CN114897872A - Method and device suitable for identifying cells in cell cluster and electronic equipment - Google Patents
Method and device suitable for identifying cells in cell cluster and electronic equipment Download PDFInfo
- Publication number
- CN114897872A CN114897872A CN202210617409.2A CN202210617409A CN114897872A CN 114897872 A CN114897872 A CN 114897872A CN 202210617409 A CN202210617409 A CN 202210617409A CN 114897872 A CN114897872 A CN 114897872A
- Authority
- CN
- China
- Prior art keywords
- cell
- image
- region image
- sample
- cells
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
- G06T7/0014—Biomedical image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
- G06V20/695—Preprocessing, e.g. image segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
- G06V20/698—Matching; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The application provides a method and a device for identifying cells in a cell mass and electronic equipment, wherein the method for identifying the cells in the cell mass comprises the following steps: identifying at least one cell mass region image from the digital slice images to be detected; for each cell mass region image, identifying each single cell region image and a standard cell image from the cell mass region images; correspondingly comparing the single cell region image with the standard cell image aiming at each single cell region image in each cell mass region image to determine whether the single cell in the single cell region image is a suspicious cell; and inputting suspicious cell images corresponding to all suspicious cells into the trained cell classification model, and determining whether the target cells of the target type exist in the digital slice images. The method and the device can improve the accuracy of identifying the target cells in the cell mass.
Description
Technical Field
The present disclosure relates to the field of cell identification technologies, and in particular, to a method and an apparatus for identifying cells in a cell cluster, and an electronic device.
Background
With the rapid development of biological industry, cells are an important component of a living body, and more people are invested in the research on the cells, however, in the existing cell recognition field, most of research and recognition are directed to single cells, but in addition to free single cells, a large number of cell clusters exist in the living body, and the analysis on the cell clusters in the prior art is less.
The cell mass is a mass tissue with a large number of cells, and the cell mass has a large number of cells, so that the staining is deep, the overlapping is serious, and the information of cell nuclei is unclear, so that the cell types in the cell mass cannot be accurately identified.
Disclosure of Invention
In view of the above, an object of the present invention is to provide a method, an apparatus and an electronic device for identifying cells in a cell mass, which can improve the accuracy of identifying target cells in the cell mass.
The embodiment of the application provides an identification method suitable for cells in a cell mass, which comprises the following steps:
identifying at least one cell mass region image from the digital slice images to be detected;
for each cell mass region image, identifying a single cell region image and a standard cell image from the cell mass region image;
correspondingly comparing the single cell region image with the standard cell image aiming at each single cell region image in each cell mass region image to determine whether the single cell in the single cell region image is a suspicious cell;
and inputting the suspicious cell image corresponding to each suspicious cell into a trained cell classification model, and determining whether the digital slice image has the target cell of the target type.
Further, for each of the cell mass region images, identifying a single cell region image from the cell mass region images includes:
for each cell mass region image, identifying each cell nucleus from the cell mass region image, and determining the cell image corresponding to each cell nucleus as a single cell region image;
judging whether the target distance between any two single cell region images in the cell mass region images is greater than or equal to a preset threshold value or not;
and if so, determining the target region image between any two single cell region images in the cell mass region image as the single cell region image.
Further, the comparing, for each single cell region image in each cell mass region image, the single cell region image with the standard cell image to determine whether the single cell in the single cell region image is a suspicious cell includes:
inputting the single cell region image and a standard cell image into a trained cell screening model aiming at each single cell region image in each cell mass region image, and determining the similarity between each single cell and a standard cell in the cell mass region image;
and if the similarity is smaller than a preset threshold value, determining the single cell corresponding to the similarity as a suspicious cell in the cell mass region image.
Further, the trained cell screening model is determined by:
obtaining different types of sample single cell region images in a sample digital slice image, type labels of the sample single cell region images and sample standard cell images corresponding to the sample digital slice image; the type label is used for representing the real sample similarity between the sample single cells and the preset sample standard cells;
inputting the sample single cell region image and the label of the sample single cell region image into an initial cell screening model, and determining the preset sample similarity between the sample single cell and a sample standard cell;
and when the loss value between the preset sample similarity and the real sample similarity between the sample single cell and the sample standard cell is smaller than a preset threshold value, training is stopped, and a trained cell screening model is determined.
Further, determining a loss value between the preset sample similarity and the real sample similarity by the following formula:
Y＝|A|，A＜0；
Y＝0，A＞0；
wherein the content of the first and second substances,
y is used for representing a loss value between the preset sample similarity and the real sample similarity, i is used for representing the number of the feature vectors of the sample single cells, and j is used for representing the number of the feature vectors of the sample standard cells; v. of bj A feature vector for characterizing a standard cell of the sample; v. of ai And v ci Respectively characterizing the feature vectors of the first type of sample single cells and the feature vectors of the second type of sample single cells; n is used for characterizing the number of single cells in the sample of the first type; 3N is the number of single cells in the second type of sample.
The embodiment of the present application also provides an identification apparatus suitable for a cell in a cell cluster, and an identification apparatus suitable for a cell in a cell cluster includes:
the first identification module is used for identifying at least one cell mass area image from the digital slice image to be detected;
the second identification module is used for identifying each single cell region image and a standard cell image from the cell mass region images aiming at each cell mass region image;
a first determining module, configured to compare the single cell region image with the standard cell image correspondingly for each single cell region image in each cell mass region image, and determine whether a single cell in the single cell region image is a suspicious cell;
and the second determination module is used for inputting suspicious cell images corresponding to the suspicious cells into a trained cell classification model and determining whether the digital slice images have target cells of a target type.
Further, the second identification module is specifically configured to:
for each cell mass region image, identifying each cell nucleus from the cell mass region image, and determining the cell image corresponding to each cell nucleus as a single cell region image;
judging whether the target distance between any two single cell region images in the cell mass region images is greater than or equal to a preset threshold value or not;
and if so, determining the target region image between any two single cell region images in the cell mass region image as the single cell region image.
Further, the first determining module is specifically configured to:
inputting the single cell region image and a standard cell image into a trained cell screening model aiming at each single cell region image in each cell mass region image, and determining the similarity between each single cell and a standard cell in the cell mass region image;
and if the similarity is smaller than a preset threshold value, determining the single cell corresponding to the similarity as a suspicious cell in the cell mass region image.
An embodiment of the present application further provides an electronic device, including: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is operated, the machine-readable instructions, when executed by the processor, performing the steps of the method for identifying cells in a cell mass as described above.
Embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the method for identifying cells in a cell mass as described above.
Compared with the prior art, the method, the device and the electronic equipment for identifying the cells in the cell mass provided by the embodiment of the application compare the single cell area image in the digital slice image to be detected with the standard cell image, further accurately identify the suspicious cells after comparison, determine the target type of the target cells in the digital slice image, and improve the accuracy of identifying the target cells in the cell mass.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
Fig. 1 is a flowchart illustrating a method for identifying cells in a cell mass according to an embodiment of the present disclosure;
FIG. 2 is a diagram illustrating a schematic diagram of an image of a single cell region in a method for identifying cells in a cell mass according to an embodiment of the present application;
FIG. 3 is a second flow chart of another method for identifying cells in a cell mass according to the embodiment of the present disclosure;
FIG. 4 is a flow chart illustrating an embodiment of a method for identifying cells in a cell mass according to the present disclosure;
fig. 5 is a schematic structural diagram illustrating an identification apparatus suitable for identifying cells in a cell mass according to an embodiment of the present application;
fig. 6 shows a schematic structural diagram of an electronic device provided in an embodiment of the present application.
In the figure:
500-an identification means suitable for cells in a cell mass; 510-a first identification module; 520-a second identification module; 550-a first determining module; 540-a second determination module; 600-an electronic device; 610-a processor; 620-memory; 630-bus.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. Every other embodiment that can be obtained by a person skilled in the art without making creative efforts based on the embodiments of the present application falls within the protection scope of the present application.
First, an application scenario to which the present application is applicable will be described. Research shows that in the prior art, a cell mass is a mass tissue with a large number of cells, and the cell mass aggregates a large number of cells, so that the staining is deep, the overlapping is serious, and the information of cell nucleuses is unclear, so that the cell types in the cell mass cannot be accurately identified.
Based on this, the embodiments of the present application provide a method and an apparatus for identifying cells in a cell mass, and an electronic device, which can improve the accuracy of identifying target cells in the cell mass.
Referring to fig. 1, fig. 1 is a flowchart illustrating a method for identifying cells in a cell mass according to an embodiment of the present disclosure. As shown in fig. 1, the identification method applied to the cells in the cell mass provided in the embodiment of the present application includes:
s101, identifying at least one cell mass area image from the digital slice image to be detected.
The method comprises the steps of firstly decoding and digitally converting a slice image to be detected through a corresponding decoding program to generate a digital slice image to be detected, then processing the digital slice image to be detected according to a plurality of resolutions to determine a target digital slice image corresponding to each resolution, then respectively inputting the digital slice image to be detected corresponding to each resolution into a trained cell mass detection model under the corresponding resolution, determining cell mass detection results under each resolution, remapping the cell mass detection results under the plurality of resolutions to the current maximum resolution, merging the cell mass detection results under each resolution, and identifying at least one cell mass area image.
The cell mass in the cell mass region image can be a cell mass consisting of a normal cell structure or a cell mass consisting of an atrophied cell structure, and because the atrophied cell mass often gathers a large number of atrophied cells, the staining is deep, the overlapping is serious, and it is difficult to accurately identify the target cells of the target type, the method for identifying the cells in the cell mass is adopted to identify the target cells of the target type.
Here, the resolution division may be customized according to an actual reference scene, and the multiple resolutions in the embodiment provided by the present application may be 40 times, 20 times, and 10 times, and thus, the maximum resolution in the embodiment provided by the present application is 40 times.
Wherein, the cell mass detection result under each resolution includes but is not limited to the position and size relationship of the cell mass; and methods for combining the cell mass measurements at each resolution include, but are not limited to, combining using non-maxima suppression methods.
In this way, in the embodiment provided by the present application, the digital slice image to be detected with the resolution of 40 times may be specifically set to be the digital slice image to be detected of 20000x20000, the digital slice image to be detected with the resolution of 20 times may be specifically set to be the digital slice image to be detected of 10000x10000, and the digital slice image to be detected with the resolution of 10 times may be specifically set to be the digital slice image to be detected of 5000x 5000.
Here, the cell mass detection model trained at each resolution is determined by:
and cutting the sample digital slice image under each resolution according to a preset size (such as 512x512), labeling a significant cell mass region of each cut image, and determining a sample cell mass image.
And inputting the sample cell mass image into an initial cell mass detection model for training, and determining the trained cell mass detection model under each resolution.
The initial cell mass detection model can be, but is not limited to, a detection network model such as fast RCNN and SSD series.
In the above, the detection network models such as fast RCNN and SSD series require that the sample cell mass image under the corresponding resolution cannot exceed the standard image size (e.g. 256 × 256).
And S102, aiming at each cell mass region image, identifying each single cell region image and one standard cell image from the cell mass region images.
In the step, for each cell mass region image, each single cell region image is identified from the cell mass region image through a trained cell segmentation model, and for each cell mass region image, a self-defined standard cell image is marked, wherein the self-defined standard cell image is a cell of the same cell type.
The trained cell segmentation model can adopt, but is not limited to, network segmentation models such as Mask-RCNN, Unet and Deeplab series.
Thus, a trained cell segmentation model is determined by:
and labeling the sample cell mass image, and determining a sample single cell region image in the sample cell mass image.
Inputting the sample single cell region image and the label of the sample single cell region image into an initial cell segmentation model, training the initial cell segmentation model, and determining each sample cell nucleus in the sample cell mass image.
In the above, the label of the sample single cell region image represents the position information and the region size of the sample single cell region image, and the output of the position information and the region size of each sample cell nucleus is the initial cell segmentation model in the training process to obtain the preset position information and the preset region size of the sample single cell region image.
And performing loss value training on the preset position information, the preset area size and the label of the truly labeled sample single cell area image, stopping training when the loss value is smaller than a preset threshold value, and determining a trained cell segmentation model.
In the practical scene application process, a worker can use the initial cell segmentation models with different network parameters to perform model training, and selects the cell segmentation model with the optimal training effect as the final trained cell segmentation model.
In step S102, for each of the cell mass region images, identifying a single cell region image from the cell mass region images includes:
step 1021, identifying each cell nucleus from the cell mass region images according to each cell mass region image, and determining the cell image corresponding to each cell nucleus as a single cell region image.
And aiming at each cell mass region image, determining each cell nucleus in each cell mass region image through a trained cell segmentation model, and determining the cell image corresponding to each cell nucleus as a single cell region image.
Thus, the cell image corresponding to each cell nucleus includes the position information corresponding to the cell nucleus and the size of the region dividing the cell nucleus.
And 1022, judging whether the target distance between any two single cell region images in the cell mass region images is greater than or equal to a preset threshold value.
The preset threshold value can be adjusted and set in a user-defined manner according to the actual conditions of different application scenes, for example, according to the slice size of the digital slice image to be detected, and here, the preset threshold value in the embodiment provided by the application can be set to be 1.5 of the size of the single cell region image.
And 1023, if the cell mass area image is larger than the single cell area image, determining the target area image between any two single cell area images in the cell mass area image as the single cell area image.
Fig. 2 is a structural diagram of a single cell region image in the identification method for cells in a cell cluster according to an embodiment of the present disclosure, where fig. 2 shows that when it is determined that a target distance between any two single cell region images in the cell cluster region image is greater than or equal to a preset threshold, the target region between any two single cell region images in the cell cluster region image is also determined as the single cell region image, and the single cell region image at this time includes an image corresponding to the target region in addition to the stored single cell region image in the cell cluster region image.
Therefore, the purpose of determining the target region as the single cell region image is to prevent the certainty and omission of single cells when the cell cluster region image is segmented, ensure the accuracy of the subsequent cell cluster region image and further improve the identification precision of the target type of the target cells in the digital section image to be detected.
S103, correspondingly comparing the single cell region image with the standard cell image aiming at each single cell region image in each cell mass region image, and determining whether the single cell in the single cell region image is a suspicious cell.
In the step, each single cell region image in each cell mass region image is compared with the standard cell image, the distance between each single cell region image and the standard cell image is determined, and the distance is compared with a preset distance to determine whether the single cell in the single cell region image is a suspicious cell.
Thus, when the distance is less than the distance, the single cell in the single cell region image is determined to be a suspicious cell.
Here, the preset distance may be set in a user-defined manner according to different application scenarios, actual requirements, and actually required suspicious cell accuracy, and each cell mass region image has only one standard cell image.
S104, inputting the suspicious cell image corresponding to each suspicious cell into a trained cell classification model, and determining whether the digital slice image contains the target cell of the target type.
In this step, the target cell is a cell of a target type that the worker wants to identify from the digital slice image to be detected, for example: the embodiment provided by the application takes the digital slice image to be detected for identifying the cervical epithelial cells as an example, the target cells with positive target type and the negative cells with negative target type are determined in the atrophic clumped cells, and the embodiment is used for helping the staff as the auxiliary medical reference.
Therefore, suspicious cell images corresponding to all suspicious cells are input into the trained cell classification model, the confidence value (between 0 and 1.0) of the target cell with the positive target type corresponding to each suspicious cell is output from the trained cell classification model, and then the suspicious cells with the confidence values exceeding the preset confidence value are determined as the target cells.
In the above, the trained cell classification model may be, but is not limited to, network classification models of ResNet series, inclusion series, and SeNet, and the embodiment provided in this application takes a network classification model of ResNet-101 type as an example.
Here, the training dataset and the validation dataset of the trained cell classification model are determined by:
marking target cells of the sample single cell region image according to target types, and setting the marked target types into a training data set and a verification data set according to certain preset proportion.
In the above, the labeled target type is used to perform target cell, different initial cell classification models using different network parameters are trained, and then the cell classification model with the most accurate training result is selected as the trained cell classification model in the embodiment of the present application.
Compared with the cell identification method in the prior art, the identification method suitable for the cells in the cell mass provided by the embodiment of the application compares the single cell region image in the digital slice image to be detected with the standard cell image, further accurately identifies the compared suspicious cells, determines the target type of the target cells in the digital slice image, realizes the adoption of a dual detection and identification model, and improves the accuracy of identifying the target cells in the normal cell mass and the atrophic cell mass.
Referring to fig. 3, fig. 3 is a second flowchart of a method for identifying cells in a cell mass according to another embodiment of the present application. As shown in fig. 3, the identification method applied to cells in a cell mass provided in the embodiments of the present application includes:
s301, identifying at least one cell mass region image from the digital slice image to be detected.
S302, aiming at each cell mass region image, identifying each single cell region image and a standard cell image from the cell mass region images.
S303, aiming at each single cell region image in each cell mass region image, inputting the single cell region image and a standard cell image into a trained cell screening model, and determining the similarity between each single cell and a standard cell in the cell mass region image.
In this step, for each single cell region image in the cell cluster region image, inputting the single cell region image and the standard cell image into a trained cell screening model, extracting an image feature of each single cell region image and an image feature of the standard cell image, and determining a similarity between each single cell in the cell cluster region image and the standard cell according to a cosine distance or an euclidean distance between the image feature of each single cell region image and the image feature of the standard cell image, that is, the trained cell screening model outputs the cosine distance or the euclidean distance between the trained cell screening model and the standard cell image.
The trained cell screening model can adopt but is not limited to a network screening model of ResNet-51 type.
Optionally, the trained cell screening model is determined by:
obtaining different types of sample single cell region images in a sample digital slice image, labels of the sample single cell region images and sample standard cell images corresponding to the sample digital slice image; the label of the sample single cell region image is used for representing the position information and the region size of the sample single cell region image.
Inputting the sample single cell region image and the label of the sample single cell region image into an initial cell screening model, and determining preset sample distances between different types of sample single cells and sample standard cells.
The preset sample distance between the sample single cells of different types and the sample standard cells can be specifically a cosine distance between the sample single cells and the sample standard cells.
And when the loss value between the preset sample distance and the real sample distance between the sample single cells of different types and the sample standard cells is smaller than a preset threshold value, training is stopped, and a trained cell screening model is determined.
Wherein a loss value between the preset sample similarity and the true sample similarity is determined by the following formula:
Y＝|A|，A＜0；
Y＝0，A＞0；
wherein the content of the first and second substances,
y is used for representing a loss value between the preset sample similarity and the real sample similarity, i is used for representing the number of the feature vectors of the sample single cells, and j is used for representing the number of the feature vectors of the sample standard cells; v. of bj A feature vector for characterizing a standard cell of the sample; v. of ai And v ci Respectively characterizing the feature vectors of the first type of sample single cells and the feature vectors of the second type of sample single cells; n is used for characterizing the number of single cells in the sample of the first type; 3N is the number of single cells in the second type of sample.
Here, "10" in the formula is used to characterize the number of standard cells of the sample as 10.
In the examples provided in this application, N > -32.
In the above, the distances between the images of different types of sample single cell regions are different from the images of sample single cell regions, and in the present application, taking the positive cells and the negative cells of the atrophic cell mass in the digital cervical epithelial cell section image as an example, the distance between the image of the positive cell region and the image of the sample single cell region is greater than the distance between the negative cells and the image of the sample single cell region.
S304, if the similarity is smaller than a preset threshold value, determining the single cell corresponding to the similarity as a suspicious cell in the cell mass region image.
In this step, if the similarity between the sample single cells of different types and the sample standard cells is smaller than a preset threshold, that is, the cosine distance or the euclidean distance between the cell screening model and the standard cell image is smaller than the preset threshold, it is indicated that the probability that the sample single cells of the type are suspicious cells is relatively high.
The preset threshold value can be set by self according to the actual application scene.
S305, inputting suspicious cell images corresponding to the suspicious cells into a trained cell classification model, and determining whether target cells of a target type exist in the digital slice images.
The descriptions of S301 to S302 and S305 may refer to the descriptions of S101 to S102 and S104, and the same technical effects can be achieved, which is not described in detail.
Compared with the cell identification method in the prior art, the identification method suitable for the cells in the cell mass provided by the embodiment of the application compares the single cell region image in the digital slice image to be detected with the standard cell image, further accurately identifies the compared suspicious cells, determines the target type of the target cells in the digital slice image, realizes the adoption of a dual detection and identification model, and improves the accuracy of identifying the target cells in the normal cell mass and the atrophic cell mass.
The following describes a flow of the identification method applied to the cells in the cell mass according to an embodiment of the present application, and as shown in fig. 4, fig. 4 shows a flow chart of an embodiment of the identification method applied to the cells in the cell mass, and the specific flow chart is as follows:
s401, a digital section to be detected related to cervical cytology is acquired.
Wherein, the scanning multiple of the digital slice to be detected is 40 times, and the scanned section size is 20000x 20000.
S402, decoding the digital slice to be detected according to a corresponding decoding program to obtain a digital slice image to be detected, and determining a standard cell image of the digital slice image to be detected.
Wherein the standard cell image can be used to represent I b To indicate.
And S403, zooming the digital slice image to be detected to obtain two zoomed digital slice images to be detected.
Wherein, the zoom times are 20 times and 10 times, and the sizes of the two digital slice images to be detected after zooming are 10000x10000 (corresponding to 20 times) and 5000x5000 (corresponding to 10 times).
And S404, respectively carrying out image blocking processing on the digital slice images to be detected under different resolutions to obtain image blocks with preset sizes.
Wherein, the size of each image block is 512x512, and there is 16 pixel overlap between each image block in the horizontal and vertical directions.
S405, aiming at each image block, selecting a trained fine cell mass detection model corresponding to the resolution from the three corresponding trained cell mass detection models to detect cell masses, respectively obtaining position information and size relation of the cell masses in the digital slice images to be detected under three respective rates, and combining through a non-maximum inhibition method to obtain each single cell region image.
After the position information and the size relationship of the cell mass are obtained, the position information and the size relationship of the cell mass need to be mapped back to the resolution of the current maximum multiple, and each single cell region image can be used as { I } i I 1.., M }.
S406, judging whether the target distance between any two single cell region images is larger than or equal to a preset threshold value, and if so, determining the target region image between any two single cell region images in the cell mass region image as the single cell region image.
Wherein each single cell region image is represented as I i，j J 1.. and R, and the preset threshold may be designed to be, for example, 1.5 times of the image of the single cell region according to the actual application scenario.
S407, judging each single cell region image I i，j And standard cell image I b Selecting two single cell region images I with cosine distance smaller than preset distance i，j As suspicious positive cells.
Wherein each single cell region image I i，j And standard cell image I b The distance between may be a cosine distance or a euclidean distance.
S408, inputting the suspicious positive cells into a trained cell classification model, determining whether target cells of a target type exist in the digital slice image, if so, determining a confidence value of the target cells, determining the target type of the target cells according to the confidence value of the target cells, and feeding back an identification result to a worker.
And the staff can judge whether the target positive cells or the target negative cells exist in the digital slice image according to the feedback result.
Referring to fig. 5, fig. 5 is a schematic structural diagram of an identification apparatus for cells in a cell mass according to an embodiment of the present application, where the identification apparatus 500 for cells in a cell mass includes:
a first identification module 510 for identifying at least one cell mass region image from the digital slice images to be detected.
A second identifying module 520, configured to identify, for each of the cell mass region images, a single cell region image and a standard cell image from the cell mass region images.
Optionally, the second identifying module 520 is specifically configured to:
for each cell mass area image, identifying each cell nucleus from the cell mass area image, and determining the cell image corresponding to each cell nucleus as a single cell area image.
And judging whether the target distance between any two single cell region images in the cell mass region images is greater than or equal to a preset threshold value.
And if the cell mass area image is larger than the cell mass area image, determining the target area image between any two single cell area images in the cell mass area image as the single cell area image.
A first determining module 550, configured to compare the single cell region image with the standard cell image correspondingly for each single cell region image in each cell mass region image, and determine whether a single cell in the single cell region image is a suspicious cell.
Optionally, the first determining module 550 is specifically configured to:
and inputting the single cell region image and a standard cell image into a trained cell screening model aiming at each single cell region image in each cell mass region image, and determining the similarity between each single cell in the cell mass region image and a standard cell.
The trained cell screening model is obtained by training sample single cell region images of different types in a sample digital slice image and sample standard cell images in the sample digital slice image.
And if the similarity is smaller than a preset threshold value, determining the single cell corresponding to the similarity as a suspicious cell in the cell mass region image.
Optionally, the trained cell screening model is determined by:
obtaining different types of sample single cell region images in a sample digital slice image, type labels of the sample single cell region images and sample standard cell images corresponding to the sample digital slice image; the type label is used for characterizing the real sample similarity between the sample single cells and the preset sample standard cells.
Inputting the sample single cell region image and the label of the sample single cell region image into an initial cell screening model, and determining the preset sample similarity between the sample single cell and a sample standard cell.
And when the loss value between the preset sample similarity and the real sample similarity between the sample single cell and the sample standard cell is smaller than a preset threshold value, training is stopped, and a trained cell screening model is determined.
Optionally, the loss value between the preset sample similarity and the real sample similarity is determined by the following formula:
Y＝|A|，A＜0；
Y＝0，A＞0；
wherein the content of the first and second substances,
y is used for representing a loss value between the preset sample similarity and the real sample similarity, i is used for representing the number of the feature vectors of the sample single cells, and j is used for representing the number of the feature vectors of the sample standard cells; v. of bj A feature vector for characterizing a standard cell of the sample; v. of ai And v ci Respectively characterizing the feature vectors of the first type of sample single cells and the feature vectors of the second type of sample single cells; n is used for characterizing the number of single cells in the sample of the first type; 3N is the number of single cells in the second type of sample.
The second determining module 540 is configured to input the suspicious cell image corresponding to each suspicious cell into the trained cell classification model, and determine whether a target cell of the target type exists in the digital slice image.
In the above, the application scenarios of the identification apparatus 500 suitable for the cells in the cell mass provided by the embodiments of the present application are determined by different cell mass types, and here, include, but are not limited to, the following two application scenarios:
the application scene one:
the recognition device 500 suitable for the cells in the cell mass provided by the application can be directly used for recognizing the target type of the target cells in the digital slice image to be detected, and is used for providing the recognition result for workers to further distinguish.
Application scenario two:
the identification device 500 for cells in a cell mass provided by the present application can be used in combination with an auxiliary diagnostic product or system of existing cytology, and the identification result can be used as auxiliary data to be provided to the worker for reference in combination with the detection result of the auxiliary product of existing cytology.
Compared with the cell identification method in the prior art, the identification device 500 suitable for the cells in the cell mass provided by the embodiment of the application compares the single cell region image in the digital slice image to be detected with the standard cell image, further accurately identifies the compared suspicious cells, determines the target type of the target cells in the digital slice image, realizes the adoption of a dual detection and identification model, and improves the accuracy of identifying the target cells in the normal cell mass and the atrophic cell mass.
Referring to fig. 6, fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure. As shown in fig. 6, the electronic device 600 includes a processor 610, a memory 620, and a bus 630.
The memory 620 stores machine-readable instructions executable by the processor 610, when the electronic device 600 runs, the processor 610 communicates with the memory 620 through the bus 630, and when the machine-readable instructions are executed by the processor 610, the steps of the method for identifying cells in a cell cluster in the method embodiments shown in fig. 1 and fig. 2 may be performed.
An embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program may perform the steps of the method for identifying cells in a cell cluster in the method embodiments shown in fig. 1 and fig. 2.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and the like.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the exemplary embodiments of the present application, and are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method for identifying cells in a cell mass, the method comprising:
identifying at least one cell mass region image from the digital slice images to be detected;
for each cell mass region image, identifying a single cell region image and a standard cell image from the cell mass region image;
correspondingly comparing the single cell region image with the standard cell image aiming at each single cell region image in each cell mass region image to determine whether the single cell in the single cell region image is a suspicious cell;
and inputting the suspicious cell image corresponding to each suspicious cell into a trained cell classification model, and determining whether the digital slice image has the target cell of the target type.
2. The method according to claim 1, wherein the identifying the single cell region image from the cell mass region images for each cell mass region image comprises:
for each cell mass region image, identifying each cell nucleus from the cell mass region image, and determining the cell image corresponding to each cell nucleus as a single cell region image;
judging whether the target distance between any two single cell region images in the cell mass region images is greater than or equal to a preset threshold value or not;
and if so, determining the target region image between any two single cell region images in the cell mass region image as the single cell region image.
3. The method according to claim 1, wherein the comparing the single-cell region image with the standard cell image for each single-cell region image in each cell mass region image to determine whether the single cell in the single-cell region image is a suspicious cell comprises:
inputting the single cell region image and a standard cell image into a trained cell screening model aiming at each single cell region image in each cell mass region image, and determining the similarity between each single cell and a standard cell in the cell mass region image;
and if the similarity is smaller than a preset threshold value, determining the single cell corresponding to the similarity as a suspicious cell in the cell mass region image.
4. The method of claim 3, wherein the trained cell screening model is determined by:
obtaining different types of sample single cell region images in a sample digital slice image, type labels of the sample single cell region images and sample standard cell images corresponding to the sample digital slice image; the type label is used for representing the real sample similarity between the sample single cells and the preset sample standard cells;
inputting the sample single cell region image and the label of the sample single cell region image into an initial cell screening model, and determining the preset sample similarity between the sample single cell and a sample standard cell;
and when the loss value between the preset sample similarity and the real sample similarity between the sample single cell and the sample standard cell is smaller than a preset threshold value, training is stopped, and a trained cell screening model is determined.
5. The method according to claim 4, wherein the loss value between the similarity of the predetermined sample and the similarity of the real sample is determined by the following formula:
Y＝|A|，A＜0；
Y＝0，A＞0；
wherein the content of the first and second substances,
y is used for representing a loss value between the preset sample similarity and the real sample similarity, i is used for representing the number of the feature vectors of the sample single cells, and j is used for representing the number of the feature vectors of the sample standard cells; v. of bj A feature vector for characterizing a standard cell of the sample; v. of ai And v ci Respectively characterizing the feature vectors of the first type of sample single cells and the feature vectors of the second type of sample single cells; n is used for characterizing the number of single cells in the sample of the first type; 3N is the number of single cells in the second type of sample.
6. An apparatus for identifying cells in a cell mass, the apparatus comprising:
the first identification module is used for identifying at least one cell mass area image from the digital slice image to be detected;
the second identification module is used for identifying each single cell region image and a standard cell image from the cell mass region images aiming at each cell mass region image;
a first determining module, configured to compare the single cell region image with the standard cell image correspondingly for each single cell region image in each cell mass region image, and determine whether a single cell in the single cell region image is a suspicious cell;
and the second determination module is used for inputting suspicious cell images corresponding to the suspicious cells into a trained cell classification model and determining whether the digital slice images have target cells of a target type.
7. The device according to claim 6, characterized in that said second identification module is specifically configured to:
for each cell mass region image, identifying each cell nucleus from the cell mass region image, and determining the cell image corresponding to each cell nucleus as a single cell region image;
judging whether the target distance between any two single cell region images in the cell mass region images is greater than or equal to a preset threshold value or not;
and if so, determining the target region image between any two single cell region images in the cell mass region image as the single cell region image.
8. The device according to claim 6, wherein the first determining module is specifically configured to:
inputting the single cell region image and a standard cell image into a trained cell screening model aiming at each single cell region image in each cell mass region image, and determining the similarity between each single cell and a standard cell in the cell mass region image;
and if the similarity is smaller than a preset threshold value, determining the single cell corresponding to the similarity as a suspicious cell in the cell mass region image.
9. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating via the bus when the electronic device is operated, the machine-readable instructions being executed by the processor to perform the steps of the method for identifying cells in a cell mass according to any one of claims 1 to 5.
10. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, is adapted to carry out the steps of the method for identifying cells in a cell mass according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210617409.2A CN114897872A (en) | 2022-06-01 | 2022-06-01 | Method and device suitable for identifying cells in cell cluster and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210617409.2A CN114897872A (en) | 2022-06-01 | 2022-06-01 | Method and device suitable for identifying cells in cell cluster and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114897872A true CN114897872A (en) | 2022-08-12 |
Family
ID=82726753
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210617409.2A Pending CN114897872A (en) | 2022-06-01 | 2022-06-01 | Method and device suitable for identifying cells in cell cluster and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114897872A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115700821A (en) * | 2022-11-24 | 2023-02-07 | 广东美赛尔细胞生物科技有限公司 | Cell identification method and system based on image processing |
| CN117115816A (en) * | 2023-10-24 | 2023-11-24 | 深圳市美侨医疗科技有限公司 | Identification method and system for clue cells in leucorrhea microscopic image |
-
2022
- 2022-06-01 CN CN202210617409.2A patent/CN114897872A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115700821A (en) * | 2022-11-24 | 2023-02-07 | 广东美赛尔细胞生物科技有限公司 | Cell identification method and system based on image processing |
| CN115700821B (en) * | 2022-11-24 | 2023-06-06 | 广东美赛尔细胞生物科技有限公司 | Cell identification method and system based on image processing |
| CN117115816A (en) * | 2023-10-24 | 2023-11-24 | 深圳市美侨医疗科技有限公司 | Identification method and system for clue cells in leucorrhea microscopic image |
| CN117115816B (en) * | 2023-10-24 | 2024-02-09 | 深圳市美侨医疗科技有限公司 | Identification method and system for clue cells in leucorrhea microscopic image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111241947B (en) | Training method and device for target detection model, storage medium and computer equipment | |
| CN110020592B (en) | Object detection model training method, device, computer equipment and storage medium | |
| CN109886928B (en) | Target cell marking method, device, storage medium and terminal equipment | |
| CN109977191B (en) | Problem map detection method, device, electronic equipment and medium | |
| CN114897872A (en) | Method and device suitable for identifying cells in cell cluster and electronic equipment | |
| EP3220353B1 (en) | Image processing apparatus, image processing method, and recording medium | |
| CN110390229B (en) | Face picture screening method and device, electronic equipment and storage medium | |
| CN112200081A (en) | Abnormal behavior identification method and device, electronic equipment and storage medium | |
| CN105405130B (en) | License image highlight detection method and device based on cluster | |
| US20190138840A1 (en) | Automatic ruler detection | |
| CN113780201B (en) | Hand image processing method and device, equipment and medium | |
| CN112651953B (en) | Picture similarity calculation method and device, computer equipment and storage medium | |
| CN112036384B (en) | Sperm head shape recognition method, device and equipment | |
| CN116416190A (en) | Flaw detection method and device, electronic equipment and storage medium | |
| CN111950566A (en) | Rotation-invariant HOG infrared image power equipment identification method | |
| CN116596875B (en) | Wafer defect detection method and device, electronic equipment and storage medium | |
| CN108596032B (en) | Detection method, device, equipment and medium for fighting behavior in video | |
| CN111507957A (en) | Identity card picture conversion method and device, computer equipment and storage medium | |
| CN111372042B (en) | Fault detection method and device, computer equipment and storage medium | |
| CN110298302B (en) | Human body target detection method and related equipment | |
| EP3176726A1 (en) | Method and device for positioning human eyes | |
| CN112287905A (en) | Vehicle damage identification method, device, equipment and storage medium | |
| CN116824135A (en) | Atmospheric natural environment test industrial product identification and segmentation method based on machine vision | |
| CN115761842A (en) | Automatic updating method and device for human face base | |
| CN114758384A (en) | Face detection method, device, 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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: Room 601-605, 6th Floor, Building J, Building 5, Yunmi City, No. 19, Ningshuang Road, Yuhuatai District, Nanjing, Jiangsu Province, 210000 Applicant after: Jiuyisanluling medical technology Nanjing Co.,Ltd. Address before: Room 305, building A3, Nanhai Science Park, 180 software Avenue, Yuhuatai District, Nanjing City, Jiangsu Province, 210000 Applicant before: Jiuyisanluling medical technology Nanjing Co.,Ltd. |