CN100588941C - System for determining the stain quality of slides using scatter plot distributions - Google Patents

System for determining the stain quality of slides using scatter plot distributions Download PDF

Info

Publication number
CN100588941C
CN100588941C CN200480016334A CN200480016334A CN100588941C CN 100588941 C CN100588941 C CN 100588941C CN 200480016334 A CN200480016334 A CN 200480016334A CN 200480016334 A CN200480016334 A CN 200480016334A CN 100588941 C CN100588941 C CN 100588941C
Authority
CN
China
Prior art keywords
sample
screening system
processor
biological
biological screening
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN200480016334A
Other languages
Chinese (zh)
Other versions
CN1806166A (en
Inventor
卡姆·L·翁
路易丝·伊森施泰因
戴维·扎赫尼塞尔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Cytyc Corp
Original Assignee
Cytyc Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Cytyc Corp filed Critical Cytyc Corp
Publication of CN1806166A publication Critical patent/CN1806166A/en
Application granted granted Critical
Publication of CN100588941C publication Critical patent/CN100588941C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Investigating Or Analysing Biological Materials (AREA)

Abstract

Systems for determining the stain quality of a plurality of biological specimens. A number of objects of interest are identified in a biological specimen. A first feature of each object of interest (e.g., nuclear area) and a second feature of each object of interest (e.g., nuclear integrated optical density) are measured, and a scatter plot of the first and second features is generated. The stainquality of the specimens are determined based on the spread of the distribution of points within the scatter plot.

Description

The system that uses scatter plot distributions that microslide is classified
Technical field
The present invention relates to be used for the system of analysis of biological samples, more specifically, relating to and being used to use scatter plot distributions is normal, unusual or suspicious system with classifying biological samples.
Background technology
In health care industry, often need laboratory technician (for example, cytotechnologist) pair cell sample to check the existence of particular cell types.For example, at present need be to cervical guide Pasteur (Papanicolaou, Pap) the smear slides inspection malignant cell or the precellular existence of cancerating.Since its since introducing before five more than ten years, the Pap smear has become the strong instrument that is used to check cervix cancer and precancerous lesion.In that time, the Pap smear has obtained trust owing to the mortality ratio with cervix cancer has reduced up to 70%.Yet falling suddenly that this mortality ratio once had slowed down, and the mortality ratio of this disease that can prevent in fact kept constant in the U.S., and from the eighties approximately annual 5,000 people since mid-term.Therefore, because cancer is found too late, annual 15,000 quilt diagnosis suffer among the women of cervix cancer, still have about 1/3rd death.Another reason that causes concern is, national cancer association (National Cancer Institute) data presentation, and since 1986, the white race women's of the right side of fifty the cervix cancer incidence of disease is annual to increase by 3%.
Many factors may cause this current threshold value to rise, and wherein important factor is many women, and especially the people at highest risk does not still participate in conventional cervix cancer examination (screening).Another has caused that the factor that works of very big concern is the limitation of traditional Pap smear method itself.
The reliability of cervix screening method and validity avoid the ability (specificity) of false positive diagnoses to weigh by the ability (sensitivity) of pathology before its cancer diagnosis simultaneously.Further, these standards depend on the accuracy that cytology is explained.Traditionally, the virologist can come biological sample is carried out single cell analysis by observing each nuclear characteristic, or comes biological sample is carried out structure analysis (contextual analysis) by seeking characteristic pattern in the eucaryotic cell structure on being presented on microslide.
Even when not only carrying out single cell analysis but also carrying out structure analysis, traditional Pap smear also has the false negative rate of scope between the 10-50%.This be to a great extent since the technician must to check a large amount of cells and object (normally nearly 100,000 to 200,000) a small amount of pernicious or worsen precellular may the existence and cause to determine.Therefore, the Pap smear tests, and other need the chemical examination of detailed inspection biomaterial, owing to the fatigue that the technician bears has been suffered high false negative rate.
In order to make this checking process easy, developed automatic screening system.In a canonical system, the operation imager is to provide a series of images of cytological specimen slide, and each image illustrates the different piece of microslide.Processor is handled these images then and is prejudged information to provide about the quantitative and state of an illness of this sample.Processor can fill order's cell analysis or structure analysis during this diagnostic message is provided, or the two is all carried out.When fill order's cell analysis, processor can be defined as such cell characteristic the tenuigenin area than (cytoplasmicarea ratio), nuclear integrated optical density (nuclear integrated optical density), mean nuclear optical density (average nuclear optical density), nucleus structure (nuclear texture) and tenuigenin tone (cytoplasmic hue).When execution architecture is analyzed, processor such architectural feature can be defined as in (i) bunch (cluster) the nucleus average headway from; The (ii) standard deviation of tenuigenin gray-scale value; The (iii) mean value of bare nucleus (barenucleus) area; The (iv) standard deviation of bare nucleus area; (v) bunch average cell matter area; And (vi) iuntercellular inverse distance.In conjunction with the result of single cell analysis and structure analysis, to obtain diagnostic result.
In some automatic screening systems, processor uses diagnostic message to be described as " normally " or " suspicious " microslide.In clinical laboratory, cytotechnologist only checks being labeled as those suspicious microslides, has reduced workload like this.In other automatic screening systems, processor uses diagnostic message to be depicted as normal or suspicious biomaterial in each sample.That is, processor focuses on the most relevant cell with the notice cytotechnologist, thereby may realize getting rid of the further inspection to remaining cell.In the case, screening apparatus uses diagnostic message to determine the most relevant biological object (for example, most probable has cell and the nucleus that meets malignant cell or worsen precellular attribute), and the position on microslide.This positional information is sent on the microscope then, and microscope turns to these positions automatically and the center is scheduled on these biological objects, is used for cytotechnologist and checks.Then that mark electronically is the most relevant biological object of cytotechnologist (for example, have meet pernicious or worsen the object of precellular attribute) is used for virologist's further inspection.
Usually, owing to make technician's notice concentrate on those suspicious slides or concentrate on the more relevant object of minority in each microslide, therefore proved that the use of automatic screening system is successful.Yet, in fact do not have aut.eq. and reached with abundant low cost and eliminated false negative and false-positive result significantly.In addition, also must consider commercial aspect.Expense by the inspection cell sample (for example Pap smear sample) of laboratory burden depends on that at least in part the technician checks the time that each microslide spends.That is, the microslide that cytotechnologist must check is many more, and perhaps the technician checks that the time of each microslide cost is many more, and then the labour costs that bears of laboratory is just many more.On the contrary, the microslide that the technician must check is few more, and the technician checks that the time of microslide cost is few more, and then the money that can save of laboratory is many more.
False-positive generation is to be caused by the biological sample that has used the substandard products coloring agents sometimes.For example, the technician may fail to follow the correct rules that are used for using to biological sample suitable coloring agent, even if perhaps correctly used, also may surpass the storage life of coloring agent.Frequently, sample just is marked as after they are by cytotechnologist's hand inspection and has used the substandard products coloring agent, thereby has wasted valuable time.
Summary of the invention
In one embodiment of the invention, a kind of method that biological sample is classified comprise a plurality of objects of being identified in the biological sample (objects of interest, OOI).This can be by obtaining biological sample image and handle these images and realize with identification OOI.This method also comprises first (for example, nuclear area) and second (for example, nuclear integrated optical density (the IOD)) feature of measuring each OOI, generates the scatter diagram (scatter plot) of the first measured feature to the second measured feature then.Then at least in part based on the distribution of putting in the scatter diagram with sample classification (for example, be divided into normal or suspicious).In one embodiment, be normal based on the scope (spread) of a distribution with sample classification, the scope of this point distribution has under the situation of normal cell attribute less relatively at sample, and has under the situation of abnormal cell attribute relatively large at sample.In the case, be lower than threshold value as the fruit dot distribution range, then sample will be classified as normally.Can also be suspicious based on the distribution of putting in the scatter diagram with sample classification at least in part.In the case, surpass threshold value as the fruit dot distribution range, then sample will be classified as suspicious.This classification can be applied in the various situations any one, for example, removes normal slide or provide standard for other devices from inspection subsequently.
In one embodiment, normal specimens is marked as does not need further inspection, in the case, can skip the subsequent examination of this sample easily, thereby has reduced the quantity of the microslide that cytotechnologist must check.On the contrary, suspicious specimen is marked as needs further to check.In can be in every way any one carried out the subsequent examination of sample, but preferably produces a plurality of fields of interest (field of interest) that comprise OOI, so that cytotechnologist's notice focuses on suspect object.
In one embodiment, use in the various technology any one to measure a distribution range.For example, quantification step can comprise with a kind of geometric configuration (for example, ellipse) coming these points of match (fit), determining at least one dimension (dimension) of this geometric configuration and the functional value that calculates dimension.For example, (under the situation of two dimension) can calculate synthetic scoring (composite score).Then can be with value and the threshold of being calculated.
This method comprises at least one additional features of measuring samples alternatively, wherein, is normal based on other measurement features with sample classification further.As limiting examples, this other information can be used for determining whether scatter diagram is reliable.Additional features can be any feature that comprises in the mean nuclear optical density of OOI sum in the sample, the artefact's (artifact) in the sample sum and OOI.
In another embodiment of the present invention, provide a kind of biological screening system that biological sample is classified of being used for.This system comprises the imaging station, is used to obtain the image of sample and generates view data from image.This system also comprises at least one processor, be used for handling (for example, filtering) view data be identified in a plurality of OOI in the biological sample, measure each OOI first feature, measure each OOI second feature, generate first measurement features to the scatter diagram of second measurement features and based on the scope of a distribution sample is classified at least in part.Can carry out these functions with the mode identical with aforesaid way.If produce fields of interest (FOI) alternatively, then this system comprises automatic or semi-automatic inspection post, be used for respect to each FOI scanning field of view (field of view, FOV), in each FOI, to present one or more OOI.
Description of drawings
Accompanying drawing shows the design and the effectiveness of embodiments of the invention, wherein, and identical reference number indication similar elements, and in the accompanying drawings:
Fig. 1 is the planimetric map that is loaded with the standard microscope slide of biological sample;
Fig. 2 is the planimetric map of the biological screening system of constructing according to one embodiment of present invention;
Fig. 3 is the scatter diagram of nuclear area (y-axle) the pair cell nuclear integral optical density (x-axle) of the objects found in normal biological sample;
Fig. 4 is the scatter diagram of nuclear area (y-axle) the pair cell nuclear integral optical density (x-axle) of the objects found in suspicious biological sample;
Fig. 5 is another planimetric map that is loaded with the standard microscope slide of biological sample;
Fig. 6 is the planimetric map of the biological screening system of constructing according to another embodiment of the present invention; And
Fig. 7 is by the fields of interest shown in the microscopical visual field of using in the system of Fig. 5 and the view of mark indicator.
Embodiment
With reference to Fig. 2, the biological screening system 10 of structure has according to one embodiment of present invention been described.System 10 is set and is used to handle a series of microslides 12, biological sample 14 (the best illustrates in Fig. 1) is categorized as " normally " (will not need cytotechnologist's inspection) or " suspicious " (will need cytotechnologist's inspection).
Although it is normal or unusual that system 10 can be used for any classifying biological samples, system 10 is specially adapted to presenting of cell cervix or vagina material, such as common those that find on the Pap smear slides.In the case, these cells (for example can reflect abnormality, cytolysis, atrophy, infection, impaired), pernicious or precancer, such as Low Grade Squamous Intraepithelial Lesions (LGSIL) or HighGrade Squamous Intraepithelial Lesions (HGSIL), and by other cell types of TheBethesda System for Reporting Cervical/Vaginal Cytologic Diagnosis definition.Biological sample 14 will be placed on the microslide 12 as thin cellular layer usually.Preferably, the cover glass (not shown) is attached to sample 14, thereby sample 14 is positioned on the microslide 12.Sample 14 can be with any suitable coloring agent dyeing, such as pap staining agent (Papanicolaou stain).
In processing procedure, system 10 obtains the image by the biological sample 14 of microslide 12 carryings, and microslide is contained in the box (not shown) usually.In imaging process, microslide 12 in a continuous manner from box separately taking-up, digital imagery, turn back in the box then.Then, system's 10 analysing digital images (analyzing continuously when handling each microslide, perhaps off-line analysis when finishing the processing of all microslides) are normal or suspicious so that sample 14 is categorized as.
In carrying out the imaging operation process, system 10 comprises camera 16, microscope 18 and motorized stage 20.Camera 16 is caught the enlarged image of microslide 12 by microscope 18.Camera 16 can be any one in the various traditional cameras, such as the charge-coupled device (CCD) camera, it is cooperated separately or with other elements (for example modulus (A/D) converter), the numeral output that can produce enough resolution for example has the digital picture of the resolution of 640 * 480 pixels to allow the handling image of being caught.Preferably, each pixel is converted to eight place values (0 to 255) according to its optical transmittance, and " 00000000 " is the designated value that is used for through the minimum amount of light of pixel, and " 11111111 " are the designated values that is used for through the maximum amount of pixel.Microslide 12 is fixed on the motorized stage 20, and motorized stage is with respect to the observation domain scanning microslide 12 of microscope 18, and camera 16 is caught image above the zones of different of biological sample 12 simultaneously.The shutter speed of camera 16 is preferably higher relatively, makes the quantity of image of sweep velocity and/or acquisition to maximize.
In the process of execution analysis function, system 10 comprises image processor 22, and it carries out various operations to the digital picture that obtains from camera 16.For example, image processor 22 can be carried out tentatively and cut apart (primary segmentation), secondary splitting (secondarysegmentation), classification and contextual analysis operation, is normally or suspicious to determine each microslide 12.
In preliminary cutting operation, image processor 22 is removed the artefact from further consider item.This by image processor 22 by those in (mask) digital picture of shielding from further consider item because its brightness and unlikely be that nuclear pixel is finished.Residual pixel in the digital picture forms " spot (blob) " with different shape and size.The 22 pairs of spots of image processor are carried out to corrode and are handled then, only are the spot of less pixel and the arrowband that stretches out or be connected with adjacent spot from spot to remove diameter from further consider item.Image processor 22 determines that according to the quantity of pixel in the spot each pixel in the image is individual subject or bunch (clustered) object then.For example, the spot that has more than 590 pixels can be considered to a bunch object, and has 590 or can be considered to individual subject less than the spot of 590 pixels.For individual subject, will no longer further consider the spot that can not satisfy specific criteria, this specific criteria relates to the total area, girth area ratio, optical density standard deviation and gray level average pixel value.
In secondary segmentation operation, it unlikely is the spot of individual cells or bunch cell that image processor 22 is removed.For individual subject, image processor 22 is carried out a series of erosion operations of removing small object and elimination protrusion (projection) from the residue spot, and removes the dilation procedure in hole from the residue spot.For bunch object, the edge of image processor 22 sharpenings (sharpen) object is to provide the border of qualification.Image processor 22 selects to have the individual subject of the highest nuclear integrated optical density (IOD) and the individual subject that satisfies specific shape need then from bunch object that limits.The individual subject of extracting from bunch object will be marked as a bunch extraction object.
In sort operation, image processor 22 is measured various features for each individual subject and a bunch object, calculates the object score of each object then based on the measured value of these features.Based on this scoring, it may be artefact's individual subject and bunch object that image processor 22 is removed.Those remaining individual subject are considered to objects (OOI).Image processor 22 is that its nucleus IOD estimates OOI ((corrected) of correction or uncorrected) then, and according to the IOD value OOI is graded.Image processor 22 is also measured the nuclear area of each OOI for structure analysis, this will be described below.For each digital picture, image processor 22 then at storer 24 stored OOI and grading thereof and the nuclear area of measuring and nucleus IOD as frame data record (FDR).In the illustrated embodiment, for each microslide 12 obtains about 2000 digital pictures, and therefore in storer 24, will be about 2000 FDR of each microslide 12 storage.
In contextual analysis operation, image processor 22 obtains nuclear area and the nucleus IOD of the OOI of 100 high ratings from FDR, and generates the scatter diagram of nuclear area (y-axle) the pair cell nuclear IOD (x-axle) of each OOI.For example, scatter diagram shown in Figure 3 is to generate from the microslide that is born with Normocellular biological sample, and the scatter diagram shown in Fig. 4 is to generate from the microslide that is born with paracytic biological sample.Based on this scatter diagram, image processor 22 will be categorized as sample 14 normal and biological sample 14 will be labeled as the further inspection that does not need cytotechnologist, perhaps be categorized as sample 14 suspicious and biological sample 14 will be labeled as the further inspection that needs cytotechnologist.
Generally speaking, as shown in Figure 4, the biological sample that contains the cell that reflects abnormality often all has broad distribution (that is, spaced point trends towards forming the loosely bounded domain) aspect two of nuclear area and nucleus IOD.On the contrary, as shown in Figure 3, contain Normocellular biological sample and trend towards aspect two of nuclear area and nucleus IOD, all having narrow and small distribution (that is, spaced point trends towards forming closely bounded domain).Therefore, the biological sample 14 that is positioned on the microslide 12 is provided is normally or suspicious indication to the scope of scatter point distribution.The difference of the scatter diagram shown in Fig. 3 and Fig. 4 comes from abnormal cell will be often have the fact of the nucleus IOD (because increase of chromatin quantity) of bigger nuclear area and Geng Gao than normal cell, and the scope that therefore will make spaced point away from produce by normal cell scatter point distribution limited is the zone that clearly limits on the contrary.
In can making in various manners any one measured scatter point distribution, but in a preferred embodiment, as shown in Figure 3 and Figure 4, spaced point carried out geometric fit.Especially, often will be slightly oval from the scatter point distribution of normal slide, therefore oval-shaped match is preferred.Closely bounded is oval different and make biological sample little by little become more suspicious along with scatter point distribution and this, utilizes this total principle, and will provide biological sample by the size of the measured ellipse of the minor axis of ellipse and major axis is normally or suspicious indication.The dimension of major axis and minor axis can be weighted and be used to calculate synthesize and mark.
In case calculate scoring, then itself and threshold value compared.Suppose in some way to calculate scoring, make the suspicion of biological sample increase along with the increase of scoring, the scoring that is higher than threshold value is suspicious with the indicator organism sample, and the scoring that is lower than threshold value is normal with the indicator organism sample.In the case, if mark less than threshold value, image processor 22 will be categorized as sample 14 normally, and if mark greater than threshold value, will be categorized as sample 14 suspicious.
Two competition factors are depended in selection that it should be noted that threshold value: the generation of the generation of false negative (that is, contain be identified as normal paracytic microslide) and false positive (that is, contain be identified as suspicious Normocellular microslide).Usually, because the possibility of error diagnosis disease wants character serious than the possibility that needs cytotechnologist to carry out unnecessary inspection, therefore false-negative generation more can't stand than false-positive generation.Thus, then threshold value preferably is chosen as, and makes false-negative number percent remain on low-level (for example 5%), allows false-positive number percent to be increased to high level (for example 50%) simultaneously.In the case, although check subsequently and double required microslide,, in fact the quantity of the microslide of being checked by cytotechnologist becomes partly to have reduced.
Especially, because scatter diagram may produce from concentrated group (the enriched set) of suspicious OOI, therefore the finiteness of the quantity of the OOI that analyzes (that is, only generating scatter diagram with 100 OOI) has not only reduced processing demands, also helps to prevent false negative.
For false negative is minimized, can use other information to guarantee that the scatter diagram that generates from the OOI of high ratings is reliable.For example, can generate based on the nuclear area of all OOI and the scatter diagram of nucleus IOD, and be used to generate another scoring.The scoring that the scatter diagram that generates with the OOI from high ratings is derived is the same, and this scoring can compare with threshold value, and it is normal or suspicious to be used for determining that sample 14 should be classified as.In the case, if from any of the scoring of two scatter diagrams on threshold value separately, then sample 14 will be classified as suspicious.Therefore, be classified as normally in order to make sample 14, two scorings must be all under threshold value separately.Alternatively, the scoring of deriving from two scatter diagrams can be weighted and be merged into overall score, this overall score and single threshold value can be compared then.
Can use out of Memory to determine whether the scatter diagram that generates from the OOI of high ratings is reliable.For example, can measure the quantity (quantity of its rising may cause false scoring) of the artefact in the biological sample and with itself and threshold.If artefact's quantity is higher than this threshold value, then sample 14 can be classified as suspiciously, is normal even the scoring of deriving from scatter diagram may be indicated microslide 12.
Equally, can measure the quantity of OOI and it is compared with threshold value or threshold range.For example, the OOI of lesser amt can indicate from its set (pool) of OOI that generates scatter diagram too little and precise results can not be provided on statistics.Perhaps the OOI of relatively large amount may produce a large amount of artefacts, and it may make the distortion as a result that obtains from scatter diagram.In the case, if OOI quantity less than low threshold value or greater than high threshold, that is, the quantity of OOI is outside threshold range, then sample 14 can be classified as suspiciously, is normal even the scoring of deriving from scatter diagram may be indicated microslide 12.
Can measure a small amount of OOI or all OOI average optical ((corrected) of correction or uncorrected) and it is compared with threshold value or threshold range.For example, lower optical density may indicate sample dyeing abundant inadequately, and higher relatively optical density may be indicated the sample overstain.Any situation all may make nucleus IOD measure distortion, thereby produces false scoring.In the case, if average optical less than low threshold value or greater than high threshold, that is, average optical is outside threshold range, then sample 14 can be classified as suspiciously, is normal even the scoring of deriving from scatter diagram may be indicated microslide 12.
Should be noted that system 10 can be used for sample classification except that determining whether microslide should be by other purpose the further inspection of cytotechnologist.For example, the normal specimens that is obtained by system's 10 classification can be used to make equipment (such as needing cytotechnologist to check the screening system of all microslides) standardization.For example, can handle normal specimens so that standardized mean nuclear optical density to be provided by screening system.Then can be by the not processed sample of screening system processing subsequent.If sample subsequently has the measurement mean nuclear optical density (for example, if sample is subsequently overstained or dyes inadequately) that is different from a certain specified rate of standardized mean nuclear optical density, then will refuse this result.
The purposes of scatter diagram can also be as the screening system that is described as normal or suspicious cells nuclear matter in each sample.Especially, with reference to Fig. 6, show the biological screening system 110 of structure according to another embodiment of the present invention.Except needing cytotechnologist, system 110 checks subsequently all microslides 112 (figure 5 illustrates), screening system 110 is roughly the same with previously described system 10, makes cytotechnologist's notice only focus on those suspicious regions (being called fields of interest (FOI)) of sample 114.In other words, before checking, cytotechnologist do not have microslide 112 (even suspicious) to be excluded.It is suspicious that system 110 uses scatter diagram to indicate which microslide 112, thereby makes cytotechnologist pay special attention to suspicious slides.
System 110 generally includes: (1) imaging station 118 is used to obtain the scan image of the biomaterial that holds, and generates the electronic image data from this image on microslide 112; (2) server 120, are used for the filtering image data to obtain OOI and to be used to each FOI to distribute one or more OOI; And (3) a plurality of inspection posts 122 (showing 3), its each all provide and scanned to present the field of view (FOV) (shown in Figure 7) of OOI to cytotechnologist with respect to each FOI.System 110 can also comprise the user interface (not shown), comprises monitor, keyboard and mouse (all not shown), makes that cytotechnologist can be mutual with system 110.
Imaging station 118 is set and is used for to make microslide 112 imagings with the similar mode of aforementioned mode about screening system 10.Therefore, imaging station 118 comprises and above-mentioned camera 16, microscope 18 and motorized stage 20 similar cameras 124, microscope 126 and motorized stage 128.Yet, the x-y coordinate when motorized stage 128 is gone back the tracking map picture and caught by camera 124.For example, the scrambler (not shown) can be coupled to each motor of motorized stage 128, to follow the trail of the clear distance that on x direction and y direction, moves during the imaging.Measure these coordinates with respect to the reference point 116 that pastes microslide 112 (shown in Figure 5).Just as will be described in detail below, these reference points 116 also will be used by each inspection post 122, can be associated with the x-y coordinate of the microslide 112 that obtains during imaging process at the x-y of the microslide during the checking process 112 coordinate guaranteeing.
Server 120 comprises: (1) image processor 130 is set the view data that is used for from being obtained by camera 124 and obtains OOI; (2) the FOI processor 132, are set to be used for distributing OOI to each FOI; (3) path processor 134, are set to be used for the drawing path route, and inspection post 122 will use this path route to come to scan the next one from a FOI; And (4) storer 136, be designed to store the grading of OOI and FOI, OOI and the path route of x-y coordinate and FOI.Should be appreciated that by each processor 130,132 and 134 functions of carrying out and can carry out by single processor, perhaps alternatively, by carrying out more than three processor.Similarly, be further appreciated that storer 136 can be divided into a plurality of storeies.
Image processor 130 is similar to previously described image processor 22, except it is graded with beyond the FDR storage together with it with ABC (that is abnormal clusters) grading and with ABC.Image processor 130 is additionally at the coordinate of FDR stored OOI and ABC.Image processor 130 is in the mode identical with image processor 22, obtains nuclear area and the nucleus IOD of the OOI of 100 high ratings from FDR, and generates the scatter diagram of nuclear area (y-axle) the pair cell nuclear IOD (x-axle) of each OOI.Image processor 130 is categorized as microslide 112 normal or suspicious based on scatter diagram and any alternatively other information then.If it is suspicious that microslide 112 is classified as, then image processor 130 will be stored the mark of making this indication in storer 136.
FOI processor 132 limits FOI based on the x-y coordinate of OOI.In the illustrated embodiment, define 22 FOI, wherein 20 are the OOI reservation, and 2 is to keep for ABC (that is abnormal clusters).Under the simplest situation, 20 FOI are centered on the OOI of 20 high ratings, and remaining 2 FOI are centered on the ABC of 2 high ratings.Alternatively, to avoid in this FOI, comprising that the OOI that is included among another FOI and the mode of ABC limit FOI.In this way, OOI and ABC can divide into groups in the coordinate mode in FOI, thereby are included in OOI in the FOI and the quantity of ABC can maximize.Alternatively, if the quantity of FOI is unfixing, comprise that then the quantity of the FOI that all OOI and ABC are required can minimize.
After generating FOI, FOI processor 132 is stored in the x-y coordinate of all FOI in the storer 136, so that the use after a while of path processor 134.Especially, path processor 134 uses suitable routing algorithm (such as, " traveling salesman (travelingsalesman) " algorithm) to draw the x-y coordinate of FOI, and it has determined to be used in the inspection post the 122 the most effective observation paths that present FOI.Path processor 134 then with the x-y coordinate of FOI together with route scheme (in the illustrated embodiment, it is realized by simply FOI being placed on checked order in the tabulation according to it) be stored in the storer 136, so that the subsequent access in inspection post 122.
Still with reference to Fig. 6, in one embodiment, show totally three inspection posts 122 with server 120 couplings, make nearly three cytotechnologists to visit simultaneously to be stored in the server 120 for information about.Especially, system 110 can observe microslide than cytotechnologist usually and handle microslide 112 quickly.Check speed even the sample preparation speed of system 110 is lower than cytotechnologist's sample, system 110 can work 24 hours in usually one day, and typical cytotechnologist only can work 8 hours in one day.Therefore, the bottleneck in the examination process appears at people's class hierarchy,, is contained in the detailed inspection of the biomaterial on the microslide 112 that is.Therefore, can understand, the use in a plurality of inspection posts 122 has alleviated this bottleneck, thereby more effectively process is provided.
Before the details of describing inspection post 122, with reference to Fig. 7, it shows the exemplary FOV that each inspection post 122 is centered in the FOI top.In the illustrated embodiment, the diameter of FOV is 2.2mm, and FOI is enclosed a next 0.4mm * 0.4mm square by being restricted to by FOV.In practical embodiments, the border of FOI is that fabricate and sightless, thereby cytotechnologist all can not intercepted the observation of any OOI.In order to concentrate cytotechnologist's notice on FOI quickly and the reference of the accurate regional extent that common indication limits by the imaginary border of FOI to be provided, provide L shaped mark indicator 142.Mark indicator 142 is caught FOI (that is, the left side of the square opening part 144 of mark indicator 142 and FOI and base adjacency).The space of 0.05mm is set between the empty border of the border of mark indicator 142 and FOI, makes OOI extend to the part of the outside of the left margin of FOI and bottom boundaries (by being included in the FOI but the left margin and near the OOI the bottom boundaries that are centered at FOI produce) and can not be labeled indicator 142 and hinder.Mark indicator 142 also is used to provide a kind of method, (for example be used for when the needs virologist further checks, if OOI has pernicious attribute or the preceding attribute that cancerates), make cytotechnologist flag F OI (for example, by pressing the button of painting electronically) electronically to mark indicator 52.
Refer back to Fig. 6, each inspection post 122 comprises microscope 138 and motorized stage 140.Microslide 112 (after Flame Image Process) is placed on the motorized stage 140, make motorized stage based on route scheme that obtains from storer 136 and the x-y coordinate of FOI, with respect to the mobile microslide 112 in the observation district of microscope 138.Microslide 112 (after Flame Image Process) is placed on the motorized stage 140, make motorized stage based on route scheme that obtains from storer 136 and the x-y transformation of coordinates of FOI, with respect to the mobile microslide 112 in the observation district of microscope 138.Especially, the x-y coordinate that these obtain with respect to the x-y coordinate system at imaging station 118 appends to the reference point 116 (shown in Figure 5) of microslide 112 with use, changes into the x-y coordinate system in inspection post 122.Therefore, guaranteed to be associated with the x-y coordinate of microslide 112 during imaging process at the x-y of the microslide during the checking process 112 coordinate.Motorized stage 140 will move according to the x-y coordinate after the FOI conversion as the route scheme indication then.In the illustrated embodiment, in order to advance to another from a FOI, cytotechnologist presses the starting switch (not shown).On this meaning, inspection post 122 is automanual.Alternatively, automatically a FOI is advanced to the next one.In the case, automatic station 140 can optionally suspend the time of scheduled volume to each FOI.On this meaning, inspection post 122 is full automatic.
When the FOI that selects was presented among the FOV of microscope 138, cytotechnologist checked that the degree of FOI and pair cell unusual (if any) decisions making.Cytotechnologist incites somebody to action the suspicious FOI of mark electronically then.The FOI that observed before cytotechnologist can turn back to, and manually move to (and observation) position that is not surrounded on microslide by FOI.Next check microslide 112, if any of FOI done mark by cytotechnologist, the whole biological sample 112 of inspection post 122 autoscans then, thus guaranteed 100% range of observation.If desired, cytotechnologist can end autoscan and transfer table 140, with reorientate and access slide 112 on the position.
Especially, if it is suspicious that sample 114 has been classified as, then cytotechnologist (for example will be notified in inspection post 122, can listen or optical signal by providing), and in the case, to will impel cytotechnologist that microslide is carried out more completely and check carrying inspection that the microslide that is classified as normal sample carries out comparatively speaking with cytotechnologist.
The use of scatter diagram also can be used in determines measuring the dyeing quality of a batch sample.Have been found that, the biological sample of inferior quality dyeing (for example, surpassed its storage life and also gone bad the tenure of use of not following dyeing rules or coloring agent) often all have broad distribution (that is, spaced point trends towards forming loose bounded domain) aspect nuclear area and the nucleus IOD two.On the contrary, the biological sample of high quality stains often all has narrower and small distribution (that is, spaced point trends towards forming tight bounded domain) aspect nuclear area and the nucleus IOD two.Therefore, the scope of scatter point distribution provides the indication to the dyeing quality that is positioned at the biological sample 14 on the microslide 12.Difference between the scatter diagram that generates from the dyeing of high quality stains and inferior quality comes from the inferior quality uneven fact that dyes, therefore and will make spaced point away from the zone that clearly limits, this zone is by all evenly limiting clear and definite nucleus and produce with respect to tenuigenin.
Can make in ining all sorts of ways any one measure the scatter point distribution of each sample.In one embodiment, spaced point is carried out geometric fit and being that mode identical normally or in suspicious is come calculated value with the above-mentioned product of random sample really.(for example, 100 samples) calculated value calculates the merging value, thereby can integrally measure the dyeing quality of this batch sample from a batch sample then.Can calculate the intermediate value (median) of the some distribution value of these samples.Alternatively, can calculate the intermediate value of the some distribution value (for example, minimum 10 some distribution values) of the sample of certain percentage.Can use mode of distribution value (mode) or average, and not use intermediate value.
In case calculate intermediate value, it can be used for determining whether current sample is suitably dyeed.This can realize in various manners.For example, can compare with intermediate value and by the baseline value (for example, by getting an average of distribution value) that is known as the correct sample calculating of dyeing.Preferably, the quantity of sample that is used to calculate the baseline intermediate value is enough big, so that the regular sampling by the handled sample in position that is provided with imaging system to be provided.
Suppose that with certain mode calculation level distribution value make that value raises along with the reduction of the dyeing quality of sample, the intermediate value that then exceeds the baseline value specified rate will indicate current batch sample may have low-quality dyeing.In the case, image processor 22 error message or the mark that are used to indicate the sample of nearest analysis correctly not dyeed generation.The child group of then can the artificially checking these suspicious specimen is visually to confirm dyeing quality.
Alternatively, can use card side (Chi-Square) statistical method to assess the intermediate value of the current batch of sample of comparing with the intermediate value of front batch sample.For example, provide following equation: c 2=∑ (f o-f e) 2/ f e, wherein, c 2Equal card side, f oThe mean value that equals to obtain, and f eEqual to expect mean value, can determine whether that a nearest batch sample has inferior quality dyeing.For example, if the mean value of last five batch samples is respectively 130,150,110,100 and 170, f then eEqual
Figure C20048001633400211
And
c 2 = ( 130 - 132 ) 2 132 + ( 150 - 132 ) 2 132 + ( 110 - 132 ) 2 132 + ( 100 - 132 ) 2 132 + ( 170 - 132 ) 2 132 =
0.0303 + 2.4545 + 3.6667 + 7.7575 + 10.9393 = 24.8483
Use the canonical statistics handbook, then can will the side of card c with given ground confidence level 2Compare with cutoff, with the mean value of determining many batch samples whether within normal the distribution.If do not exist, determine that then a nearest batch sample has low-quality dyeing.
Alternatively, some distribution value can directly be compared with baseline value, rather than determines mean value and then mean value is compared with baseline value from the some distribution value of calculating.In the case, can calculate quantity, and it is compared with number of threshold values above the some distribution value of giving determined number of baseline value.For example, give fixed number, can determine that then the quality of coloring agent is suspicious, produce error message or mark thereby impel if 20% some distribution value surpasses threshold value.

Claims (16)

1. one kind is used for biological screening system that biological sample is classified, comprising:
The imaging station is used to obtain the scan image of described sample, and generates view data from described scan image; And
At least one processor, be used to filter described view data to obtain objects, discern a plurality of objects in the described biological sample, measure first feature of each objects, measure second feature of each objects, generate the scatter diagram of described first measurement features to described second measurement features, described scatter diagram has the point that has scope and distributes, and described sample is categorized as normally based on the described scope that described point distributes at least in part.
2. biological screening system according to claim 1, wherein, described at least one processor also is used for normal specimens is labeled as does not need further inspection.
3. biological screening system according to claim 1, wherein, it is normal or suspicious that described at least one processor also is used for distributing described sample classification based on described point.
4. biological screening system according to claim 3, wherein, described at least one processor also is used for normal specimens is labeled as does not need further inspection, and suspicious specimen is labeled as needs further to check.
5. biological screening system according to claim 1 wherein, if described some distribution range is lower than threshold value, is normal with described sample classification then.
6. biological screening system according to claim 1, wherein, described at least one processor also is used to measure described some distribution range, the definition threshold value, and, wherein, described sample relatively is categorized as normally based on described some distribution range and the described threshold measured.
7. biological screening system according to claim 6, wherein, described at least one processor by to described some match geometric configuration, determine described geometric configuration at least one dimension, and calculate the value of the function be described at least one dimension, measure described some distribution range, and wherein, value and the described threshold of described at least one processor by being calculated is with some distribution range and the described threshold of being measured.
8. biological screening system according to claim 7, wherein, described geometric configuration is oval.
9. biological screening system according to claim 6, wherein, described at least one processor is measured described some distribution range by the value of calculating, wherein, with value and the described threshold of being calculated.
10. biological screening system according to claim 1, wherein, described at least one processor also is used to measure at least one additional features of described biological sample, wherein, described sample further is categorized as normally based on described at least one other measurement features.
11. biological screening system according to claim 10, wherein, described at least one processor also is used for determining that based on described at least one other measurement features described scatter diagram is reliable.
12. biological screening system according to claim 10, wherein, described at least one other measurement features is the sum of the objects in the described biological sample.
13. biological screening system according to claim 10, wherein, described at least one other measurement features is the sum of the artefact in the described biological sample.
14. biological screening system according to claim 10, wherein, described at least one other measurement features is the mean nuclear optical density of described objects.
15. biological screening system according to claim 1, wherein, the quantity of described objects is scheduled to.
16. biological screening system according to claim 1, wherein, described at least one processor also is used for being classified as under the suspicious situation at described sample and produces a plurality of fields of interest to comprise described objects, described system also comprises automatic or semi-automatic inspection post, is used for respect to each fields of interest scanning field of view to present described objects.
CN200480016334A 2003-06-12 2004-05-19 System for determining the stain quality of slides using scatter plot distributions Expired - Fee Related CN100588941C (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US47843403P 2003-06-12 2003-06-12
US60/478,434 2003-06-12
US10/794,334 2004-03-05

Publications (2)

Publication Number Publication Date
CN1806166A CN1806166A (en) 2006-07-19
CN100588941C true CN100588941C (en) 2010-02-10

Family

ID=36867506

Family Applications (1)

Application Number Title Priority Date Filing Date
CN200480016334A Expired - Fee Related CN100588941C (en) 2003-06-12 2004-05-19 System for determining the stain quality of slides using scatter plot distributions

Country Status (1)

Country Link
CN (1) CN100588941C (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104732230A (en) * 2015-03-27 2015-06-24 麦克奥迪(厦门)医疗诊断系统有限公司 Pathology image local-feature extracting method based on cell nucleus statistical information
CN105259095A (en) * 2015-10-14 2016-01-20 南昌西尔戴尔医疗科技有限公司 Negative-exclusion-method intelligent screening system for cervical cancer cellpathology

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Optical microscope system for standardized cell measurements and analyses. James W.Bacus,et al.APPLIED OPTICS,Vol.26 No.16. 1987
Optical microscope system for standardized cell measurements and analyses. James W.Bacus,et al.APPLIED OPTICS,Vol.26 No.16. 1987 *

Also Published As

Publication number Publication date
CN1806166A (en) 2006-07-19

Similar Documents

Publication Publication Date Title
US4175860A (en) Dual resolution method and apparatus for use in automated classification of pap smear and other samples
AU2009251162B2 (en) Method for classifying slides using scatter plot distributions
EP3088511B1 (en) Image processing device, program, information storage medium, and image processing method
JP4948647B2 (en) Urine particle image region segmentation method and apparatus
US9607372B2 (en) Automated bone marrow cellularity determination
CN101809589B (en) Methods and systems for processing biological specimens utilizing multiple wavelengths
CN1795375B (en) System for organizing multiple objects of interest in field of interest
JP4864709B2 (en) A system for determining the staining quality of slides using a scatter plot distribution
Lerner et al. Automatic signal classification in fluorescence in situ hybridization images
Rolke et al. Size structure analysis of zooplankton samples by means of an automated image analyzing system
CN112432902A (en) Automatic detection system and method for judging cell number through peripheral blood cell morphology
CN113552126A (en) Reticulocyte detection method and system
CN100588941C (en) System for determining the stain quality of slides using scatter plot distributions
Tucker An image analysis system for cervical cytology automation using nuclear DNA content.
EP0545348A1 (en) Blood cell analyzer
CN114223016A (en) Scanning/pre-scanning quality control of slides
JPH1090163A (en) Particle analyzer
Netten et al. Automation of fluorescent dot counting in cell nuclei
JP2001510894A (en) Inspection system with sample inspection function
CN113945558B (en) Method for judging myeloproliferation degree
Lerner et al. Automatic signal classification in fluorescence in-situ hybridization images

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
C17 Cessation of patent right
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20100210

Termination date: 20130519