CN113295692A - Cell analysis method based on cell nucleus DNA and TBS double analysis method, computer equipment and storage medium - Google Patents
Cell analysis method based on cell nucleus DNA and TBS double analysis method, computer equipment and storage medium Download PDFInfo
- Publication number
- CN113295692A CN113295692A CN202110570957.XA CN202110570957A CN113295692A CN 113295692 A CN113295692 A CN 113295692A CN 202110570957 A CN202110570957 A CN 202110570957A CN 113295692 A CN113295692 A CN 113295692A
- Authority
- CN
- China
- Prior art keywords
- cells
- value
- abnormal
- positive
- image
- 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.)
- Granted
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 56
- 210000004027 cell Anatomy 0.000 title claims description 260
- 210000003855 cell nucleus Anatomy 0.000 title claims description 27
- 230000002159 abnormal effect Effects 0.000 claims abstract description 98
- 206010020718 hyperplasia Diseases 0.000 claims abstract description 46
- 230000002390 hyperplastic effect Effects 0.000 claims abstract description 46
- 238000000034 method Methods 0.000 claims abstract description 26
- 230000002062 proliferating effect Effects 0.000 claims abstract description 22
- 239000012472 biological sample Substances 0.000 claims abstract description 18
- 108091093105 Nuclear DNA Proteins 0.000 claims abstract description 6
- 108020004414 DNA Proteins 0.000 claims description 42
- 238000013528 artificial neural network Methods 0.000 claims description 31
- 206010058314 Dysplasia Diseases 0.000 claims description 12
- 238000011176 pooling Methods 0.000 claims description 12
- 239000002131 composite material Substances 0.000 claims description 9
- 150000001875 compounds Chemical class 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 9
- 230000001086 cytosolic effect Effects 0.000 claims description 7
- 230000000007 visual effect Effects 0.000 claims description 7
- 238000013329 compounding Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 210000000805 cytoplasm Anatomy 0.000 claims description 4
- 238000000338 in vitro Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 2
- 230000035755 proliferation Effects 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 230000000877 morphologic effect Effects 0.000 claims 2
- 210000004940 nucleus Anatomy 0.000 claims 2
- 230000004807 localization Effects 0.000 claims 1
- 239000000203 mixture Substances 0.000 claims 1
- 230000035945 sensitivity Effects 0.000 abstract description 13
- 230000005856 abnormality Effects 0.000 abstract 1
- 239000010410 layer Substances 0.000 description 40
- 238000001514 detection method Methods 0.000 description 15
- 230000006870 function Effects 0.000 description 14
- 238000004891 communication Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 239000000523 sample Substances 0.000 description 6
- 241000193830 Bacillus <bacterium> Species 0.000 description 5
- 238000003745 diagnosis Methods 0.000 description 5
- 238000007477 logistic regression Methods 0.000 description 4
- 210000001519 tissue Anatomy 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 244000005700 microbiome Species 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 238000002955 isolation Methods 0.000 description 2
- 238000003064 k means clustering Methods 0.000 description 2
- 239000002346 layers by function Substances 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000010186 staining Methods 0.000 description 2
- 238000007447 staining method Methods 0.000 description 2
- 230000001502 supplementing effect Effects 0.000 description 2
- FFRBMBIXVSCUFS-UHFFFAOYSA-N 2,4-dinitro-1-naphthol Chemical compound C1=CC=C2C(O)=C([N+]([O-])=O)C=C([N+]([O-])=O)C2=C1 FFRBMBIXVSCUFS-UHFFFAOYSA-N 0.000 description 1
- 241000894006 Bacteria Species 0.000 description 1
- 241000606161 Chlamydia Species 0.000 description 1
- 208000035473 Communicable disease Diseases 0.000 description 1
- 241000233866 Fungi Species 0.000 description 1
- 241000238631 Hexapoda Species 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 241000204031 Mycoplasma Species 0.000 description 1
- 238000005481 NMR spectroscopy Methods 0.000 description 1
- 102000029797 Prion Human genes 0.000 description 1
- 108091000054 Prion Proteins 0.000 description 1
- 241000606701 Rickettsia Species 0.000 description 1
- 241000589970 Spirochaetales Species 0.000 description 1
- 241000700605 Viruses Species 0.000 description 1
- KSCQDDRPFHTIRL-UHFFFAOYSA-N auramine O Chemical compound [H+].[Cl-].C1=CC(N(C)C)=CC=C1C(=N)C1=CC=C(N(C)C)C=C1 KSCQDDRPFHTIRL-UHFFFAOYSA-N 0.000 description 1
- 238000001574 biopsy Methods 0.000 description 1
- 238000004195 computer-aided diagnosis Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 235000013601 eggs Nutrition 0.000 description 1
- 210000003527 eukaryotic cell Anatomy 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 230000007762 localization of cell Effects 0.000 description 1
- 244000000010 microbial pathogen Species 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000009206 nuclear medicine Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 244000045947 parasite Species 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 210000001236 prokaryotic cell Anatomy 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000001356 surgical procedure Methods 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 210000004916 vomit Anatomy 0.000 description 1
- 230000008673 vomiting Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
Landscapes
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention discloses a cell analysis method based on a nuclear DNA and TBS double analysis method, computer equipment and a storage medium, wherein the cell analysis method comprises the following steps: s1: obtaining an image of an ex vivo biological sample; s2: first appearance morphology, and cells were classified into four categories: normal cells, suspected abnormal cells, highly suspected abnormal cells, and positive abnormally proliferating cells; s3: and (4) performing second appearance shape judgment on the positive abnormal hyperplastic cells or the highly suspicious abnormal hyperplastic cells obtained in the step (S2), and dividing the cells into three types: suspicious hyperplastic cells, highly suspicious dysplastic cells, and positive dysplastic cells; step S4: the analysis results of step S2 and step S3 are used to obtain a comprehensive cell analysis result, and the abnormality indexes of the cells are displayed in order from high to low when the result is output. The method of the invention can give consideration to both sensitivity and specificity during cell analysis, and effectively improve the analysis efficiency of workers.
Description
Technical Field
The invention relates to a cell analysis method, in particular to a cell analysis method based on a nuclear DNA and TBS double analysis method, a computer device and a storage medium.
Background
Many lesions of an organism cause abnormal changes in normal cells, and therefore, it is often clinically necessary to determine whether a tissue or an organism is diseased or not by means of microscopic examination of a biological sample.
Along with the development of the AI technology, the AI technology is increasingly relied on for cell analysis, however, the existing AI technology for biological sample analysis has the following problems: the sample analysis error is large, and the conditions of false detection and missing detection often occur; the sample data volume is big, and sample analysis work load is big, and is not convenient for medical staff to carry out looking up of biological sample image, and is inefficient to the analysis of sample.
Disclosure of Invention
In order to solve the technical problems, the invention provides a cell analysis method, a computer device and a storage medium for a cell nucleus DNA and TBS double analysis method.
The cell analysis method based on the nuclear DNA and TBS double analysis method comprises a first appearance shape judgment step and a second appearance shape judgment step, and the cells are divided into four types through the first appearance shape judgment step: normal cells, suspected abnormal cells, highly suspected abnormal cells, and positive abnormally proliferating cells; the second appearance shape judging step is to realize cell analysis by calculating and analyzing the nuclear plasma ratio of the positive abnormal hyperplastic cells or the highly suspicious abnormal hyperplastic cells obtained in the first appearance shape judging step, defining the nuclear plasma ratio as a DNA parameter 2, respectively presetting a threshold D and a threshold E for the DNA parameter 2, and dividing the DNA parameter 2 into three numerical value ranges, thereby dividing the cells to be analyzed into three types: firstly, the DNA parameter 2 is less than or equal to a threshold value D, and the cells are suspicious hyperplastic cells; the threshold D is more than the DNA parameter 2 and less than or equal to the threshold E, and the cells are highly suspicious abnormal hyperplastic cells; ③ if the DNA parameter 2 is more than the threshold E, the cells are positive abnormal hyperplastic cells; judging the cells which are judged to be the positive abnormally proliferating cells by the first appearance shape judging step and the second appearance shape judging step as the first positive abnormally proliferating cells; judging the abnormal hyperplastic cells which are judged to be positive by the first appearance shape judging step as well as the highly suspicious abnormal hyperplastic cells which are judged to be second positive abnormal cells by the second appearance shape judging step; judging the cells which are judged to be highly suspicious abnormal cells by the first appearance shape judging step and judging the cells which are judged to be positive abnormally proliferated cells by the second appearance shape judging step as third positive abnormal cells; judging the cells which are judged to be highly suspicious abnormal cells by the first appearance shape judging step and judging the cells which are judged to be highly suspicious abnormally proliferating cells by the second appearance shape judging step as fourth positive abnormal cells; judging the cells which are judged to be the positive abnormal hyperplastic cells or the highly suspicious abnormal cells by the first appearance shape judging step, and judging the cells which are judged to be the suspicious hyperplastic cells by the second appearance shape judging step as fifth positive abnormal cells; and judging the cells which are judged to be suspicious abnormally hyperplastic cells by the first appearance shape judging step and judging the cells which are judged to be suspicious abnormally hyperplastic cells or positive abnormally hyperplastic cells by the second appearance shape judging step as fifth suspicious cells.
Further, the first appearance shape judging step is realized by acquiring a DNA parameter 1 of the size of the cell nucleus area, defining the DNA parameter 1 as the first appearance shape judging step, presetting a threshold value a, a threshold value B and a threshold value C for the DNA parameter 1, and dividing the parameter 1 into four numerical value ranges, thereby classifying the cells to be analyzed into four types: 1, abnormal cells, wherein the parameter 1 is less than or equal to a threshold value A; suspicious abnormal cells, wherein the parameter is more than the threshold A and less than or equal to the threshold B; ③ highly suspicious abnormal cells, wherein the threshold B is more than the parameter 1 and less than or equal to the threshold C; fourthly, the positive abnormal proliferation cells, the parameter 1 is larger than a threshold value C, and the calculation and the analysis of the parameter 1 are realized through a pre-trained neural network;
the neural network comprises an input layer, a hidden layer and an output layer, the output layer of the neural network outputs scores corresponding to the in-vitro biological samples, and the highest score is used as an analysis index of the first appearance form judgment step;
the neural network comprises a first neural network module, a second neural network module and a classifier, wherein the first neural network module is a Darknet-53 network, the second neural network module comprises 5 convolution layers, a ReLU layer, a pooling layer, a BN + LeakyReLU layer and a convolution layer which alternately appear, and the classifier adopts an LR logical regression classifier.
Further, before the first appearance shape judging step, the method further comprises a step of acquiring and preprocessing an image of the cell to be analyzed, and specifically comprises the following steps:
(1) amplifying the cells under a microscope by different times, automatically focusing the cells through the microscope and acquiring a target picture to be analyzed; dividing each picture of a group of pictures shot by cells in the same visual field of a microscope into three channels of R (red), G (green) and B (blue), compounding any two channels to obtain a first compound image, calculating Variance of each picture relative to the first compound image to obtain Variance value, comparing the Variance values of each picture, and selecting the picture with the maximum Variance value relative to the first compound image as a first target picture of the group of pictures;
(2) and fusing a plurality of first target pictures of different levels in the same field to obtain an image of the cell to be analyzed.
Further, the step (2) is realized according to the following steps: defining one first target picture as A (X, y), defining the previous picture as A _ Pre (X, y), defining the Next picture as A _ Next (X, y), defining the fused image as Fuse _ image (X, y), respectively setting the pixel values of the three images at the point (X, y) as A (X, y) value, respectively setting the A _ Pre (X, y) value and the A _ Next (X, y) value, comparing the 3X3 areas, respectively calculating the sum of the 3X3 area pixel values of the three first target pictures to obtain A _ add _ value, respectively calculating the sum of the A _ Pre _ add _ value and the A _ Next _ add _ value, and respectively obtaining the fused image at the point (X, y) value, which is the integral value of the fused image after traversing the point (X, y) corresponding to the maximum value.
Furthermore, the cell nucleus of the image of the cell to be analyzed is positioned and identified, and the area size of the cell nucleus is calculated.
Further, the second appearance shape judging step includes a step of calculating the size of the cytoplasmic area of the positive dysplasia cells or highly suspicious dysplasia cells located by the first appearance shape judging step.
Furthermore, the size of the cytoplasmic area is calculated by separating the positioned and segmented positive dysplasia cells or highly suspicious dysplasia cells to obtain a red channel image R (x, y), a green channel image G (x, y) and a blue channel image B (x, y), obtaining a second composite image by using the images of any two channels, and calculating the Mean value Mean _ value of the second composite image; subtracting each pixel of the second composite image from Mean _ value 0.8, the new pixel value being equal to each pixel value of the second composite image-Mean _ value 0.8, and assigning it 255 when the new pixel value is greater than 0; and when the new pixel value is less than 0, assigning the new pixel value as 0, obtaining a binary image Threshold (x, y) according to the value, obtaining a convex hull of the Threshold (x, y) image, obtaining the area of the convex hull, and finally finding out the largest area, namely the cytoplasm area of the cell.
Furthermore, the second appearance shape judging step further comprises the step of dividing the DNA parameter 1 of the positive abnormal hyperplastic cell or the highly suspicious abnormal hyperplastic cell obtained in the first appearance shape judging step by the corresponding cytoplasmic area of the cell to obtain the nuclear plasma ratio, and calculating to obtain the DNA parameter 2.
The invention also provides a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of the aforementioned method.
The invention also provides a storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the aforementioned method.
The invention has the following beneficial effects:
firstly, analyzing the area of a cell nucleus by a first appearance shape judging step, so that abnormal cells can be quickly analyzed, then further analyzing highly suspicious abnormal cells and positive abnormally proliferating cells in the abnormal cells by a second appearance shape judging step, further dividing the cells to be analyzed into suspicious proliferating cells, highly suspicious abnormally proliferating cells and positive abnormally proliferating cells by a DNA parameter 2, separating the abnormal cells by the first appearance shape judging step, and further analyzing the abnormal cells by adopting a cell nucleus plasma ratio TBS relationship, so that the abnormal cells are further diagnosed and confirmed, the specificity of abnormal cell diagnosis is further improved, and the sensitivity and the specificity of exfoliative cytology are provided by combining and mutually supplementing the two methods.
Secondly, the method of the invention arranges the corresponding visual fields of the sequences from front to back by sequencing the cell nucleus area from big to small, the cell nucleus serous ratio TBS positive to the suspicious cells, so that analysts can conveniently and quickly find the visual field of the abnormal cells and quickly look up and locate the abnormal cells, the analysis efficiency of the cells is improved, and the burden of the staff is reduced.
The method can also adjust the threshold (such as threshold A, threshold B or threshold C) set by the DNA parameter 1 and the cell nuclear plasma ratio TBS analysis threshold (such as threshold D or threshold E) of the area size of different types of exfoliated cytology cell nuclei in a targeted manner according to the characteristics of different types of exfoliated cytology cell nuclei, thereby carrying out targeted analysis on the cell nuclei of different types of exfoliated cells and providing detection sensitivity and detection specificity in different ranges.
The method can search a balance point which can be researched on the specificity and the sensitivity of the detection, not only improves the sensitivity of the analysis, but also improves the specificity of the analysis and the diagnosis, solves the problems of the sensitivity and the specificity of different types of exfoliative cytology, can ensure the negative elimination rate with high specificity and high sensitivity, and can provide a positive result with high accuracy, thereby effectively reducing the workload of cell analysis.
Detailed Description
The following describes the technical solution of the present invention with reference to specific embodiments.
In the present invention, "ex vivo" refers to the isolation of a target organism from an animal or human into an in vitro environment, whether by surgery or sampling or the isolation of vomit or the like; "biological sample" broadly refers to a pathogenic microorganism or tissue cell, including but not limited to: non-cell type microorganisms such as: mainly comprises viruses, prion particles and the like; prokaryotic cell type microorganisms such as: bacteria, rickettsia, chlamydia, mycoplasma, spirochete, etc.; eukaryotic cell type microorganisms such as: including fungi, protozoa, parasites (protozoa, worms, medical insects) and their eggs, etc. Tissue cells include, but are not limited to: tissue sections and various cells.
TBS in the context of the present invention is an abbreviation for descriptive diagnosis (the Thethesdasystem). The nuclear DNA and TBS double analysis method of the invention refers to the analysis of cells by combining the size of cell nucleus with the TBS analysis of cells.
In the present invention, images may be acquired by various medical image acquisition apparatuses. For example, for the detection of various infectious diseases, microscopic images of biological samples can be obtained by means of microscopic imaging. The images may also be acquired by means of ultrasound equipment, X-ray equipment, nuclear magnetic resonance equipment, nuclear medicine equipment, medical optical equipment, thermal imaging equipment, and the like.
In the embodiment of the present invention, the image may be a two-dimensional image or a three-dimensional image. The image may be a grayscale image, a binary image, or a color image.
The image of the present invention may be an image of an ex vivo biological sample observed by a biological microscope.
The first embodiment is as follows:
the cell analysis method based on the double analysis method of the nuclear DNA and TBS comprises the following steps:
s1: obtaining an image of an ex vivo biological sample;
s2: first appearance morphology, and cells were classified into four categories: normal cells, suspected abnormal cells, highly suspected abnormal cells, and positive abnormally proliferating cells;
s3: and (4) performing second appearance shape judgment on the positive abnormal hyperplastic cells or the highly suspicious abnormal hyperplastic cells obtained in the step (S2), and dividing the cells into three types: suspicious hyperplastic cells, highly suspicious dysplastic cells, and positive dysplastic cells;
step S4: using the analysis results of step S2 and step S3, a combined cell analysis result is obtained: judging the cells judged to be the positive abnormally proliferating cells by both the step S2 and the step S3 as first positive abnormal cells; judging the abnormal hyperplastic cells judged to be positive by the step S2 and the highly suspicious abnormal hyperplastic cells judged to be highly suspicious by the step S3 as second positive abnormal cells; judging the abnormal cell which is judged to be highly suspicious by the step S2 as a third positive abnormal cell which is judged to be a positive abnormally proliferating cell by the step S3; judging the cells which are judged to be highly suspicious abnormal cells by the step S2 and the cells which are judged to be highly suspicious abnormally proliferating cells by the step S3 as fourth positive abnormal cells; judging the cells which are judged to be the positive abnormal hyperplastic cells or the highly suspicious abnormal cells by the step S2 as the fifth positive abnormal cells by the step S3; the suspected abnormally proliferating cells judged in step S2 and the suspected abnormally proliferating cells or the positive abnormally proliferating cells judged in step S3 are judged as the fifth suspected cells. The display sequence of each type of cells is sequentially a first positive abnormal cell, a second positive abnormal cell, a third positive abnormal cell, a fourth positive abnormal cell, a fifth positive abnormal cell and a fifth suspicious cell from front to back, so that a cell image with the highest comprehensive abnormal risk coefficient is firstly displayed in front of a worker, and the worker can conveniently analyze the cells.
The image in step S1 may be an image of an original unstained biological sample, or an image of a stained biological sample.
In the present invention, the staining method for the biological sample may adopt any staining method in the prior art, such as wil-niemann acid-fast staining (taking the tubercle bacillus cell as an example, the tubercle bacillus is stained red under a microscope, and the non-tubercle bacillus is stained blue), auramine O fluorescence acid-fast staining (the tubercle bacillus is stained golden yellow under a microscope, and other non-tubercle bacillus is colorless or black background) so as to have better identifiability in the image.
The resolution of the image of the biological sample can be processed appropriately according to the need, for example, the resolution can be 416 x 416 or 256 x 256, etc.
The image of the ex-vivo biological sample of step S1 of the present embodiment is obtained by the following steps: dividing each picture of a group of pictures of an ex-vivo biological sample taken in the same visual field of a microscope into three channels of R (red), G (green) and B (blue), obtaining a new image B _ G by utilizing B-G, respectively carrying out Variance calculation on each picture relative to the image B _ G to obtain a Variance value, respectively comparing the Variance values of each picture, and selecting the picture with the maximum Variance value relative to the image B _ G as a first target picture of the group of pictures. The cell image to be analyzed is obtained by fusing a plurality of first target pictures of different levels in the same field of view, and is obtained according to the following steps: defining one first target picture as A (X, y), defining the previous picture as A _ Pre (X, y), defining the Next picture as A _ Next (X, y), defining the fused image as Fuse _ image (X, y), respectively setting the pixel values of the three images at the point (X, y) as A (X, y) value, respectively setting the A _ Pre (X, y) value and the A _ Next (X, y) value, comparing the 3X3 areas, respectively calculating the sum of the 3X3 area pixel values of the three first target pictures to obtain A _ add _ value, respectively calculating the A _ Pre _ add _ value and the A _ Next _ add _ value, and obtaining the fused sample of the in-vitro image (X, y) at the point (X, y) with the maximum value, thereby obtaining the fused image. The cell image to be analyzed according to the present invention can be obtained by amplifying the cell image by a plurality of different magnifications, including but not limited to: 4 times, 10 times, 20 times, 40 times, and 100 times.
Step S2 of the present embodiment includes: the image obtained in step S1 is input into a pre-trained neural network, which includes an input layer, a hidden layer, and an output layer.
The neural network of the present invention may be a connection in one or more functional layers of convolution, pooling, full connection, residual, excitation, regularization, tensor stitching, etc. in the prior art, and the function to be performed by each functional layer is known to those skilled in the art. The convolutional layer may be used to perform a convolution operation to extract feature information of an input image (e.g., with a size of 227 × 227) to obtain a feature map featuremap (e.g., with a size of 13 × 13); the pooling layer may perform a pooling operation on the input image, such as a max-pooling (max-pooling) method, a mean-pooling (mean-pooling) method, etc.; the activation layer introduces nonlinear factors through activation functions, such as adopting correction unit (ReLU, Leaky-ReLU, P-ReLU, R-ReLU) functions, S-type functions (Sigmoid functions) or hyperbolic tangent functions (tanh functions) and the like. And the full connection layer is used for converting the feature map output by convolution into a one-dimensional vector. The loss function is used for evaluating the degree of inconsistency between the predicted value f (x) and the true value Y during the neural network training, and can be a log-log loss function, a square loss function, an exponential loss function, a Hinge loss function and the like.
Depending on the specific function to be performed, various types of neural networks may be used, for example, a deep convolutional neural network CNN such as Lenet, Alexnet, VGG, etc. may be used to extract features in the image and output the result of classifying the physical or pathological properties of the image by a classifier such as softmax, svm, etc.
The pre-trained neural network of the embodiment comprises a first neural network module, a second neural network module and a classifier, wherein the first neural network module is a Darknet-53 network, the second neural network module comprises 5 convolutional layers, a ReLU layer, a pooling layer, a BN + LeakyReLU layer and a convolutional layer which alternately appear, and the classifier adopts an LR logistic regression classifier.
The Darknet-53 network obtains three characteristic graphs with different sizes corresponding to each image from biological sample images by adopting a multi-scale fusion method, performs dimension clustering on the labeled data set by adopting a K-means clustering algorithm to obtain a plurality of prior frames (anchors) with different sizes, performs bbox prediction by adopting the K-means clustering, and performs target scoring on the part surrounded by the anchors by adopting a plurality of Logistic regression (finding out that the part with the highest score is the final prediction category).
The first neural network module comprises a convolutional layer and five residual modules which are sequentially connected, the first residual module comprises a convolutional layer and a residual unit, the second residual module comprises a convolutional layer and two residual units, the third residual module comprises a convolutional layer and eight residual units, the fourth residual module comprises a convolutional layer and eight residual units, the fifth residual module comprises a convolutional layer and four residual units, each residual unit comprises two convolutional layers and a residual connecting layer, and the five residual modules are connected in a ResNet layer-skipping connection mode. The output ends of the fourth residual error module and the fifth residual error module are input to the same first tensor splicing layer concat, and the output end of the third residual error module is input to the other second tensor splicing layer concat; the fifth residual error module outputs convolution processing (Conv2D) to obtain a first feature map, and the output end of the fourth residual error module outputs convolution processing (Conv2D) after up-sampling processing and combination output convolution processing at the output end of the second tensor splicing layer and the fifth residual error module to obtain a second feature map; the output of the output end of the third residual error module at the tensor splicing layer and the first tensor splicing layer is subjected to up-sampling processing and combination and then output convolution processing
(Conv2D) obtaining a third feature map.
The second neural network module includes 5 convolutional layers Conv, ReLU layers, Pooling layers Pooling, DBL layers, and convolutional layers Conv alternately.
The classifier adopts an LR logistic regression classifier, and utilizes logistic regression to perform an objective score (objectnessscore) on the content enclosed by each anchor of the first neural network module and the feature map output by the second neural network module, wherein the highest score is the final prediction category.
The invention locates the position of the cell nucleus by yolov3 algorithm and segments the cell nucleus, separates the channel of the segmented cell nucleus image to obtain the red channel image R (x, y), the green channel image G (x, y) and the blue channel image B (x, y), compounds any two channel images to obtain the first compound image, compounds the blue channel image B (x, y) and the green channel image G (x, y) to obtain the new image-the first B _ G (x, y), calculates the Mean value of the first B _ G (x, y) image to obtain the first Mean _ value; and (3) subtracting each pixel of the first B _ G (x, y) from the first Mean value, wherein the new pixel value is equal to each pixel value of the first B _ G (x, y) -the first Mean value, when the new pixel value is larger than 0, the new pixel value is assigned 255, when the new pixel value is smaller than 0, the new pixel value is assigned 0, so as to obtain a first B _ G _ Threshold (x, y) of the binary image, the convex hull of the first B _ G _ Threshold (x, y) image is obtained, the area of the convex hull is obtained, and finally the largest area is found to be the area of the cell nucleus, so that the DNA parameter 1 is obtained.
DNA parameter 1 was analyzed by step S2, and the cells were classified into four categories: the DNA parameter 1 is less than or equal to the threshold value A and is a normal cell; suspicious abnormal cells with the threshold A being more than the DNA parameter 1 and less than or equal to the threshold B; ③ the cells with the threshold B being more than the DNA parameter 1 and less than or equal to the threshold C are highly suspicious abnormal cells; and fourthly, the cells with DNA parameters of 1 > the threshold value C are the positive abnormal hyperplastic cells.
As a preferred implementation manner of this embodiment, step S2 further includes a step of forming the pre-trained neural network by training. Dividing sample images for training into positive and suspicious images; the output layer of the neural network is constructed to classify the scores in a preset scoring range, samples with the score of more than or equal to 95 percent as the optimal scheme are considered to be positive, samples with the score of more than or equal to 75 percent and less than 95 percent as the optimal scheme are considered to be highly suspicious, and samples with the score of more than or equal to 50 percent and less than 75 percent as the optimal scheme are considered to be suspicious; samples scored < 50% were considered negative.
The system can also score more than or equal to 80% as a suboptimal scheme and is judged to be positive, the optimal scheme with the score of more than or equal to 60% and less than 80% is judged to be highly suspicious, and the optimal scheme with the score of more than or equal to 35% and less than 55% is judged to be suspicious.
Therefore, through the training mode, the detection method can be used for four-classification of the images of the biological samples, compared with the traditional AI detection method for cell analysis, the method can only output positive results, and cannot further analyze suspicious cells, so that the cell analysis method is higher in accuracy, convenient for doctors to analyze the cells, and higher in working efficiency.
In step S3 of the present embodiment, the positive dysplasia cells or highly suspected dysplasia cells located and identified in the first appearance morphology determining step are further analyzed.
In step S3, the calculation of the DNA parameter 2 is to perform cell localization and independent segmentation on the positive abnormally proliferating cells or the highly suspicious abnormally proliferating cells obtained in step S2, and to implement cell analysis by calculating the nuclear plasma ratio thereof, defining the nuclear plasma ratio as the DNA parameter 2, presetting a threshold D and a threshold E for the DNA parameter 2, and dividing the DNA parameter 2 into three numerical ranges, thereby dividing the cells to be analyzed into three types: firstly, the DNA parameter 2 is less than or equal to a threshold value D, and the cells are suspicious hyperplastic cells; the threshold D is more than the DNA parameter 2 and less than or equal to the threshold E, and the cells are highly suspicious abnormal hyperplastic cells; and thirdly, if the DNA parameter 2 is larger than the threshold value E, the cells are positive abnormal hyperplastic cells. As a preferable embodiment of this embodiment, the threshold D may be set to 1, and the threshold E may be set to 1.5, or the threshold D and the threshold E may be adjusted as needed.
The size of the cytoplasm area is calculated by separating the positioned and segmented positive abnormal hyperplastic cells or highly suspicious abnormal hyperplastic cells to obtain a red channel image R (x, y), a green channel image G (x, y) and a blue channel image B (x, y), and compounding any two channels to obtain a third composite image, wherein the embodiment takes compounding of the B (x, y) channel image and the G (x, y) channel image as an example for explanation, and after compounding to obtain a second B _ G (x, y) image, a Mean value of the second B _ G (x, y) image is obtained; each pixel of the second B _ G (x, y) is subtracted from the second Mean _ value 0.8, the new pixel value is equal to each pixel of the second B _ G (x, y) -the second Mean _ value, and when the new pixel value is greater than 0, the value is assigned to 255; when the new pixel value is less than 0, the value is assigned to 0. And obtaining a second B _ G _ Threshold (x, y) of the binary image, obtaining a convex hull of the second B _ G _ Threshold (x, y) image, obtaining the area of the convex hull, and finally finding out the maximum area, namely the cytoplasm area of the cell.
In step S3, the size of the cell nucleus area of the positive dysplastic cells or highly suspicious dysplastic cells calculated in step S2 is divided by the cytoplasmic area of the corresponding cell to obtain the nuclear plasma ratio, and DNA parameter 2 is calculated.
The invention has the following beneficial effects:
firstly, dividing cells to be analyzed into normal cells, suspicious abnormal cells, highly suspicious abnormal cells and positive abnormal hyperplastic cells by a first appearance shape judgment step and analyzing a DNA parameter 1 (cell nucleus area size); and utilizing a second appearance shape judging step to further analyze the highly suspicious abnormal cells and the positive abnormal hyperplastic cells, further dividing the cells to be analyzed into suspicious hyperplastic cells, highly suspicious abnormal hyperplastic cells and positive abnormal hyperplastic cells through DNA parameters 2, separating the abnormal cells through the first appearance shape judging step, and further analyzing the abnormal cells by adopting a cell nucleus plasma ratio (TBS) relation, thereby further diagnosing and confirming the abnormal cells, further improving the specificity of abnormal cell diagnosis, combining the two methodology analyses and supplementing each other, and further ensuring the sensitivity and specificity of cell analysis.
Secondly, the method of the invention arranges the corresponding visual fields from front to back by sequencing the cells from large to small in area, positive in cell nucleus plasma ratio TBS analysis to suspicious cells, so that analysts can conveniently and quickly find the visual field of abnormal cells and quickly look up the abnormal cells, thereby improving the analysis efficiency of the abnormal cells.
The method can be used for adjusting the threshold (such as threshold A, threshold B or threshold C) set by the DNA parameter 1 and the threshold (such as threshold D or threshold E) of the cell nucleus ratio TBS analysis threshold of different types of exfoliated cell nuclei in a targeted manner according to the characteristics of different exfoliated cytology cell nuclei, so that the cell nuclei of different types of exfoliated cells are analyzed in a targeted manner, the detection sensitivity and the detection specificity in different ranges are provided, and an effective research method is provided for large-scale artificial intelligent analysis of different types of isolated cells.
The method can search a balance point which can be researched on the specificity and the sensitivity of the detection, not only improves the sensitivity of the analysis, but also improves the specificity of the analysis and the diagnosis, solves the problems of the sensitivity and the specificity of different types of exfoliated cells, can ensure the negative elimination rate with high specificity and high sensitivity, and can provide a positive result with high accuracy, thereby effectively reducing the workload of cell analysis.
Example two:
the present embodiment provides a computer device, including: a processor; a memory for storing a computer program; when executed by a processor, the computer program causes the processor to implement the steps of the detection method according to one embodiment.
The processor may be a Central Processing Unit (CPU), a field programmable logic array (FPGA), a single chip Microcomputer (MCU), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), or other logic operation devices with data processing capability and/or program execution capability. One or more processors may be configured to execute the above detection method simultaneously with parallel computing processor groups, or configured to execute some steps in the above detection method with some processors, some processors execute other steps in the above detection method, and so on. The computer instructions comprise one or more processor operations defined by an instruction set architecture corresponding to the processor, which may be logically embodied and represented by one or more computer programs.
The computer program of the present embodiment may be stored on a local storage or downloaded and installed from a network via a communication component.
The memory of the present embodiment may be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, a mobile storage device, or any suitable combination of the foregoing.
In order to realize the operation of the computer device, it is easy to understand that the computer device also generally has an input/output interface, a communication interface, and the like, and the input/output interface can be configured in the computer device as a component, and can also be externally connected to the device to provide corresponding functions. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc. The communication interface is used for realizing information communication between the electronic equipment and other devices so as to realize communication interaction between the electronic equipment and other equipment. The communication module can realize communication in a wired mode (such as USB, network cable and the like) and also can realize communication in a wireless mode (such as mobile network, WIFI, Bluetooth and the like).
The computer device may include only the components necessary to implement the embodiments of the present description, and need not include all of the components shown.
The computer device provided by the embodiment of the invention can be used as a computer aided diagnosis device (CAD) in medical application, can be used as a computer system for reading images in an aided manner, and can be used for assisting medical professionals and the like to diagnose diseases of patients by combining data obtained by clinical examination, biopsy and the like and personal medical experience based on parameters or state description output by the CAD.
Example three:
the present embodiment is a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, performs the steps of the method according to the first embodiment.
The above embodiments are merely illustrative of the technical concept and features of the present invention, and the present invention is not limited thereto, and any equivalent changes or modifications made according to the spirit of the present invention should be included in the scope of the present invention.
Claims (10)
1. A cell analysis method based on a nuclear DNA and TBS double analysis method is characterized by comprising the following steps: a first appearance form judging step and a second appearance form judging step,
the cells were classified into four categories by the first appearance morphology determination step: normal cells, suspected abnormal cells, highly suspected abnormal cells, and positive abnormally proliferating cells;
the second appearance shape judging step is to realize cell analysis by calculating and analyzing the nuclear plasma ratio of the positive abnormal hyperplastic cells or the highly suspicious abnormal hyperplastic cells obtained in the first appearance shape judging step, defining the nuclear plasma ratio as a DNA parameter 2, respectively presetting a threshold D and a threshold E for the DNA parameter 2, and dividing the DNA parameter 2 into three numerical value ranges, thereby dividing the cells to be analyzed into three types: firstly, the DNA parameter 2 is less than or equal to a threshold value D, and the cells are suspicious hyperplastic cells; the threshold D is more than the DNA parameter 2 and less than or equal to the threshold E, and the cells are highly suspicious abnormal hyperplastic cells; ③ if the DNA parameter 2 is more than the threshold E, the cells are positive abnormal hyperplastic cells;
judging the cells which are judged to be the positive abnormally proliferating cells by the first appearance shape judging step and the second appearance shape judging step as the first positive abnormally proliferating cells; judging the abnormal hyperplastic cells which are judged to be positive by the first appearance shape judging step as well as the highly suspicious abnormal hyperplastic cells which are judged to be second positive abnormal cells by the second appearance shape judging step; judging the cells which are judged to be highly suspicious abnormal cells by the first appearance shape judging step and judging the cells which are judged to be positive abnormally proliferated cells by the second appearance shape judging step as third positive abnormal cells; judging the cells which are judged to be highly suspicious abnormal cells by the first appearance shape judging step and judging the cells which are judged to be highly suspicious abnormally proliferating cells by the second appearance shape judging step as fourth positive abnormal cells; judging the cells which are judged to be the positive abnormal hyperplastic cells or the highly suspicious abnormal cells by the first appearance shape judging step, and judging the cells which are judged to be the suspicious hyperplastic cells by the second appearance shape judging step as fifth positive abnormal cells; and judging the cells which are judged to be suspicious abnormally hyperplastic cells by the first appearance shape judging step and judging the cells which are judged to be suspicious abnormally hyperplastic cells or positive abnormally hyperplastic cells by the second appearance shape judging step as fifth suspicious cells.
2. The method according to claim 1, wherein the first appearance morphology determining step is implemented by obtaining a parameter of the size of the area of the cell nucleus, and defining it as DNA parameter 1, presetting a threshold a, a threshold B and a threshold C for DNA parameter 1, respectively, and dividing parameter 1 into four numerical ranges, thereby classifying the cells to be analyzed into four types: 1, abnormal cells, wherein the parameter 1 is less than or equal to a threshold value A; suspicious abnormal cells, wherein the parameter is more than the threshold A and less than or equal to the threshold B; ③ highly suspicious abnormal cells, wherein the threshold B is more than the parameter 1 and less than or equal to the threshold C; fourthly, the positive abnormal proliferation cells, the parameter 1 is larger than a threshold value C, and the calculation and the analysis of the parameter 1 are realized through a pre-trained neural network;
the neural network comprises an input layer, a hidden layer and an output layer, the output layer of the neural network outputs scores corresponding to the in-vitro biological samples, and the highest score is used as an analysis index of the first appearance form judgment step;
the neural network comprises a first neural network module, a second neural network module and a classifier, wherein the first neural network module is a Darknet-53 network, the second neural network module comprises 5 convolution layers, a ReLU layer, a pooling layer, a BN + LeakyReLU layer and a convolution layer which alternately appear, and the classifier adopts an LR logical regression classifier.
3. The method according to claim 1, characterized in that before the first appearance morphology determining step, it further comprises a step of acquisition and pre-processing of an image of the cells to be analyzed, in particular comprising the following steps:
(1) amplifying the cells under a microscope by different times, automatically focusing the cells through the microscope and acquiring a target picture to be analyzed; dividing each picture of a group of pictures shot by cells in the same visual field of a microscope into three channels of R (red), G (green) and B (blue), compounding any two channels to obtain a first compound image, calculating Variance of each picture relative to the first compound image to obtain Variance value, comparing the Variance values of each picture, and selecting the picture with the maximum Variance value relative to the first compound image as a first target picture of the group of pictures;
(2) and fusing a plurality of first target pictures of different levels in the same field to obtain an image of the cell to be analyzed.
4. The method of claim 3, wherein step (2) is implemented as follows: defining one first target picture as A (X, y), defining the previous picture as A _ Pre (X, y), defining the Next picture as A _ Next (X, y), defining the fused image as Fuse _ image (X, y), respectively setting the pixel values of the three images at the point (X, y) as A (X, y) value, respectively setting the A _ Pre (X, y) value and the A _ Next (X, y) value, comparing the 3X3 areas, respectively calculating the sum of the 3X3 area pixel values of the three first target pictures to obtain A _ add _ value, respectively calculating the sum of the A _ Pre _ add _ value and the A _ Next _ add _ value, and respectively obtaining the fused image at the point (X, y) value, which is the integral value of the fused image after traversing the point (X, y) corresponding to the maximum value.
5. The method of claim 4, wherein the nuclei of the image of the cell to be analyzed are located and identified and the size of the area of the nuclei is calculated.
6. The method according to claim 5, wherein the second appearance morphological assessment step comprises a step of calculating the size of the cytoplasmic area of the positive dysplasia cells or highly suspicious dysplasia cells located by the first appearance morphological assessment step.
7. The method according to claim 6, wherein the size of the cytoplasmic area is calculated by separating the positive dysplasia cells or highly suspicious dysplasia cells after localization and segmentation to obtain a red channel image R (x, y), a green channel image G (x, y) and a blue channel image B (x, y), obtaining a second composite image by using the image composition of any two channels, and obtaining a Mean value Mean _ value of the second composite image; subtracting each pixel of the second composite image from Mean _ value 0.8, the new pixel value being equal to each pixel value of the second composite image-Mean _ value 0.8, and assigning it 255 when the new pixel value is greater than 0; and when the new pixel value is less than 0, assigning the new pixel value as 0, obtaining a binary image Threshold (x, y) according to the value, obtaining a convex hull of the Threshold (x, y) image, obtaining the area of the convex hull, and finally finding out the largest area, namely the cytoplasm area of the cell.
8. The method according to claim 7, wherein the second appearance morphology determining step further comprises dividing the DNA parameter 1 of the positive dysplasia cell or the highly suspicious dysplasia cell obtained by the first appearance morphology determining step by the cytoplasmic area of the corresponding cell to obtain the nuclear plasma ratio, and calculating the DNA parameter 2.
9. A computer arrangement comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to carry out the steps of the method according to any one of claims 1-8.
10. A storage medium, characterized in that a computer program is stored which, when being executed by a processor, causes the processor to carry out the steps of the method according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110570957.XA CN113295692B (en) | 2021-05-25 | 2021-05-25 | Cell analysis method based on cell nuclear DNA and TBS double analysis method, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110570957.XA CN113295692B (en) | 2021-05-25 | 2021-05-25 | Cell analysis method based on cell nuclear DNA and TBS double analysis method, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113295692A true CN113295692A (en) | 2021-08-24 |
| CN113295692B CN113295692B (en) | 2024-06-21 |
Family
ID=77324731
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110570957.XA Active CN113295692B (en) | 2021-05-25 | 2021-05-25 | Cell analysis method based on cell nuclear DNA and TBS double analysis method, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113295692B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011075278A (en) * | 2009-09-29 | 2011-04-14 | Olympus Corp | Method of analyzing structure composing cell nucleus and method of analyzing form of cell nucleus |
| CN106875404A (en) * | 2017-01-18 | 2017-06-20 | 宁波摩视光电科技有限公司 | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image |
| CN109145941A (en) * | 2018-07-03 | 2019-01-04 | 怀光智能科技（武汉）有限公司 | A kind of irregular cervical cell group's image classification method and system |
| CN109190567A (en) * | 2018-09-10 | 2019-01-11 | 哈尔滨理工大学 | Abnormal cervical cells automatic testing method based on depth convolutional neural networks |
| CN110120040A (en) * | 2019-05-13 | 2019-08-13 | 广州锟元方青医疗科技有限公司 | Sectioning image processing method, device, computer equipment and storage medium |
| CN110705583A (en) * | 2019-08-15 | 2020-01-17 | 平安科技（深圳）有限公司 | Cell detection model training method and device, computer equipment and storage medium |
| CN111797786A (en) * | 2020-07-09 | 2020-10-20 | 郑州中普医疗器械有限公司 | Detection method for in vitro biological samples, four-classification, computer device and computer-readable storage medium |
-
2021
- 2021-05-25 CN CN202110570957.XA patent/CN113295692B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011075278A (en) * | 2009-09-29 | 2011-04-14 | Olympus Corp | Method of analyzing structure composing cell nucleus and method of analyzing form of cell nucleus |
| CN106875404A (en) * | 2017-01-18 | 2017-06-20 | 宁波摩视光电科技有限公司 | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image |
| CN109145941A (en) * | 2018-07-03 | 2019-01-04 | 怀光智能科技（武汉）有限公司 | A kind of irregular cervical cell group's image classification method and system |
| CN109190567A (en) * | 2018-09-10 | 2019-01-11 | 哈尔滨理工大学 | Abnormal cervical cells automatic testing method based on depth convolutional neural networks |
| CN110120040A (en) * | 2019-05-13 | 2019-08-13 | 广州锟元方青医疗科技有限公司 | Sectioning image processing method, device, computer equipment and storage medium |
| CN110705583A (en) * | 2019-08-15 | 2020-01-17 | 平安科技（深圳）有限公司 | Cell detection model training method and device, computer equipment and storage medium |
| CN111797786A (en) * | 2020-07-09 | 2020-10-20 | 郑州中普医疗器械有限公司 | Detection method for in vitro biological samples, four-classification, computer device and computer-readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113295692B (en) | 2024-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10083340B2 (en) | Automated cell segmentation quality control | |
| Ortega et al. | Hyperspectral imaging and deep learning for the detection of breast cancer cells in digitized histological images | |
| Meijering et al. | Tracking in molecular bioimaging | |
| JP6192747B2 (en) | Machine learning system based on tissue objects for automatic scoring of digital hall slides | |
| Deng et al. | A classification–detection approach of COVID-19 based on chest X-ray and CT by using keras pre-trained deep learning models | |
| Poostchi et al. | Malaria parasite detection and cell counting for human and mouse using thin blood smear microscopy | |
| Wu et al. | A hematologist-level deep learning algorithm (BMSNet) for assessing the morphologies of single nuclear balls in bone marrow smears: algorithm development | |
| CN111797786B (en) | Detection method for in vitro biological sample, four-classification, computer device and computer-readable storage medium | |
| WO2012041333A1 (en) | Automated imaging, detection and grading of objects in cytological samples | |
| Chakrabortya et al. | A combined algorithm for malaria detection from thick smear blood slides | |
| Liu et al. | Platelet detection based on improved YOLO_v3 | |
| Acar et al. | Coupled analysis of in vitro and histology tissue samples to quantify structure-function relationship | |
| US20180040120A1 (en) | Methods for quantitative assessment of mononuclear cells in muscle tissue sections | |
| JP2022547722A (en) | Weakly Supervised Multitask Learning for Cell Detection and Segmentation | |
| Song et al. | Hybrid deep autoencoder with Curvature Gaussian for detection of various types of cells in bone marrow trephine biopsy images | |
| Radha | Analysis of COVID-19 and pneumonia detection in chest X-ray images using deep learning | |
| Matias et al. | Segmentation, detection, and classification of cell nuclei on oral cytology samples stained with papanicolaou | |
| Wang et al. | Automated identification of analyzable metaphase chromosomes depicted on microscopic digital images | |
| Yang et al. | Advances in AI‐based cancer cytopathology | |
| Hu et al. | Automatic detection of tuberculosis bacilli in sputum smear scans based on subgraph classification | |
| Yan et al. | HLDnet: Novel deep learning based artificial intelligence tool fuses acetic acid and Lugol’s iodine cervicograms for accurate pre-cancer screening | |
| CN113178228B (en) | Cell analysis method based on nuclear DNA analysis, computer device, and storage medium | |
| Hao et al. | DBM-ViT: A multiscale features fusion algorithm for health status detection in CXR/CT lungs images | |
| Durkee et al. | Pseudo-spectral angle mapping for automated pixel-level analysis of highly multiplexed tissue image data | |
| CN113295692B (en) | Cell analysis method based on cell nuclear DNA and TBS double analysis method, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |