JPS6132182A - Device for classifying cell - Google Patents

Abstract

PURPOSE:To decrease a measuring time by classifying cells roughly in the range of cells considered to be correct on the accuracy of classification by means of a low magnification microscope at first and applying accurate classification with high accuracy by means of a high magnification microscope as to cells not classified surely by the low magnification microscope. CONSTITUTION:A sample 2 subjected to normal dying is moved by a sample automatic loader 1, mounted on an XY stage of a low magnification microscope 4 and the XY stage is controlled by a controller 13. After automatic focusing, a camera 6 and a white corpuscle detector provided in the microscope 4 detect the position of white corpuslces scatterd in the visual field at each visual field and the picture of the white corpuscles, and after the picture signal is subjected to A/D conversion 7, the result is stored in a memory 8. The caracteristic such as size and shape is extracted from the picture in the memory 8 by a characteristic extracging device 9, and the extracted data is subjected to classifying processing by a white corpuscle identification computer 10 and the position of an abnormal corpuscle is stored in a memory 14. After a prescribed number of white corpuscle are idenditied, the sample 2 is carried to a high magnification microscope 18 and accurate information is picked up from the abnormal corpuscles.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a cell classification device, and particularly to a cell classification device that recognizes cell images and classifies normal or abnormal cells.

[Background of the invention]

Conventional cell sorting devices, especially blood cell sorting devices, place a blood-stained specimen on a high-magnification microscope with a magnification of 600x to i,00x, and use a TV camera to detect the blood cells scattered on the specimen. Blood cells were classified using a means of analyzing morphological images and identifying and classifying them. However, when identifying blood cells using a high-magnification microscope, the number of blood cells within the field of view of the microscope decreases, and the time required to find blood cells increases in proportion to the square of the field area. Therefore, by searching for blood cells, especially white blood cells, scattered at a rate of one cell per 300 μm square on a blood specimen, and then detecting and classifying at least 100 white blood cells per sample, the processing time is 0.5 times.
It takes 1 minute to 1 minute. Classifying 300 or more blood cells per sample, which is required for the statistical accuracy of test results, requires a longer processing time because the processing capacity of the classification device is limited. Therefore, in order to improve the statistical accuracy of the test results of the above-mentioned classification device, a new method for classifying white blood cells, which solves the problem of a decrease in the processing capacity of the classification device due to the long processing time, has been developed. An enzymatic chemical reaction method (or flow method) has been developed that uses an enzyme test sample that reacts with blood to identify blood samples from the light scattering intensity of white blood cells that are irradiated with laser light while in a fluid state. It was put into practical use as (for example, Japanese Patent Publication No. 59-853).

However, the aforementioned blood cell classification device that classifies blood cells morphologically uses the morphological characteristics of the blood cells, so accurate information can be obtained, whereas the flow-type blood cell classification device uses only light intensity. Because blood cells are classified by measurement, the test results correspond to the morphological classification method for normal leukocytes, but not for abnormal or immature leukocytes. Therefore, there is a problem that current clinical tests cannot be accepted as is.

[Purpose of the invention]

An object of the present invention is to provide a cell sorting device with improved processing ability that can classify as many cells as required for test accuracy even within a short measurement time.

[Summary of the invention]

As a result of various studies on cell image analysis, the present inventors found that when cells are identified and classified using a low-magnification microscope, the number of cells that fall within the field of view increases and the number of cells within the field of view increases. Since the time it takes to find one cell from another is proportional to the magnification of 2, the measurement time can be shortened. On the other hand, with a low-magnification microscope, the amount of light information from the cells is small, so it is possible to accurately detect the cells. Focusing on the fact that the characteristics of cells cannot be extracted and cell classification cannot be performed, and that using a high-magnification microscope increases the amount of optical information obtained from the cells but also increases the measurement time, we first use a low-magnification microscope. The present invention can be achieved by roughly classifying cells within the range of cells that are considered reliable in terms of classification accuracy, and by using a high-magnification microscope to perform precise classification with a high-magnification microscope for cells that cannot be reliably classified with a low-magnification microscope. The present invention was developed based on the belief that the object could be achieved. That is, the present invention includes a first detection means for irradiating light onto a cell specimen containing at least 1':) cells and detecting light information generated from the cells through a low magnification microscope; a first identifying means for comparing the amount of optical information obtained with a predetermined value indicating the characteristics of the cell and selectively classifying the cells that fall within the range of the predetermined value; storage means for storing positions on the cell specimen of cells that do not fall within the range; irradiating light to the cells existing at the positions stored in the storage means;
a second detection means for detecting light information generated from the cell through a high-magnification microscope, and comparing the amount of light information detected by the second detection means with a predetermined value indicating the characteristics of the cell, and determining the occupation of the cell. The present invention further comprises a second identification means for performing classification, and a display means for displaying the identification and classification results of the cells by the first identification means and the second identification means, so that a desired number of cells can be quickly identified in terms of statistical accuracy. It is characterized by being able to do things. The light may be selected depending on the type and characteristics of cells for which classification and measurement are to be performed. Depending on the type of light, a microscope and a detector for extracting optical information from this microscope are specified. For example, when a laser beam is used, the microscope is a fluorescence microscope, and when the inspection is performed, a fluorescence detector is used, and when visible light is used, the microscope is an ordinary microscope, and the detector is a TV left camera.

In the above-mentioned present invention, the principle on which the present invention is based will be further explained in detail using a cell sorting device that sorts blood as an example.

There are five types of white blood cells distributed in blood specimens: neutrophils, lymphocytes, monocytes, eosinophils, and basophils as normal white blood cells, and in addition, blast cells, myelocytes, and red blood cells in pathological specimens. Immature blood cells such as blasts, atypical cells, and lymphocytes appear. In a normal blood sample, the white blood cells are 200-300μ
They are scattered at a ratio of one per m square, and about 90% of them are occupied by neutrophils and lymphocytes, with a low incidence of monocytes, eosinophils, and basophils. Also, depending on the type of white blood cells stained with ordinary staining methods such as Wright staining and May-Giemsa staining,
Morphological characteristics are shown in the color of the cytoplasm and granules, the shape and size of the nucleus, etc. The average diameter of white blood cells on a specimen is approximately 10 μm.
~20 μm, and conventionally, white blood cell images were obtained using an optical microscope using an original lens with magnification of 40x to 100x.
The method involves dividing white blood cells into minute pixels of 5 μm square, analyzing the color density information of each pixel, extracting the characteristics of blood cells, and accurately classifying white blood cells.

Therefore, by using the typical characteristics of white blood cells, such as the color of cytoplasmic granules, the shape of the nucleus, and the size of the nucleus and cytoplasm, we first roughly and accurately capture only typical white blood cells, and even normal blood cells can be degenerated and inhibited. By selecting cells of unknown speed, most typical neutrophils and lymphocytes, which make up the majority of white blood cells, can be accurately processed at the rough identification stage. It is considered that a pixel resolution of about 1 μm or less is sufficient for rough identification processing, so for example, high-speed image capture using a combination of a low-magnification microscope using a 20x objective lens and a high-resolution linear array semiconductor detector with a resolution of 1 μm or less is recommended. The device can be expected to process blood cells 10 to 20 times faster than conventional methods.

Next, the positions on the specimen of the blood cells that were made unidentifiable during the rough separation process are memorized, and the specimen is transferred to a conventional high-resolution microscope to precisely analyze only the marked unknown cells. . The number of blood cells that require this precise identification process is generally small, and is 115 to 1/10 of the number of white blood cells processed per sample.
It is considered to be a degree.

In addition to the above, white blood cells can also be classified at high speed using a double-stained specimen of fluorescent staining and normal staining. Fluorescent staining is, for example, a staining method that has different degrees of staining depending on the type of white blood cells, or is specific to the type of white blood cells, and when irradiated with light such as blue laser light, it emits fluorescence specific to white blood cells. By analyzing the wavelength and intensity of the emitted fluorescence, the type of white blood cells and pathological information can be determined. Fluorescent staining uses fluorescent reagents such as acridine orange, which stains the nuclear material of white blood cells and has different staining intensity depending on the type, and monoclonal antibodies that display immune phenomena such as 1977 cells, 8977 cells, and young cells. be. The important point of this double staining method is that the blood is fluorescently stained and then normal stained to prepare a double stained specimen. Analyze and classify leukocytes by fluorescence analysis and immature cells, 1
Identify specific blood cells such as 977 cells and 8977 cells. At this fluorescence analysis stage, it is possible to roughly identify five categories of normal white blood cells, abnormal blood cells, specific blood cells to be examined, etc., and mark the position of each blood cell. Next, blood cells that cannot be distinguished using a fluorescence microscope and specific blood cells are morphologically identified using a high-magnification microscope. The combination of fluorescence analysis and morphological identification methods enables highly accurate age classification using composite information from the same blood cells that cannot be obtained with conventional classification, and also provides new pathological diagnostic information. It has the advantage that it can be obtained by comparing it with morphology.

Unlike conventional morphological classification, blood cell classification using a fluorescence microscope does not require extracting the shape of the blood cells, so the magnification of the microscope can be lowered even further, increasing sample processing capacity several times more than conventional methods. You can expect it to go up high.

[Embodiments of the invention]

Next, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

Figure 3 shows the field of view captured by a microscope in which white blood cells on a blood cell specimen are roughly classified and classified using a low-magnification microscope or a low-magnification fluorescence microscope using image processing or fluorescence analysis, and the fine classification and classification performed using a high-magnification microscope. Shows the field of view taken in.

In FIG. 3, 31 shows a blood specimen, 32 shows white blood cells scattered on the specimen and their positional coordinates xiyi, and 33 shows a low-magnification microscopic field. -34 shows a high magnification microscope field, and 35 shows a low magnification fluorescence microscope field. For example, if the magnification of the objective lens of a microscope is x20, the field of view is approximately 1000 μm square, and an average of 8 to 10
white blood cells will be scattered. In order to identify these white blood cells by image processing, the minimum unit of image capture, that is, the pixel 36, is 1 μm! There is a method of scanning the viewing screen using, for example, a 10241J near-array semiconductor detector as a detector for image capture. Another example is a method in which a 1024×1024 area array semiconductor camera is used to capture images of white blood cells at a higher speed while at the same time detecting their positions while moving the field of view intermittently.

In the case of a high magnification field of view of 34, the pixel 37 is (0, 25
μm)2 to (0.5 μm)2 is required, and a microscope with a ×40 to ×100 objective lens similar to conventional image capture is used. The edge of the low magnification fluorescence analysis field of 35 captures white blood cells as a source of fluorescence, and since it is only necessary to know their position, a lower magnification microscope can be used. For example, if a microscope with a ×10 objective lens is used, the field of view will be 2,000 μm in diameter, and 30 to 40 white blood cells will be scattered within this field of view. In this case, the white blood cell fluorescence detector is different from the case of image capture in step 33 mentioned above, and because it is necessary to increase the sensitivity of white blood cell fluorescence detection as much as possible, the pixel 38 is smaller than the size of the white blood cells. It is preferable. Therefore, a low-resolution, inexpensive semiconductor detector with 512 channels to 256 channels can be used as a near array detector corresponding to pixels (4 μm) 2 to (8 μm) 2 that are slightly smaller than the diameter of a white blood cell nucleus. 512X512 or 25 to further increase processing speed
A camera with a 6x256 area array semiconductor detector can also be used. Generally, when using a camera, the scanning time of the field of view is 16 m (8), so × 10
When classifying leukocytes by fluorometry using a fluorescence microscope with a double objective, approximately 30 leukocytes in the field of view can be analyzed within a time of at least 15 m. Even if one field of view is analyzed in 32 m5ec, which is twice that time, including the time required for moving the microscope stage, automatic focusing, and base information acquisition processing, approximately 900 white blood cells per group will be analyzed. Therefore, the processing speed of the classification system of the present invention is considered to be influenced by the classification speed of the high-power microscope at the subsequent stage. Now, assuming that the total number of leukocytes to be classified in the second stage is 10, when 1,000 white blood cells are processed in the first stage fluorescence analysis, 100 white blood cells will be subjected to image processing in the second stage. The image processing speed of the second stage is shorter than that of the conventional method because there is no need to search for white blood cells scattered on the specimen as in the conventional method, and the stage movement can be controlled in the shortest distance to the white blood cell position marked in the first stage. However, a minimum of 0.1 (8) per blood cell is still required. It takes 1 (Hec'l) to process 100 blood cells, and even including the specimen exchange time, it is possible to process more than 200 specimens per hour. The processing capacity is approximately 20 times higher than the high-speed processing of 100 samples per hour when processing white blood cells. However, the overall processing capacity is determined by the number of blood cells that require subsequent image processing. Ru.

FIG. 4 shows the intensity of fluorescence emitted when leukocytes stained with acridine orange as fluorescent staining are irradiated with 488 nm blue light from an Ar laser as excitation light and the frequency distribution of various leukocytes. In the figure, (a) shows a wavelength of 530n.
The spectrum of leukocytes for green fluorescence near m is (b
) The figure shows the spectrum of white blood cells for red fluorescence around a wavelength of 650 nm. Also, L is lymphocyte, B is basophil,
M indicates monocyte, E indicates eosinophil, N indicates neutrophil, and IMM indicates immature cell.

From the two-dimensional intensity spectra of these two fluorescences, it is possible to classify five typical types of white blood cells and immature cells.

FIG. 1 shows an implementation of the cell sorting device according to the present invention, in which a blood sample is normally stained and white blood cells are classified at high speed using two microscopes with low magnification and high magnification and an image processing method. FIG. 1 is a configuration diagram showing a system.

In the figure, each specimen 2 is moved from a specimen autoloader 1 to a stage of a low magnification microscope 4 via an ID reader 3 on the way. Low magnification microscope 4 has the same microscope @4.
A light source 5 that emits visible light and a camera 6 are connected. The camera 6 is connected to an A/D converter 7,
The A/D converter 7 is connected to an image memory 8. The image memory 8 is connected to a feature extractor 9;
This feature extractor 9 is connected to the blood cell identification computer IOK. The blood cell differentiation computer 1o is connected to an ID classification result memory, and this memory is connected to a printer 12.
It is connected to the.

An X-Y controller 13 is connected to the low magnification microscope 4, and a sample ID unknown blood cell position memory 14 is connected to the controller 13.

The ID reader 3 and the blood cell identification computer 1° are the specimen I.
D is connected to an unknown blood cell position memory 14, and this memory 14 is connected to an ID checker 15. An ID reader 16 is connected to this checker 15. An XY controller 16 is connected to the ID checker 15, and a high magnification microscope 17 to which a light source 19 is connected is connected to the controller 16. A camera 18 is connected to the high magnification microscope 17.
is connected to the A/D conversion group 20.

The A/D converter 20 is connected to an image memory 21, which in turn is connected to the feature extractor 9.

Next, the operation of this embodiment will be explained. First, a white liquor specimen 2 stained with May and Giemsa is moved by a specimen autoloader 1, and while the specimen ID is read by an ID reader 3 along the way, it is transferred to the X of a low magnification microscope 4 which is irradiated with visible light from a light source 3. -Mounted on Y stage. This X-
The Y stage is operated under the control of the controller 13, and after automatic focusing, the camera 6 detects the positions of white blood cells scattered within the field of view and images of the white blood cells for each field of view by a white blood cell detector provided in the microscope 4. . The image signal of the white blood cells detected by the camera 6 is divided into three color density signals of red (R), green (G), and t(B) within the camera 6 for each position of each white blood cell. are respectively digitized by an A/D converter 7 in order to process them in digital quantities.

This digitized signal is stored in the image memory 8. Reveal! JP! While the stage 4 moves to the next field of view and prepares to capture the image, features such as the size, shape, and color density of the white blood cells are extracted from the white blood cell image in the image memory 8 by the feature extraction device 9. .

Next, the data extracted by the feature extractor 9 is stored in the blood cell identification computer 10, where it is subjected to calculations such as white blood cell area calculation, maximum and minimum value detection, concentration histogram, perimeter, and number of segments, and is processed using a discrimination function. Classification of various white blood cells is performed by comparing with predetermined values stored in advance. Through such a process, a designated number of white blood cells required for the accuracy of blood cell classification test is identified, and the results are stored in the classification result memory 11 in correspondence with the ID of the specimen. At the same time, the blood cell classification computer 10 stores the position of the blood cell determined as an abnormal or unknown cell in the position memory 14. Next, the specimen 2 is transported by the specimen autoloader 1 to a high-magnification microscope 18 that is irradiated with visible light from a light source 19, and the specimen I
D is confirmed by ID IJ-da 16 and checker 15. Next, according to the classification by the low magnification microscope 4, the position on the specimen specimen is classified as an unknown ball or an abnormal ball by the unknown ball position memory 14.
The X-Y stage controller 16 calls out the blood cells stored in the X-Y stage controller 16, and the detailed information of the blood cell image is sent to the image memory 21 through the camera 18 and A/D converter 20.
Incorporate into. The movement of the X=Y stage is performed by the control device in the X-Y controller 16, but since the positions of the unknown spheres and abnormal spheres on the specimen are known in advance in the memory 14, the movement of the X-Y stage is fast. It can be carried out. X-
While the Y stage controller 16 calls the next unknown or abnormal ball and prepares to capture its image, the white blood cells in the image memory are processed by the feature extraction 9 and the blood cell job specific computer 10.
The timeshares are precisely identified and classified by the timeshare, and the results are sent to 15 result memories.

The operations of the camera 18, A/D converter 20, and image memory 21 are the same as those of the camera 6, A/D converter 7, and image memory 8 used for inspection classification using the low magnification microscope 4.

In the above manner, all unknown balls or abnormal balls are X-Y,
7 is read by the controller 16 and finely identified by the high magnification microscope 17, and the result is reported as the final result of the specimen by the printer 16.As described above, according to this embodiment, rough identification classification of blood cells is performed. By using the two-stage method of precise identification and classification in parallel for a large number of samples, it is possible to expect a processing capacity that is twice to several times that of conventional methods, even when processing 9,300 or more white blood cells per sample. Therefore, cell classification of a large number of blood samples can be performed quickly and accurately in clinical tests at hospitals. This is extremely effective, especially during emergency inspections.

In this embodiment, the specimen 2 is transferred to the high magnification microscopes 1 and 7 by the specimen autoloader 1, but the high magnification microscope can also be transferred to the specimen portion if necessary.

FIG. 2 is a configuration diagram showing an embodiment different from the embodiment shown in FIG. 1 of the cell sorting device according to the present invention.

The difference between this example and the example shown in FIG. 1 is that specimen 2 was double-stained with fluorescent staining and ordinary staining, the cell specimen was irradiated with laser light, and white blood cells were detected using a low-magnification fluorescence microscope. The point is that it performs a rough classification.

The components of this embodiment and the embodiment of FIG. 1 are as follows.

That is, a laser light source 22 is provided in place of the visible light light source 5 in the embodiment of FIG. 1, and this light source 22 is connected to a low magnification fluorescence microscope 23. The microscope 23 is equipped with a fluorescence detector 2 instead of the camera 6 shown in FIG.
4 is connected to the detector 24, and an A/D converter 7 is connected to the detector 24. A fluorescence intensity analyzer 25 is connected to the A/D converter 7 instead of the image memory 8 in FIG. 1, and this analyzer 25 is not connected to the feature extractor 9 in FIG. Connected to blood cell identification computer.

Next, the points of operation of this embodiment that are different from the operation of the embodiment shown in FIG. 1 will be explained. Fluorescent pigments within the white blood cells excited by the excitation light of the laser beam from the light source 22 emit fluorescence, which is detected by the fluorescence detector 24 .

The fluorescence signal detected by the detector 24 is analyzed into various fluorescence intensity signals by green, red, etc. fluorescence filters provided in the detector 24, and digitally processed by the A/D converter 7 to convert into digital quantities. is converted to The various fluorescence intensity signals converted into digital quantities are
The intensity is analyzed by the fluorescence analyzer 25, and the type of each white blood cell is classified by the blood cell identification computer 26 based on the fluorescence intensity spectrum of each wavelength, and the results are stored in the memory 11. The location of blood cells that are determined to be difficult to classify or immature cells by the blood cell type computer 26 is stored in the memory 14, and images of the white blood cells are obtained using the high magnification microscope 17 as in the embodiment shown in FIG. A detailed analysis will be carried out.

Classifying blood cells by a combination of fluorescence analysis and image processing as described above is not only a two-step microscopic analysis as in this example, but also a combination of fluorescence analysis and image processing performed simultaneously on one microscope. It is also possible by In such a case, rather than increasing the processing speed of blood cell classification, it is possible to increase the clinical diagnostic information based on the composite information of blood cells, thereby increasing the accuracy of sample diagnosis.

As described above, according to this embodiment, since blood cells are classified using a laser beam, a microscope with lower magnification than that of the first embodiment can be used at the stage of rough cell classification.

Therefore, classification processing time can be shortened while maintaining classification accuracy even more than in the first embodiment.

〔Effect of the invention〕

As explained above, according to the present invention, it is possible to quickly classify the number of cells required for the accuracy of cell classification test results.

Brief description of objects Figure 1 Figure 2 is a configuration diagram showing the system of an embodiment of the cell sorting device according to the present invention, Figure 3 is a diagram showing the position of blood cells in a blood sample and the field of view under a microscope. FIG. 4 is a graph showing the relationship between leukocyte frequency in blood and fluorescence intensity.

4...Low magnification microscope, 6...Camera, 8.21...
・Image memory, 9...% slight oil utility, 10...Blood cell identification computer, 11...ID, classification result memory, 1
4... Unknown original position memory from the specimen, 17... High magnification microscope, 23... Low magnification fluorescence microscope, 24... Fluorescence extractor, 25... Fluorescence intensity analyzer.

Claims (1)

[Claims] 1. A first detection means for irradiating light onto a cell specimen containing at least one cell and detecting light information generated from the cell through a low magnification microscope; and detection by the first detection means. comparing the amount of optical information obtained with a predetermined value indicating the characteristics of the cell,
A first step of identifying and classifying the cells that fall within the range of the predetermined value.
an identification means, a storage means for storing a position on the cell specimen of a cell that does not fall within the range of the predetermined value in the first identification means, and a method for applying light to the cell existing at the position stored in the storage means. a second detection means for irradiating the cell and detecting optical information generated from the cell through a high magnification microscope;
a second identification means for identifying and classifying the cells by comparing the amount of optical information detected by the detection means with a predetermined value indicating the characteristics of the cells; and a second identification means for identifying and classifying the cells; 1. A cell classification device comprising: display means for displaying identification classification results. 2. In the invention as set forth in claim 1, the light irradiated onto the cell specimen by the first and second detection means is visible light, and the optical information is image information from the cells. Characteristic cell sorting device. 3. In the invention as set forth in claim 1, the light irradiated onto the cell specimen by the first detection means is a laser beam, and the optical information is fluorescence information from cells that have absorbed the laser beam. A cell sorting device featuring: 4. A cell sorting device according to any one of claims 1 to 3, wherein the cells are blood cells.