into the data processor over and over again.

" .
As described above, system oscilloscope 714 can be used to observe the output of video recorder summer 294 on a real-time basis. However, it is very difficult to extract much useful information from the display because the display consist~ of many pulses close together. Even if the annotation pulses are removed from the trace by closing switch 648 on FIG. 20, the line scan in-formation essentially appears as a series of spikes. It would be highly desirable to be able to examine the profile . . .
of a single line scan, that is, a signal of the form shown adjacent to conductor 52 in FIG. 8. This can be accomplished on the recorder oscilloscope 708 of FIG. 23 if the same field - is played back over and over again.

The composite playback signal includes X and Y
sync information. The signal is axtended to the input of ~L ' ~ -123~

:: - . . .

-~03638S
X-sync stripper 702 and Y-sync stripper 703. At the end of each playback of the field, Y-sync stripper 703 pulses the reset input of line counter 704. Thareafter, in the next playback of the same field, each X sync pulse increments the line counter. Manually adjustable decoder 705 can be set so that its output, SELECTED LINE conductor 706, goes high when the count in counter 704 equals the manually selected - .'' ' .'.

-123a-~,~,.. ... ~ ... . .. .. .... . .. . .. .... .

10363~S
value. Conductor 706 goes high only as long as this count is maintained in the counter. Consequently, conductor 706 remains high only while a single line i9 being played back.
Conductor 706 is extended to the sync input of the recorder oscilloscope. The positive step at the start of the line scan is used to sync the oscilloscope. Since the start of the same line scan always triggers the oscilloscope, it is possible to adjust the time base such that only the selected line scan (or two, etc.) appears on the scope.

The pulse on conductor 706 is also extended to an input of video summer 715. The output of the summer on conductor 707 is extended to the input of each of recorder oscilloscope 708 and recorder monitor 709. The input to the video summer on conductor 88 is the composite playback'" - - ' signal, and includes the line scan information, the sync information and the annotation information. Added to this composite signal is a step corresponding to the selected line.
In ten~s of the display on the recorder oscilloscope, this step is seen as a pedestal on which the selected line infor-mation appears. The main function of the pulse on conduc-tor 707, however, is to generate a dark line;`on the recorder monitor while the line of interest is being formed. This iine is shown as line 710 on the display of the recorder monitor.
As the decoder setting is manually adjusted, line 710 may be observed to move up and down on the display. In this way any line through the cell can be selected simply by changing the setting of the decoder while at the same time observing the selected line on the recorder monitor. This allows any line through the cell to be selected with a minimum of effort in order to view the profile of the line information in detail.

, - ~03638S
The pulses on pedestal 713 on the recorder oscilloscope display correspond to the nine vertical lines which appear on the recorder monitor display since these vertical lines are generated b~ the generation of nine .2-microsecond pulses during each line scan. The negative going pulse at the trailing edge of the pedestal corresponds to an inverted X-SAMPLE pulse. Similarly, the reduced level at the leading edge of the pedestal corresponds to an X-SAMPLE pulse at the start o~ the line scan~

Once a display such as that shown on FIG. 23 i9 achieved, it is possible to check the operation of the data processor in the following manner. It is often desir-able to determine which line scan through a cell causes a particular decision to be made. The input to the oscillo-scope can be moved from conductor 707 to a number of points in the data processor. For exa~ple, the display on the re-corder monitor indicates that the shoulders test was failed.
Suppose it i9 de~ired to determine which line scan (or ~cans) resulted in the test failure. The signal input to the oscilloscope can be connected to the Q output of flip-flop 341 on FIG. 13. It will be recalled that this flip-flop is set in the 1 state at the end of any line scan during which the shoulders test fails. Line 710 on the recorder monitor display indicates the line which is controlling the triggering of the trace on oscilloscope 708. Decoder 705 is manually changed until the trace, instead of being at the ground level, includes a positive-level (1) pulse. This is an indication that the particular line to which the oscilloscope is synced is that during which flip-flop 341 is set in the 1 state. This, in .. . . . . . ..
: :

~. ~ ... .. .:.. -0363t~5 turn, implies that it is the previous line during which the shoulders test was failed. Similarly, many test points throughout the data processor can be checked in this way.

Start/Stop Control, Counter Bank And Data Acquisition System --FIG. 24 . _ _ Start/stop control 37, counter bank 40 and data acguisition system 42, all shown within box 86 on FIG. 1, are shown in further detail on FIG. 24. The system standby-off switch 750 is connected to one of the inverting inputs of OR
gate 751. This input o the OR gate is thus ordinarily low.
To start the system, operation switch 750 is closed momentarily to extend a pulse through OR gate 751 to the inverting input of one-minute sample change timer 752. This timer is triggered by a negative step as soon as switch 750 is released.

With the triggering of the one-minute sample change timer, a one-minute positive pulse appears on conductor 61d. This conductor is extended to the automatic cell handling system. The pulse notiies the cell handling system to change the sample under test. This requires approximately 30 seconds.
After the sample is changed, the next solution with new cell - samples start to flow through the capillary tube. The first 30 seconds of the ce~l flow is allotted for cleaning the tube of cells from the previous sample. It is only after a flow of cells has occurred for approximately 30 seconds that the data processor is triggered into operation.

.. . . .. . .. . . . . .
.... .
~" '~ ' .

10363~5 The negative step at the output of timer 752 at the end of the one-minute pulse triggers one-shot multivibrator 753. A short pulse appears on START OF Rl~
conductor 61e. This pulse resets flip-flops 756 and 759, and all of counters 762-765. The flip-flops and counters must be cleared prior to the acquisition of data con-cerning the new sample. At the termination of the short pulse on conductor 61e, one-minute run timer 754 is triggered. The positive one-minute pulse on conductor 61c 0 i8 extended to an input of gate 422 on FIG. 15, and as described above enables this gate for a one-minute run interval. Until this gate is enabled none of the counter inputs, conductors 69a, 69b, 69c and 69d, can be pulsed~
by the data processor even though the circuitry within the data processor functions whenever cells flow through the capillary tube. Similarly, conductor 61c is extended to the circuit o FIG. 18 where it controls the re~etting of counter 447 and the inhibiting o gate 450.

At the end of the one-minute run, a negative step appears on conductor 61c and triggers one-shot multi-vibrator 755. A short positive pulse is generated on -END OF RUN conductor 61a. This pulse is extended to an input of OR gate 751. At the termination of the pulse, the negative step causes gate 751 to apply to a negative step to the inverting input of one-minute sample change timer 752, and the cycle begins all over again with the pulse on conductor 61d causing the automatic cell handling system to change samples.

,.. ~ ,,, ,. - ~ . .. .
, ~ ~
:,. . . :

'10363~35 Counter bank 40 includes the four counters 762-765 described above (the non-normal counter, the ambiguous counter, the cell counter and the histiocyte counter). It also includes a fifth counter 766, the sample counter. Each of the samples fed into the automatic cell handling system includes an identifying number, with all of the samples fed into the system in sequence having successive numbers. The present input of sample counter 766 is adjusted such that the count in the counter represents the identifying number of the first sample in any sequence, less unity. The first START OF
RUN pulse on conductor 61e, applied to the increment input of counter 766, increments the count in counter 766 so that the stored number corresponds to the identifying number of the first sample. Thereafter, each START OF RUN pulse on conductor 61e increments the counter so that the count in the counter will correspond to the identifying number of the sample under test. This procedure can be facilitated by having a preset input dial which is internally offset by a count of one so that the operator is not requlred to mentally subtract each time hP feeds a present number into the system.

Before the data acquisition system 42 is set into operation at the end of any run, it is necessary to verify that a sufficient number of cells have been processed for the results to be meaningful. The output of cell counter 764 ... . . . . .
- ~ , , on cable 75c is extended over cable 87 to the input of ad-justable adequate cell count decoder 760. The pre-set value stored in this counter represents the predetermined cell count which, if exceeded, is an indication that a sufficient number of cells have been examined. If a sufficient number of cells have been examined, decoder 760 energizes output conductor 767 which sets flip-flop 756 in the 1 state. At the start of any run flip-flop 756 is reset, and the Q output goes high tc enable one input of ~ate 757 and the Q output goes low to de-energize one input of gate 758. However, if during the run a sufficient number of cells are examined, gate 758 i6 enabled and gate 757 is disabled.

At the end of the run, the pulse on conductor 61a is extended to one input of each of the two gates. If a sufficient number of cells have been examined, conductor 61b i9 pulsed through gate 758. This conductor is extended to the trigger input of the data acquisition system. A
pulse on thi~ conductor in~tructs the data acquisition system to store all of the data contained in counters 762-766.
All of the data on cables 75a-75e is extended over cable 64 .. ~. . .
to the data acquisition system. The identifying number of-the sample just tested is recorded, together with the individual counts stored in counters 762-765. Also, as described above, it is of interest to verify that a particular sample was a lung sample if this was the case. In such a case, conductor 69e is high and this information is recorded as well.-Although not shown in the drawing, it is also possible to record the Vcyto and Vnuc or any other threshold levels used during the test of .

10363~3s each samp]e via a digital voltmeter and a programmable stepping switch.

On the other hand, if an insufficient number of cells have been examined, gate 757 extends the END OF RUN pulse from conductor 61a to the set input of flip-flop 759. The flip-flop is originally reset by the STAR~ OF RUN pulse on conductor 61e. However, at the end of the run, it switches to the 1 state and the Q output goes high. The Q output causes lamp circuit 761 to energize to inform the operator that the capillary tube must be cleaned, or that the sample is too sparse of cells.

, Alternate Non-Normal Loqic -- FIG. 27 . .
Referring to FIG. 15, the five conductors-on which non-normal pulses can appear are all coupled to five respective inputs of OR gate 418. These conductors represent respectively enlarged nuclear diameter ~276), in~ufficient cytoplasm shoulders ~334), enlarged N/C
ratio (380), non-normal cell shape (480) and excessive peak nuclear area (310). Each of the five~conditions is treated equally in that any one of them can control the categorization of a cell as non-normal. However, it is sometimes preferred to perform combinational weighting of the non-normal criteria in the decision-making process.
It may even be desirable to provide proportional weighting as well so that certain decisions are given more significance than others in the categorization of a cell.
The circuit o FIG. 27 outlines a scheme of this type as an alternate method for dexiving a non-normal signal - on conductor ~27 of FIG. 15.

.
.. .. . .. .... . ..

. . , .: - , . .

~036385 The five conductors just described are shown tied to six resistors to the left of FIG. 27.
There are only two efective inputs to OR gate 472, the energization of either of which causes a cell to be categorized as non-normal. Operational amplifier 470, with feedback resistor 469, serves as a summer. Po-tentiometer 468 controls the D.C. bias at the input of the summer. Three of the non-normal conductors (276, 310 and 334) are connected through respecti~e potentiometers 465, 466 and 467 to the input of the summer. The potentiometers may be set so that each input is given a different weight. The output of the summer is extended to one input of comparator 471, the other input to-which is ...
connected to potentiometer tap 476. The output of comparator 471 i8 ordinarily low. The output goes high only if the summer input to the comparator exceeds the threshold voltage determined by the setting of potentiometer 476. The setting of potentiometers 465-468 determine which combinations of the three inputs cause the output of the comparator to go high. For example, the potentiometer settings can be adjusted such that pulses on any two of the three inpu*
conductors, or pulses on all three, trigger OR gate 472, whereas a pulse on only one input conductor cannot trigger the OR gate.

Similarly, the other input to OR gate 472 is connected to the output of another comparator 473. The three ~ -131-`X

; ~ '`' ` `` ~'' -' -` `' 10363~s data inputs to ~ummer 474, whose output is connected to one input of comparator 473, are another combination of three of the non-normal conductor~ (310, 380 and 480). D~fferent weightings can be assigned to the pulses on these three con-ductors for controlling the final energization of conductor 427 .. .;, . .
--131a-.... ~ , . ., . ., , .. .. . . , . . _ ~ , .. ... .. . . . . .. . . . . .

i ~ - , . -~ ~ .. : . . .

1~363~s which is used to categorize the cell as non-normal. These weight-ings can clearly be completely different from those associated with the inputs to swmmer 470. Moreover, there can be a large number of these combinatorial weightings made, all feeding into OR gate 472.

Automatic SamPle Handlinq SYStem -- FIG. 25 The automatic sample handling system is shown in FIG. 25. Capillary tube 46 is shown to the right o~ the drawing. The tube (shown greatly enlarged in FIG. 6) is shown simply by dotted line 46' as extending to mounting plate 135. The capillary tube then extends downward below the mounting plate as shown in FIG. 25A.

The various samplés are contained in individual containers or tubes 866 which are open at the top. The upper end of each tube flares outward as shown most clearly for tube 866-2 in FIG. 25D to enable the tube to be trans-ported on guides such as 851 and 857. The guides are of two types -- stationary rails and conveyor belts. There are two conveyor belts (857 and 858) and they a~e preferably . ~ ,. .
coiled springs connected to form two closed loops.

There are six guide rail segments 851-856.
All of the rail segments are stationary except rails 855 and 856. Theue are fixed on elevator 773 which can be moved up and down as will be described below. The conveyor belts are extended around two sets of five pulleys 859, each mounted - ~ -132-. ,~

. .

on a shaft 860 extending to support surface 850. The two pulley shafts at the far right end of the support surface are extended, as shown by dotted lines 861, to respective pulleys 862. These pulleys are connected by respective belst 863 to drive pulley 864, which latter pulley is driven by motor 865. The two conveyor belts 857 and 858 thus rotate continuously .

.
.

., .
'' ' ' ;

,. . . .,~. " .;, : ., . . .

,' , ':' ~.; . .

.,., ' - :

,,' ' .
, .
-132a-~ .
.~, ' .

The sample tubes are placed at the start of the automatic transport system, tube 866-1 in FIG. 25A
being shown in this position. As belt 857 rotates, the tubes are moved in the direction shown by the arrows. Tube 866-2 is shown in a position nearing the working station.
Tube 866-3 is shown in phantom outline adjacent stop rail 853 just prior to elevator 773. As more and more sample tubes are placed at the starting end of the auto-matic transport system, they are conveyed toward the .
working station. They simply line up in back of tube 866-4 with conveyor belt 857 slipping past the queued tubes.
The conveyor belt system is capable of moving a tube only if it is not blocked by a tube or rail in front of it. .. .-, Tube 866-4 is shown on guide rails 855, 856 on elevator 773 A tube is transferred from the waiting position of tubé 866-3 to the working position of tube 866-4 by the energization of actuator 776. Piston 777 is forced out of the actuator, below.belt 857, to push the tube in the waiting position to the working position.
This is shown most clearly in FIG. 25B, a vièw from.
the right side of FIG. 25A. Tube 866-3 (shown in dotted lines) is in the waiting position up against guide rail 853. When piston 777 is pulsed outward in the direction shown by arrow 787, tube 866-3 is forced to the position shown by tube 866-3', the latter tube shown in section.
Thereafter, piston 777 is withdrawn so that the next sample tube can be conveyed into the waiting position.

.... . . .. . . . , . .. ~ . .... . . . . . . . .. . ..
', ' .
.
, Actuator 774 is then energized and piston 775, attached to the bottom of elevator 773, is forced upward. The entire elevator, including the pair of guide rails 855 and 856, moves upward as shown by arrow 788, such that capillary tube 46 extends down into the sampLe solution contained in the tube. Tube 866-3 ' is shown in the raised position by the dotted lines in FIG. 25B.

Following the run of the sample! actuator 774 is de-energized and piston 775 moves downward. The entire elevator drops and the tube just processed is moved to the position shown for tube 866-4 in FIG. 25A. When actuator 776 is energized once again and piston 777 moves in the direction of arrow 787, the next sample tube is pushed from the waiting position to the working position. This tube, in turn, pushes the tube t~en in the working position to the position shown by tube 866-3" in FIG. 25B. The latter tube i8 immediately transported by belt 858, in the direction of the arrows shown in FIG. 25A, to the final position shown by tube 866-6. As successi~e tubes are moved to the final position, they are taken away in order to aliow processing of additional samples.

It is desirable to maintain the solution in each sample tube in a state of agitation in order to obtain a ~ ..
.. ... . . . . .... . . . . . . . ......... . .. .. .. . . . . . .. ..
` ' . : . :, r~
10363E~s uniform dispersion of cells immediately prior to, and during, their aspiration through the capillary tube. A
series of small magnets 867 are shown on top of supporting surface 850, underneath the sample tubes, ~ust before they enter the working station. Each of these magnets is connected by a respective shaft, shown only by dotted lines 768, to a respective one of pulleys 769. The pulleys are in turn coupled by belt 770 to pulley 771 which is driven by motor 772. Magnets 867 rotate con-tinuously. Inside each sample tube, such as tube 866-2, there is a Teflon-coated magnet 778, as detalled in FIG. 25C. As each of magnets 867 rotates, it forces the magnet 778 in the tube above it to similarly rotate and agitate the cell solution. Thus immediately before each sample tube arrives at the working station, the solution it contains has been continuously agitated for at least 10 minutes. An exception to this would be the irst set of tub~s ~or the day or run. In this càse, the tubes may be kept in a standby condition by use of switch 750 in FIG.
24 to insure good cell dispersion.
. .
Another rotating magnet 867' is contained on the elevator itself, as shown in FIG. 25B, immediately below the sample being processed. The magnet is connected by shaft 768' to motor 786 contained in the elevator. The motor moves up and down with the elevator and insures that the solution which is being drawn through the capillary tube is continuously agitated.

'; . ' '` '''' . ' ' ' ~: ~
:~ , ~ 1()363~9s It will be recalled that the signal on conductor 61d in FIG. 24 from start/stop control 37 consists of one-minute positive pulses at two-minute intervals. Conductor 61d i~ ex-tended directly to the "down" control of actuator 774. The positive step on conductor 61d at the start of each run causes -elevator 773 to be pulled do~n, immediately following the processing of the previous sample. Conductor 61d is also extended to the input of delay unit 780. After 10 second~, the positive step at the output o the delay unit is extended over conductor 781 to the "out" input of actuator 776. After 10 seconds have elapsed, the elevator is in the down position, and piston 777 is forced out of the actuator to change the sample in the working position.
.. , ~ he output of delay unit 780 is also extended to the input of delay unit 782. After an additional 10 seconds have elapsed, the positive step at the output of delay unit 782 is extended over conductor 783 to the "in" input of actuator 776. Piston 777 is pulled into the actuator inasmuch as the sample on the elevator rails has by this time been changed and the piston should be withdrawn prior to lifting the elevator.

The output of delay unit 782 is also extended to the input of delay unit 784. The delay of this unit is also 10 seconds, and consequently after 30 seconds have elapsed following the initial positive step on conductor 61d the positive step appears on conductor 785 extended to the "up" input of actuator 774. When the positive step is applied to the "up"
input of the actuator, the elevator is raised. The elevator thus starts to rise 30 seconds after the positive step at the ' ~ .

~0363~5 start of each run appears on conductor 61d. It requires only a few seconds for the elevator to rise to a height sufficient for cells to be drawn through capillary tube 46. The cells in the carrier solution are drawn through the capillary tube for approximately 30 seconds before processing begins, as described above, in order to clear the capillary tube of the previous sample. Elevator 773 remains in the up position for approximately 1.5 minutes, during the last one minute of which the data obtained by the data processor is registered. After a total of two minutes have elapsed, the next positive step on conductor 61d starts the cycle all over again with the re-lowering of the elevator.

The automatic sample handling system greatly facilitates the rapid processing of successive samples. It will be apparent to those ekilled in the art that other types of automatic transport systems can be utilized.

Similar remarks apply to the other blocks of equipment in the illustrative embodiment of the invention.
For example, the line scans through a cell are generaged by a moving spot of light in the illustrative embodiment of th~ in-vention. However, it is also possible to generate an entire line of light without distinguishing between different points on the line in real time. A line profile through a cell could be determined by providing many photodetectors to detect the light transmitted through the capillary tube corresponding to different points on the line. Alternatively, a single photo-~ -137-.... . . . . . ... .. .. .. . . . .. . .
~ '- ` . .
.

1036~s detector could be provided in which case its output at any instant of time would represent the total cellular information along the line being examined. Successive outputs o~ such a photodetector would correspond to successive lines through the cell, and only a ~ingle waveform of the type shown in FIG.

:

, ' ' , . ' . ',, . ', .
'' . ~
.

.
.

-137a-, , , ,, .... ., . , . . ,: .. . ... . ..
'-~, ' 10363t~S
9 would be generated for each cell. For such an arranyement, it can be shown that, although each line of light impinges upon a cell in a direction transverse to the direction of flow, the photodetector output in real time corresponds to a line scan through the center of the cell in the direction of the flow. Similarly, instead of generating the ultra-violet and visible light with an electron beam impinging upon the phosphor-coated face of a cathode-ray tube, it is contemplated that a laser source or sources should be adequate for the same purpose. It will also be apparent to those skilled in the art that the ultra-violet and visible light signals can be combined, other than by subtracting one from the other, in order to enhance the distinguishability between nucleus and cytoplasm. Alao, staining may be practiced to enhance discernability in those cases where non-uniformity of staining is not a problem.
~In such cases, ultra-violet light may not be necessary for scanning.) Thus it is to be understood that numerous modi-fications may be made in the illustrative embodiment of the invention and other arrangements may be devised without departing from the spirit and scope of the invention.
-: . - - , ..

Claims (23)

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A system for analyzing biological cells compris-ing a scanning station, means for controlling the movement of biological cells in essentially single file past said scanning station, means for scanning said cells at the scanning station with line scans of light, means for detecting the instantaneous intensity of the light scanning said cells at the scanning sta-tion for deriving a signal proportional thereto, means for dis-tinguishing between nuclear and cytoplasmic cellular material along any line scan through a cell in accordance with the de-rived signal, and means responsive to said distinguishing means for categorizing a cell in one of at least two groups.

2. A system in accordance with claim 1 wherein the light is ultra-violet having a substantial portion of its energy at wavelengths below 300 millimicrons.

3. A system in accordance with claim 1 wherein said distinguishing means is operative to indicate the scanning of cytoplasmic cellular material responsive to the magnitude of said derived signal exceeding a first threshold level and is operative to indicate the scanning of nuclear cellular material responsive to the magnitude of said derived signal exceeding a second threshold level.

4. A system in accordance with claim 3 wherein said categorizing means includes means for generating a pulse of a first type during each line scan during the time interval when said derived signal exceeds said first threshold level and means for generating a pulse of a second type during each line scan during the time interval when said derived signal exceeds said second threshold level, means for determining the existence of pre-selected characteristics in a cell scanned at said scanning station in accordance with the widths of said first type and second type pulses during a line scan, and means for controlling the categorizing of a cell in one of said at least two groups in accordance with the determined characteristics after the scanning of said cell has been completed.

5. A system in accordance with claim 4 further in-cluding means for counting a predetermined number of successive line scans during which said derived signal does not exceed a predetermined threshold level for determining that a cell pre-viously being scanned is no longer being scanned.

6. A system in accordance with claim 5 further in-cluding means for enabling the operation of said categorizing means only if said derived signal exceeded a fixed threshold level during each of a predetermined number of successive line scans following the categorizing of the last cell.

7. A system in accordance with claim 5 further in-cluding means for inhibiting the operation of said categorizing means at the completion of the scanning of a cell if during the scanning of the cell said derived signal exceeded a fixed thres-hold level during a number of line scans exceeding a predetermined number.

8. A system in accordance with claim 1 further in-cluding means for scanning said cells with line scans of light having a portion of its energy at wavelengths above 300 milli-microns, means for detecting the instantaneous intensity of the light at wavelengths above 300 millimicrons transmitted through said cells for generating a signal proportional thereto, means for comparing the instantaneous magnitude of said generated sig-nal to a threshold level, means responsive to the instantaneous magnitude of said generated signal exceeding said threshold level for registering that the cell being scanned is a histiocyte, counter means for counting the total number of histiocytes scanned during the scanning of cells of a sample, and means responsive to the total count in said histiocyte counter means exceeding a pre-determined number at the completion of the scanning of cells of a sample for registering that the sample is of lung cells.

9. A system in accordance with claim 4 wherein two of said second-type pulses are generated during each line scan which passes through the nucleus of a cell and further includ-ing means for comparing the sum of the widths of the two second-type pulses generated during each line scan which passes through the nucleus of a cell with a predetermined threshold width, and means responsive to the sum of the widths of said two second-type pulses being less than said threshold width for characteriz-ing the cell as having insufficient cytoplasm shoulders.

10. A system in accordance with claim 4 wherein, dur-ing a line scan which extends through the nucleus of a cell, a single one of said first-type pulses and a pair of said second-type pulses are generated, said pair of second-type pulses occur-ring respectively earlier and later than said single first-type pulse, and further including first means operative responsive to the width of said earlier second-type pulse exceeding the width of said later second-type pulse during any line scan which ex-tends through the nucleus of a cell, second means operative re-sponsive to the width of said later second-type pulse exceeding the width of said earlier second-type pulse during another line scan which extends through the nucleus of the same cell, and means responsive to the operations of both of said first and second means for characterizing the cell as having a non-normal shape.

11. A system in accordance with claim 3 further in-cluding means for integrating that portion of said derived sig-nal which exceeds a predetermined threshold level during each line scan through a cell, means for comparing the magnitude of the integrated signal to a predetermined magnitude, and means responsive to the magnitude of said integrated signal exceeding said predetermined magnitude during any line scan through a cell for characterizing the cell as having a non-normal peak nuclear area.

12. A system in accordance with claim 3 further in-cluding means for generating a digitized signal from said de-rived signal, means responsive to said scanning means for gener-ating television synchronizing pulses, and means for combining said digitized signal and said synchronizing pulses to form a video signal representative of digitized cell information.

13. A system in accordance with claim 1 wherein said scanning station includes a quartz capillary tube through which said cells move, quartz objective and condenser lenses are disposed on opposite sides of said quartz capillary tube, a quartz plate supports said quartz capillary tube adjacent to said quartz objective lens, and a droplet of oil bridges said quartz objec-tive lens and said quartz capillary tube.

14. A system in accordance with claim 1 wherein a capillary tube is disposed at said scanning station, and means controls a flow through said capillary tube of a carrier solu-tion containing therein the biological cells to be scanned.

15. A system in accordance with claim 9 wherein means is provided responsive to said derived signal exceeding a first threshold level for indicating the presence of cytoplasmic cellu-lar material, means is provided responsive to said derived signal exceeding a second threshold level for indicating the presence of nuclear cellular material, and means characterize a scanned cell in accordance with the sequence that said derived signal ex-ceeds said first and second threshold levels during the line scans through the cell.

16. A system in accordance with claim 15 further in-cluding means for categorizing a scanned cell at the end of the scanning thereof in accordance with the operation of said char-acterizing means, and means for inhibiting the operation of said categorizing means in the absence of said derived signal having exceeded a predetermined threshold level during each of a prede-termined number of successive line scans.

17. A system in accordance with claim 13 wherein vacuum pump means is connected to the other end of said tube for controlling a flow through said tube of a carrier solution containing therein the biological cells to be scanned, means support a plurality of open-top containers each containing car-rier solution with biological cells of a respective sample therein, means transport said containers below said one end of said tube, means raises each container when it is disposed be-neath said one end of said tube to a height sufficient to immerse said one end of said tube in the carrier solution to enable said vacuum pump means to draw the carrier solution in said container through said tube, and means lowers each of said containers fol-lowing the flow of a portion of the carrier solution contained therein through said tube.

18. A system in accordance with claim 1 wherein means is provided for processing the derived signal in accordance with the generated signal to increase the signal/noise ratio of said derived signal.

19. A system in accordance with claim 4 further in-cluding means for comparing the width of said first pulse gen-erated during each line sweep to a predetermined threshold width, and means responsive to the width of said first pulse exceeding said threshold width for characterizing a cell as having a non-normal nuclear diameter.

20. A system in accordance with claim 4 wherein, dur-ing a line sweep which extends through the nucleus of a cell, a single one of said first pulses and a pair of said second pulses are generated, said pair of second pulses occurring respectively earlier and later than said single first pulse, and further in-cluding first means operative responsive to the width of said earlier second pulse exceeding the width of said later second pulse during any line sweep through a cell, second means operative responsive to the width of said later second pulse exceeding the width of said earlier second pulse during another line sweep through the same cell, and means responsive to the operation of both said first and second means for characterizing the cell as having a non-normal shape.

21. A system in accordance with claim 1 wherein a moni-toring station is provided, means transmits said derived signal to said monitoring station, means is provided for quantizing said derived signal, and means selectively controls the trans-mission of the quantized derived signal to said monitoring station instead of said derived signal.

22. A system in accordance with claim 15 wherein means is provided for weighting the various characteristics identified in a cell to categorize the cell as normal or non-normal.

23. A system in accordance with claim 15 wherein means is provided for registering the category of each cell, whose characteristics are determined by said detecting means, and means is provided for preventing the operation of said reg-istering means unless said derived signal has exceeded a prede-termined threshold level during each line sweep of a predeter-mined number of successive line sweeps following the categori-zation of the last cell.