CN109859218B - Pathological graph key area determination method and device, electronic equipment and storage medium - Google Patents
Pathological graph key area determination method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN109859218B CN109859218B CN201910138631.2A CN201910138631A CN109859218B CN 109859218 B CN109859218 B CN 109859218B CN 201910138631 A CN201910138631 A CN 201910138631A CN 109859218 B CN109859218 B CN 109859218B
- Authority
- CN
- China
- Prior art keywords
- pathological
- processed
- target
- subgraph
- result
- 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.)
- Active
Links
Images
Landscapes
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The embodiment of the invention provides a method and a device for determining a key area of a pathological diagram, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring a pathological diagram to be processed; dividing the pathology diagram to be processed into a plurality of pathology subgraphs to be processed; inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed; for each target pathological subgraph to be processed, inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein a target information region is segmented out; and combining the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. The invention reduces the loss of the target information in the pathological graph and obtains all the pathological graph key areas of the target information.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method and an apparatus for determining a critical area of a pathological diagram, an electronic device, and a storage medium.
Background
With the development of computer science technology, the pathological image digitization technology has made great progress. The technology mainly comprises a WSI (white Slide Image) scanning technology and related automatic processing of computer software. Particularly in recent years, with the generation of high-resolution cameras and the continuous improvement of computer processing performance, the magnification and definition of the WSI can fully meet the daily diagnosis of pathologists, and the work of the pathologists is transferred from a microscope to a computer. Through the designated WSI processing software, a doctor can conveniently and manually drag pathological images, search focus areas, carry out corresponding labeling and the like, and the diagnosis efficiency is accelerated to a certain extent. Meanwhile, the WSI data also contains the characteristics of the traditional electronic image data, can be subjected to processing such as cropping, down-sampling scaling and the like, and is more operable.
In the prior art, the WSI is processed by a conventional convolutional neural network, and the method comprises the following steps: and directly scaling the WSI picture to a proper size, and further researching key information contained in the scaled WSI picture.
However, the inventor finds that in a key information research method of scaling the WSI picture to a proper size, target information locally contained in a pathological diagram is easily lost. Therefore, how to reduce the loss of target information in the pathological graph and obtain all the pathological graph key areas of the target information is still an urgent technical problem to be solved.
Disclosure of Invention
The embodiment of the invention aims to provide a method and a device for determining a pathological diagram key area, electronic equipment and a storage medium, so as to reduce the loss of target information in a pathological diagram and obtain the pathological diagram key area of all target information. The specific technical scheme is as follows:
in a first aspect, an embodiment of the present invention discloses a method for determining a critical area of a pathological diagram, where the method includes:
acquiring a pathological diagram to be processed;
dividing the pathology diagram to be processed into a plurality of pathology subgraphs to be processed;
inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed;
for each target pathological subgraph to be processed, inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein a target information region is segmented out;
and combining the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed.
Optionally, the target features of the target to-be-processed pathological subgraph are: in the classifier, the classifier is a convolution result of an nth convolutional layer of the first N convolutional layers, which are the same as a preset segmentation model, and the segmentation model is the preset segmentation model with the first N convolutional layers removed, where N is a natural number greater than zero.
Optionally, before the inputting each of the pathology subgraphs to be processed into a predetermined classifier for analysis, the method further includes:
and adjusting the color of each pathological subgraph to be processed into a picture consistent with the color of the standard picture.
Optionally, the inputting each to-be-processed pathological sub-graph into a predetermined classifier for analysis, and determining a target to-be-processed pathological sub-graph containing target information in each to-be-processed pathological sub-graph includes:
aiming at each pathological subgraph to be processed, judging whether the pathological subgraph to be processed contains target information or not through the classifier;
and if the pathological subgraph to be processed contains the target information, determining the pathological subgraph to be processed as a target pathological subgraph to be processed.
Optionally, the inputting the target feature of the target pathological sub-image to be processed into a predetermined segmentation model to obtain a target image corresponding to the target pathological sub-image to be processed, where the target image is a region of the target information, includes:
aiming at the target characteristics of the target pathological subgraph to be processed, carrying out first convolution operation on the target pathological subgraph to be processed through a plurality of convolution layers of the segmentation model to obtain a first result;
performing cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result;
extracting a convolution result of a designated convolution layer in the segmentation model, and performing second convolution operation on the convolution result to obtain a third result;
and summing and performing a third convolution operation on the second result and the third result to obtain a target picture corresponding to the target to-be-processed pathological subgraph divided into a target information area.
In a second aspect, an embodiment of the present invention discloses a device for determining a critical area of a pathological diagram, where the device includes:
the to-be-processed pathological image acquisition module is used for acquiring a to-be-processed pathological image;
the to-be-processed pathological graph dividing module is used for dividing the to-be-processed pathological graph into a plurality of to-be-processed pathological subgraphs;
the target pathological subgraph to be processed determining module is used for inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining target pathological subgraphs to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed;
the target picture determining module is used for inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model aiming at each target pathological subgraph to be processed to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein the target picture is divided into a target information area;
and the key region determining module is used for combining all the target pictures to obtain the key region of the pathological graph aiming at the pathological graph to be processed.
Optionally, the apparatus further comprises:
and the color adjusting module of the pathological subgraph to be processed is used for adjusting the color of each pathological subgraph to be processed into a picture with the color consistent with that of the standard picture.
Optionally, the target to-be-processed pathology subgraph determination module includes:
the target information judgment sub-module is used for judging whether each pathological subgraph to be processed contains target information or not through the classifier aiming at each pathological subgraph to be processed;
and the target pathological subgraph to be processed determining submodule is used for determining the pathological subgraph to be processed as the target pathological subgraph to be processed if the target information is contained in the pathological subgraph to be processed.
Optionally, the target picture determining module includes:
the first result determining submodule is used for carrying out first convolution operation on the target pathology sub-image to be processed through the plurality of convolution layers of the segmentation model aiming at the target feature of the target pathology sub-image to be processed to obtain a first result;
the second result determining submodule is used for carrying out cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result;
the third result determining submodule is used for extracting a convolution result of a designated convolution layer in the segmentation model and carrying out second convolution operation on the convolution result to obtain a third result;
and the target picture determining submodule is used for summing the second result and the third result and performing a third convolution operation to obtain a target picture corresponding to the target to-be-processed pathological subgraph divided into a target information area.
In a third aspect, an embodiment of the present invention discloses an electronic device, including a processor, a communication interface, a memory and a communication bus, where the processor, the communication interface, and the memory complete mutual communication through the communication bus;
a memory for storing a computer program;
and the processor is used for realizing the method steps of any one of the pathological diagram key area determination methods when executing the program stored in the memory.
In another aspect, an embodiment of the present invention discloses a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the method steps of any one of the above methods for determining a critical area of a pathological diagram are implemented.
In yet another aspect, an embodiment of the present invention discloses a computer program product containing instructions for implementing the method steps of any one of the above methods for determining critical areas of a pathological diagram when the computer program product runs on a computer.
In the method, the device, the electronic device and the storage medium for determining the key region of the pathological graph provided by the embodiment of the invention, the pathological graph to be processed is divided into a plurality of pathological subgraphs to be processed in view of the situations that the medical image has high resolution, large size and the like. And extracting the characteristics of each pathological subgraph to be processed to obtain the multi-scale characteristics of the pathological subgraph to be processed, and obtaining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and the target characteristics of the target pathological subgraph to be processed. Dividing each target pathological subgraph to be processed through a division model to obtain a target picture of a divided target information area; and finally, combining all the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. According to the embodiment of the invention, all the pathological subgraphs to be processed after the pathological graph to be processed is divided are analyzed, so that the loss of target information in the pathological graph is reduced, and the key region information of all the target information of the pathological graph is obtained. In addition, in the embodiment of the invention, the target characteristics of the target pathological subgraph to be processed are determined through the classifier, and the target characteristics of the target pathological subgraph to be processed are input into the segmentation model, so that the segmentation model is prevented from carrying out the same repeated processing as that of the segmenter on the unprocessed target pathological subgraph to be processed. And only inputting the target features of the target pathological subgraph to be processed containing the target information into the segmentation model for processing, and filtering pictures which do not need to be calculated. Therefore, the embodiment of the invention also saves the calculation amount and improves the efficiency of determining the key area of the pathological diagram.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flowchart of a method for determining a critical area of a pathological diagram according to an embodiment of the present invention;
fig. 2 is a flowchart of a method for determining a target pathological subgraph to be processed in a method for determining a critical area of a pathological graph according to an embodiment of the present invention;
fig. 3 is a flowchart of a target picture determination method in a method for determining a critical area of a pathological diagram according to an embodiment of the present invention;
FIG. 4 is a diagram of a pathological diagram critical area determination system according to an embodiment of the present invention;
FIG. 5 is a schematic structural diagram of a classifier in a system for determining a critical area of a pathological diagram according to an embodiment of the present invention;
FIG. 6 is a schematic structural diagram of a segmentation model in a pathological diagram key region determination system according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a device for determining a critical area of a pathological diagram according to an embodiment of the present invention;
fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The number of cancer cases has increased over the past decades. Studies have shown that the average incidence of cancer is still on a growing trend.
Early diagnosis and accurate diagnosis are necessary for curing cancer. In early diagnosis, Fine Needle Aspiration (FNA) biopsy is one of the most successful medical tests. FNA takes a small biopsy from a patient's tumor using a fine needle and subjects the sample to a series of procedures to make a cytopathological section. The doctor observes the characteristics of key areas in the whole section under a microscope to make corresponding diagnosis. With the increase of pathological samples, doctors need to spend much time and great effort in the diagnosis process to ensure the accuracy of diagnosis due to the characteristics of cells in the cytopathological section.
Therefore, it is very important to rapidly extract key regions in the study of the computer diagnosis technology based on the cytopathology image. The existing computer diagnosis technology for cell pathology comprises two aspects of pathology image digitization technology and pathology image processing and automatic diagnosis technology. The pathological section is digitized by the pathological image digitization technology, that is, the Whole tissue section can be scanned by a digital section scanner to generate a WSI (white slide image, a full-section digital image). The cytopathology section is scanned using higher magnification to ensure the integrity of critical regional features. The pathological image digitization technology can not only reduce the workload of pathologists and reduce the required time, but also laterally improve the accuracy of pathological diagnosis in pathological analysis.
In the prior art, the WSI is processed by a conventional convolutional neural network, and the method comprises the following steps: and directly scaling the WSI picture to a proper size, and further researching key information contained in the scaled WSI picture.
However, the inventor finds that in a key information research method of scaling the WSI picture to a proper size, target information locally contained in a pathological diagram is easily lost. Therefore, how to reduce the loss of target information in the pathological graph and obtain all the pathological graph key areas of the target information is still an urgent technical problem to be solved.
In order to solve the technical problem, an embodiment of the invention discloses a method and a device for determining a pathological diagram key region, an electronic device and a storage medium, so as to reduce the loss of target information in a pathological diagram and obtain all pathological diagram key regions of the target information. The specific method comprises the following steps:
in a first aspect, an embodiment of the present invention discloses a method for determining a critical area of a pathological diagram, as shown in fig. 1. Fig. 1 is a flowchart of a method for determining a critical area of a pathological diagram according to an embodiment of the present invention, where the method includes:
and S101, acquiring a pathological diagram to be processed.
The pathology to be processed may be the WSI of a given study, for example: WSI of thyroid-punctured cells.
And S102, dividing the pathology diagram to be processed into a plurality of pathology subgraphs to be processed.
Although the traditional workflow is transferred to a computer by the WSI scanning technology, the huge picture size of the WSI scanning technology makes it difficult for a doctor to find a lesion area. Doctors often spend a lot of time looking for critical areas and the whole process requires watching a computer monitor at all times, which is prone to fatigue and reduces the efficiency of diagnosis. Also, WSIs typically have a very large amount of data. Therefore, in the embodiment of the invention, the WSI of the pathological image to be processed is cut to obtain a plurality of image blocks of the pathological images to be processed.
S103, inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed.
In the embodiment of the invention, in order to better determine the key area of the target information in the pathological diagram to be processed, a classifier capable of identifying the target information can be determined in advance. And judging whether the input pathological subgraph to be processed contains target information or not by using the classifier, and discarding other pathological subgraphs to be processed which do not contain the target information.
The classifier is a neural network model which can accurately identify target information and is obtained by training the neural network model. The target information is the key information in the pathological diagram to be processed which needs to be identified. For example, if the pathology to be processed is WSI of thyroid-penetrating cells, the target information may be follicular cells. Follicular cells are sparsely distributed in pathological sections of thyroid cells. In the tens of thousands of panels into which a WSI is divided, only less than 10% of the panels contain follicular cells. Therefore, most pictures which do not contain follicular cells can be filtered by adopting the classifier, and the task amount of the segmentation model is reduced.
Optionally, the target features of the target to-be-processed pathological subgraph are as follows: in the classifier, the segmentation model is a preset segmentation model with the first N convolutional layers removed, and N is a natural number greater than zero.
In the embodiment of the invention, in order to better determine the key region of the target information in the pathological graph to be processed, the target pathological subgraph to be processed containing the target information determined by the classifier can be input into the segmentation model. And dividing each target information area containing the target pathological subgraph to be processed by convolution calculation of the segmentation model. The segmentation model is also used for training the neural network model, the obtained identifiable picture contains a model of a target information area, namely, the identification of the target information in the pathological subgraph to be processed is obtained through convolution calculation, the segmentation of the area containing the target information in the pathological subgraph to be processed is also obtained through convolution calculation, and the segmentation model and the area have the same shallow convolution operation in the process of processing the pathological subgraph to be processed. In order to reduce the amount of calculation, the convolutional layers having the same number of layers as the classification recognition and segmentation recognition processes are determined in advance, and the convolution operation does not need to be repeated in the later segmentation recognition process. Namely, the segmentation model in the embodiment of the invention is as follows: and in the preset divider, removing the same first N convolutional layers in the classification identification and division identification processes to obtain a preset division model.
In the embodiment of the invention, through convolution operation analysis of a classifier, a target pathology subgraph to be processed containing target information in the input pathology subgraph to be processed can be determined, and through the classifier, a target feature is output, wherein the target feature is a convolution result of an Nth convolution layer of the first N convolution layers which are the same as a preset segmentation model in the classifier.
Optionally, before each pathological sub-image to be processed is input into a predetermined classifier for analysis in S103, in order to reduce the difference in staining of WSI and facilitate effective recognition by the classifier, the image containing the target information may be used as a standard image, and a color adjustment method is used to adjust the color of each pathological sub-image to be processed to be consistent with the color of the standard image.
Optionally, in step S103, each pathological sub-graph to be processed is input into a predetermined classifier for analysis, and a target pathological sub-graph to be processed, which includes the target information, in each pathological sub-graph to be processed is determined, as shown in fig. 2. Fig. 2 is a flowchart of a method for determining a target to-be-processed pathological sub-graph in a method for determining a critical region of a pathological graph according to an embodiment of the present invention, where the method includes:
and S1031, aiming at each pathological subgraph to be processed, judging whether the pathological subgraph to be processed contains target information or not through the classifier.
For example, the pathological pattern to be processed in the embodiment of the invention is WSI of thyroid gland puncture cells, and the target information is follicular cells. In the step, the sub WSIs of the thyroid gland puncture cells are input into a classifier, and whether follicular cells exist in the sub WSIs of the thyroid gland puncture cells or not is judged through the classifier.
And S1032, if the pathology subgraph to be processed contains the target information, determining the pathology subgraph to be processed as the target pathology subgraph to be processed.
For example, in step S1032, if the classifier determines that the WSI of the thyroid gland puncturing cell contains follicular cells, the sub-WSI of the thyroid gland puncturing cell is determined as the target pathological subgraph to be processed.
In addition, the classifier discards the pathology subgraph to be processed which does not contain the target information, namely the classifier directly discards the pathology subgraph to be processed which does not contain the target information after inputting the pathology subgraph to be processed, and does not input the pathology subgraph to be processed into the segmentation model. And further effectively reduces the task processing amount of the segmentation model.
For example, the pathological pattern to be processed in the embodiment of the invention is WSI of thyroid gland puncture cells, and the target information is follicular cells. The classifier can directly discard the interfering only glial information's WSI of thyroid-penetrating cells, as well as the uninformative WSI.
And S104, inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model aiming at each target pathological subgraph to be processed to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein the target information region is segmented out.
In the embodiment of the invention, the convolution operation is carried out on the target characteristics of the input target pathological subgraph to be processed through the segmentation model to obtain the target picture corresponding to the target pathological subgraph to be processed and divided into the target information area
Optionally, in S104, the target feature of the target pathological sub-image to be processed is input into a predetermined segmentation model, so as to obtain a target image corresponding to the target pathological sub-image to be processed, which is segmented into a target information region, as shown in fig. 3. Fig. 3 is a flowchart of a method for determining a target picture in a method for determining a critical area of a pathological diagram according to an embodiment of the present invention, where the method includes:
s1041, aiming at the target feature of the target pathological subgraph to be processed, performing a first convolution operation on the target pathological subgraph to be processed through a plurality of convolution layers of a segmentation model to obtain a first result;
s1042, performing cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result;
in this step, first, a first convolution operation is performed on the target pathological subgraph to be processed by segmenting a plurality of convolution layers of the model to obtain a first result.
S1043, extracting a convolution result of the designated convolution layer in the segmentation model, and performing a second convolution operation on the convolution result to obtain a third result;
the segmentation model of the embodiment of the invention provides an algorithm for strengthening the pooling of the cavity convolution pyramid, introduces lower-scale features into the pooling of the cavity convolution pyramid, and improves the influence of the low-scale features on the segmentation result so as to improve the accuracy of the segmentation result.
Specifically, it can be seen that any one of the middle convolution layers in the segmentation model serves as a designated convolution layer. And in the process of processing the target pathology subgraph to be processed by the segmentation model, extracting the convolution result of the specified convolution layer, and further performing one-time specified convolution operation on the convolution result through the convolution layer with the convolution kernel of 256 1 × 1 and the step length of 1 to obtain a third result after convolution.
And S1044, summing the second result and the third result and performing a third convolution operation to obtain a target picture corresponding to the target to-be-processed pathological subgraph in the divided target information area.
And S105, combining the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed.
And splicing the target pictures according to the target information areas to obtain the pathological diagram key area corresponding to the complete key information of the pathological diagram to be processed.
In the method for determining the key region of the pathological graph provided by the embodiment of the invention, the pathological graph to be processed is divided into a plurality of pathological subgraphs to be processed in view of the conditions that the medical image has high resolution, large size and the like. And extracting the characteristics of each pathological subgraph to be processed to obtain the multi-scale characteristics of the pathological subgraph to be processed, and obtaining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and the target characteristics of the target pathological subgraph to be processed. Dividing each target pathological subgraph to be processed through a division model to obtain a target picture of a divided target information area; and finally, combining all the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. According to the embodiment of the invention, all the pathological subgraphs to be processed after the pathological graph to be processed is divided are analyzed, so that the loss of target information in the pathological graph is reduced, and the key region information of all the target information of the pathological graph is obtained. In addition, in the embodiment of the invention, the target characteristics of the target pathological subgraph to be processed are determined through the classifier, and the target characteristics of the target pathological subgraph to be processed are input into the segmentation model, so that the segmentation model is prevented from carrying out the same repeated processing as that of the segmenter on the unprocessed target pathological subgraph to be processed. And only inputting the target features of the target pathological subgraph to be processed containing the target information into the segmentation model for processing, and filtering pictures which do not need to be calculated. Therefore, the embodiment of the invention also saves the calculation amount and improves the efficiency of determining the key area of the pathological diagram.
To better illustrate the method for determining the key region of the pathological diagram according to the embodiment of the present invention, there may be a system architecture diagram for determining the key region of the pathological diagram according to the embodiment of the present invention shown in fig. 4.
The pathological diagram key region determining system comprises a classifier and a segmentation model. The classifier and the preset segmentation model have the same first N convolutional layers, in order to reduce the calculation amount, the first N convolutional layers can be used as a shared structure, and the segmentation model of the embodiment of the invention is the preset segmentation model with the preset segmentation model removing the N convolutional layers. That is, the first N convolutional layers of the classifier perform convolution operation on the pathology subgraph to be processed to obtain target characteristics, and the target characteristics are directly input into the segmentation model.
By the pathological diagram key area determining system, the following method can be realized:
firstly, preprocessing data of each pathology subgraph to be processed, which is divided from a pathology graph to be processed;
the method specifically comprises the following steps: the picture containing the target information can be used as a standard picture, and the color of each pathological subgraph to be processed is adjusted to be consistent with the color of the standard picture by using a color adjusting method.
And step two, inputting the preprocessed result into a classifier for classification, wherein the classifier can perform feature analysis on the input pathological subgraph to be processed, judge whether the pathological subgraph to be processed contains target information, use the pathological subgraph to be processed containing the target information as a target pathological subgraph to be processed, and discard each pathological subgraph to be processed which does not contain the target information.
The structure of the classifier in the pathological diagram key region determination system can be shown in fig. 5.
The classifier includes a plurality of convolutional layers, which correspond to block 1, block 2 through block 4 in fig. 5. Block 1 therein is a shared structure.
Take Deeplab V3 as an example to illustrate a specific sharing method. The basic model is ResNet 101, and the structure is shown in FIG. 3 (a). The shared structure is zone 1 in ResNet 101. Fig. 2 (c) shows the structure of the classifier except for the shared structure. And adding a convolution layer with the convolution kernel size of 3 multiplied by 3 to further extract the characteristics of Block 1. The number of the nodes of the two full connection layers is 4096 and 3 respectively, wherein the number of the nodes of the second full connection layer is consistent with the number of the categories to be distinguished. And further analyzing and extracting the image characteristics through a convolution layer and two full-connection layers to obtain a classification result, for example, if the pathological image to be processed is thyroid cell WSI, classifying all the images into three types on the whole, namely, an image containing follicular information, a colloid information image and an information-free image, wherein the image containing follicular information is a target pathological subgraph to be processed.
After passing through the classifier, each pathological subgraph to be processed can be labeled, the target pathological subgraph to be processed is sent to a subsequent segmentation model, and other pictures are discarded.
And thirdly, inputting the target characteristics of the pathological subgraph to be processed into the segmentation model aiming at each target pathological subgraph to be processed to obtain a target picture corresponding to the pathological subgraph to be processed, wherein the target information region is segmented out.
The structure of the segmentation model in the pathology diagram key region determination system can be shown in fig. 6.
The segmentation model includes ASPP (empty convolutional Pyramid Pooling) and global average Pooling. Due to the characteristics of the cytopathology image, the volume of target information is small and the target information is sparsely distributed, so that the segmentation accuracy is improved, more accurate characteristics with a lower scale are added into ASPP in the existing ASPP processing, and the E-ASPP is formed.
As can be seen from fig. 6, for each target pathology subgraph to be processed, a first convolution operation is performed on the target pathology subgraph to be processed by segmenting convolution layers of a plurality of blocks in the model, so as to obtain a first result; performing cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result; extracting the convolution result of the convolution layer of the block 3 in the segmentation model, and performing specified convolution operation on the convolution result to obtain a third result; and summing the second result and the third result and performing a third convolution operation to obtain a target picture corresponding to the target to-be-processed pathological subgraph divided into a target information area.
For example, the pathology map to be processed in the embodiment of the present invention is a thyroid cell pathology WSI, and the follicular cell region included in each target thyroid cell pathology map can be obtained after the processing by the classifier and the segmentation model in the pathology map key region determination system. And finally, splicing the target thyroid cell pathology subgraphs according to the positions of follicular cell areas to obtain all follicular cell areas in the thyroid cell pathology WSI.
In the embodiment of the invention, the influence of the low-scale features on the segmentation result is improved and the accuracy of the segmentation of the target information in the pathological image to be processed is improved by adding the features with lower scale in the existing ASPP processing in the segmentation model.
The method for determining the key region of the pathological graph extracts the multi-scale characteristic information of the image through CNN based on the powerful extraction capability of the convolutional neural network to the image characteristics in deep learning, and fuses the characteristic information, so that the key region containing target information is extracted from the cell pathological section WSI. Aiming at the characteristics of the cytopathology slice, in order to obtain a more accurate segmentation result, a low-scale feature is added into the original segmentation model ASPP. The invention finally realizes the reduction of the loss of the target information in the pathological diagram and obtains the key region information of all the target information of the pathological diagram.
In addition, in order to reduce the calculation time of the whole WSI, a classifier and a segmentation model are used in the embodiment of the invention, and pictures which do not need to be calculated are filtered by the classifier. And the idea of sharing the structure is provided, so that the segmentation model does not repeatedly process the bottom layer convolution operation which is the same as that of the classifier, the calculation amount is integrally saved, and the segmentation efficiency is improved. The doctor can further diagnose by observing the extracted key area, so that the time of the doctor is saved.
In a second aspect, an embodiment of the present invention discloses a device for determining a critical area of a pathological diagram, as shown in fig. 7. Fig. 7 is a schematic structural diagram of a device for determining a critical area of a pathological diagram according to an embodiment of the present invention, where the device includes:
a to-be-processed pathological diagram obtaining module 701, configured to obtain a to-be-processed pathological diagram;
a to-be-processed pathological graph partitioning module 702, configured to partition a to-be-processed pathological graph into a plurality of to-be-processed pathological subgraphs;
a target to-be-processed pathological subgraph determining module 703, configured to input each to-be-processed pathological subgraph into a predetermined classifier for analysis, and determine a target to-be-processed pathological subgraph including target information in each to-be-processed pathological subgraph and a target feature of the target to-be-processed pathological subgraph;
a target picture determining module 704, configured to, for each target pathological sub-image to be processed, input the target feature of the target pathological sub-image to be processed into a predetermined segmentation model, so as to obtain a target picture corresponding to the target pathological sub-image to be processed, where the target information region is segmented;
the key region determining module 705 is configured to combine the target pictures to obtain a pathological diagram key region for the pathological diagram to be processed.
In the device for determining the key area of the pathological diagram, which is provided by the embodiment of the invention, the pathological diagram to be processed is divided into a plurality of pathological subgraphs to be processed in view of the situations that the medical image has high resolution, large size and the like. And extracting the characteristics of each pathological subgraph to be processed to obtain the multi-scale characteristics of the pathological subgraph to be processed, and obtaining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and the target characteristics of the target pathological subgraph to be processed. Dividing each target pathological subgraph to be processed through a division model to obtain a target picture of a divided target information area; and finally, combining all the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. According to the embodiment of the invention, all the pathological subgraphs to be processed after the pathological graph to be processed is divided are analyzed, so that the loss of target information in the pathological graph is reduced, and the key region information of all the target information of the pathological graph is obtained. In addition, in the embodiment of the invention, the target characteristics of the target pathological subgraph to be processed are determined through the classifier, and the target characteristics of the target pathological subgraph to be processed are input into the segmentation model, so that the segmentation model is prevented from carrying out the same repeated processing as that of the segmenter on the unprocessed target pathological subgraph to be processed. And only inputting the target features of the target pathological subgraph to be processed containing the target information into the segmentation model for processing, and filtering pictures which do not need to be calculated. Therefore, the embodiment of the invention also saves the calculation amount and improves the efficiency of determining the key area of the pathological diagram.
Optionally, in an embodiment of the apparatus for determining a critical area of a pathological diagram of the present invention, the apparatus further includes:
and the color adjusting module of the pathological subgraph to be processed is used for adjusting the color of each pathological subgraph to be processed into a picture with the same color as the standard picture.
Optionally, in an embodiment of the apparatus for determining a critical region of a pathological diagram, the module 703 for determining a target pathological subgraph includes:
the target information judgment sub-module is used for judging whether each pathological subgraph to be processed contains target information or not through the classifier aiming at each pathological subgraph to be processed;
and the target pathological subgraph to be processed determining submodule is used for determining the pathological subgraph to be processed as the target pathological subgraph to be processed if the pathological subgraph to be processed contains the target information.
Optionally, in an embodiment of the apparatus for determining a critical area of a pathological diagram of the present invention, the target image determining module 704 includes:
the first result determining submodule is used for carrying out first convolution operation on the target pathological sub-image to be processed through a plurality of convolution layers of the segmentation model aiming at the target characteristic of the target pathological sub-image to be processed to obtain a first result;
the second result determining submodule is used for performing cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result;
the third result determining submodule is used for extracting the convolution result of the designated convolution layer in the segmentation model and carrying out second convolution operation on the convolution result to obtain a third result;
and the target picture determining submodule is used for summing the second result and the third result and performing a third convolution operation to obtain a target picture corresponding to the target to-be-processed pathological subgraph divided into a target information area.
In a third aspect, an embodiment of the present invention discloses an electronic device, as shown in fig. 8. Fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, which includes a processor 801, a communication interface 802, a memory 803, and a communication bus 804, where the processor 801, the communication interface 802, and the memory 803 complete communication with each other through the communication bus 804;
a memory 803 for storing a computer program;
the processor 801 is configured to implement the following method steps when executing the program stored in the memory 803:
acquiring a pathological diagram to be processed;
dividing the pathological graph to be processed into a plurality of pathological subgraphs to be processed;
inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed;
for each target pathological subgraph to be processed, inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein the target information region is segmented out;
and combining the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed.
The communication bus 804 mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus 804 may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.
The communication interface 802 is used for communication between the above-described electronic apparatus and other apparatuses.
The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory 803 may also be at least one storage device located remotely from the processor 801.
The Processor 801 may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but also Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components.
In the electronic device provided by the embodiment of the invention, in view of the situations of high resolution, large size and the like of the medical image, the pathology diagram to be processed is divided into a plurality of pathology subgraphs to be processed. And extracting the characteristics of each pathological subgraph to be processed to obtain the multi-scale characteristics of the pathological subgraph to be processed, and obtaining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and the target characteristics of the target pathological subgraph to be processed. Dividing each target pathological subgraph to be processed through a division model to obtain a target picture of a divided target information area; and finally, combining all the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. According to the embodiment of the invention, all the pathological subgraphs to be processed after the pathological graph to be processed is divided are analyzed, so that the loss of target information in the pathological graph is reduced, and the key region information of all the target information of the pathological graph is obtained. In addition, in the embodiment of the invention, the target characteristics of the target pathological subgraph to be processed are determined through the classifier, and the target characteristics of the target pathological subgraph to be processed are input into the segmentation model, so that the segmentation model is prevented from carrying out the same repeated processing as that of the segmenter on the unprocessed target pathological subgraph to be processed. And only inputting the target features of the target pathological subgraph to be processed containing the target information into the segmentation model for processing, and filtering pictures which do not need to be calculated. Therefore, the embodiment of the invention also saves the calculation amount and improves the efficiency of determining the key area of the pathological diagram.
In another aspect, an embodiment of the present invention discloses a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the method steps of any one of the above methods for determining a critical area of a pathological diagram are implemented.
In a computer-readable storage medium provided by an embodiment of the present invention, in view of the situations of high resolution, large size, and the like of a medical image, a pathology map to be processed is segmented into a plurality of pathology subgraphs to be processed. And extracting the characteristics of each pathological subgraph to be processed to obtain the multi-scale characteristics of the pathological subgraph to be processed, and obtaining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and the target characteristics of the target pathological subgraph to be processed. Dividing each target pathological subgraph to be processed through a division model to obtain a target picture of a divided target information area; and finally, combining all the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. According to the embodiment of the invention, all the pathological subgraphs to be processed after the pathological graph to be processed is divided are analyzed, so that the loss of target information in the pathological graph is reduced, and the key region information of all the target information of the pathological graph is obtained. In addition, in the embodiment of the invention, the target characteristics of the target pathological subgraph to be processed are determined through the classifier, and the target characteristics of the target pathological subgraph to be processed are input into the segmentation model, so that the segmentation model is prevented from carrying out the same repeated processing as that of the segmenter on the unprocessed target pathological subgraph to be processed. And only inputting the target features of the target pathological subgraph to be processed containing the target information into the segmentation model for processing, and filtering pictures which do not need to be calculated. Therefore, the embodiment of the invention also saves the calculation amount and improves the efficiency of determining the key area of the pathological diagram.
In yet another aspect, an embodiment of the present invention discloses a computer program product containing instructions for implementing the method steps of any one of the above methods for determining critical areas of a pathological diagram when the computer program product runs on a computer.
In a computer program product including instructions provided by an embodiment of the present invention, in view of the situations of high resolution, large size, and the like of a medical image, a pathology map to be processed is segmented into a plurality of pathology subgraphs to be processed. And extracting the characteristics of each pathological subgraph to be processed to obtain the multi-scale characteristics of the pathological subgraph to be processed, and obtaining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and the target characteristics of the target pathological subgraph to be processed. Dividing each target pathological subgraph to be processed through a division model to obtain a target picture of a divided target information area; and finally, combining all the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed. According to the embodiment of the invention, all the pathological subgraphs to be processed after the pathological graph to be processed is divided are analyzed, so that the loss of target information in the pathological graph is reduced, and the key region information of all the target information of the pathological graph is obtained. In addition, in the embodiment of the invention, the target characteristics of the target pathological subgraph to be processed are determined through the classifier, and the target characteristics of the target pathological subgraph to be processed are input into the segmentation model, so that the segmentation model is prevented from carrying out the same repeated processing as that of the segmenter on the unprocessed target pathological subgraph to be processed. And only inputting the target features of the target pathological subgraph to be processed containing the target information into the segmentation model for processing, and filtering pictures which do not need to be calculated. Therefore, the embodiment of the invention also saves the calculation amount and improves the efficiency of determining the key area of the pathological diagram.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the device, the electronic apparatus and the storage medium embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiments.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (9)
1. A method for determining key areas of a pathological diagram is characterized by comprising the following steps:
acquiring a pathological diagram to be processed;
dividing the pathology diagram to be processed into a plurality of pathology subgraphs to be processed;
inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining a target pathological subgraph to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed;
for each target pathological subgraph to be processed, inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein a target information region is segmented out;
combining the target pictures to obtain a pathological diagram key area aiming at the pathological diagram to be processed;
inputting the target feature of the target pathological subgraph to be processed into a predetermined segmentation model to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein the step of dividing the target information region comprises the following steps:
aiming at the target characteristics of the target pathological subgraph to be processed, carrying out first convolution operation on the target pathological subgraph to be processed through a plurality of convolution layers of the segmentation model to obtain a first result;
performing cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result;
extracting a convolution result of a designated convolution layer in the segmentation model, and performing second convolution operation on the convolution result to obtain a third result;
and summing and performing a third convolution operation on the second result and the third result to obtain a target picture corresponding to the target to-be-processed pathological subgraph divided into a target information area.
2. The method according to claim 1, wherein the target features of the target to-be-processed pathological sub-graph are: in the classifier, the classifier is a convolution result of an nth convolutional layer of the first N convolutional layers which are the same as a preset segmentation model, the segmentation model is the preset segmentation model with the first N convolutional layers removed, wherein N is a natural number greater than zero.
3. The method according to claim 1, wherein before said inputting each of said pathology sub-graphs to be processed into a predetermined classifier for analysis, said method further comprises:
and adjusting the color of each pathological subgraph to be processed into a picture consistent with the color of the standard picture.
4. The method according to claim 1, wherein the inputting each pathological sub-graph to be processed into a predetermined classifier for analysis, and determining a target pathological sub-graph to be processed containing target information in each pathological sub-graph to be processed comprises:
aiming at each pathological subgraph to be processed, judging whether the pathological subgraph to be processed contains target information or not through the classifier;
and if the pathological subgraph to be processed contains the target information, determining the pathological subgraph to be processed as a target pathological subgraph to be processed.
5. An apparatus for determining a critical region of a pathological map, the apparatus comprising:
the to-be-processed pathological image acquisition module is used for acquiring a to-be-processed pathological image;
the to-be-processed pathological graph dividing module is used for dividing the to-be-processed pathological graph into a plurality of to-be-processed pathological subgraphs;
the target pathological subgraph to be processed determining module is used for inputting each pathological subgraph to be processed into a predetermined classifier for analysis, and determining target pathological subgraphs to be processed containing target information in each pathological subgraph to be processed and target characteristics of the target pathological subgraph to be processed;
the target picture determining module is used for inputting the target characteristics of the target pathological subgraph to be processed into a predetermined segmentation model aiming at each target pathological subgraph to be processed to obtain a target picture corresponding to the target pathological subgraph to be processed, wherein the target picture is divided into a target information area;
a key region determining module for combining each target picture to obtain a pathological diagram key region aiming at the pathological diagram to be processed,
wherein, the target picture determining module comprises:
the first result determining submodule is used for carrying out first convolution operation on the target pathological sub-image to be processed through a plurality of convolution layers of the segmentation model aiming at the target characteristic of the target pathological sub-image to be processed to obtain a first result;
the second result determining submodule is used for carrying out cavity convolution pyramid pooling and global average pooling on the first result to obtain a second result;
the third result determining submodule is used for extracting a convolution result of a designated convolution layer in the segmentation model and carrying out second convolution operation on the convolution result to obtain a third result;
and the target picture determining submodule is used for summing the second result and the third result and performing a third convolution operation to obtain a target picture corresponding to the target to-be-processed pathological subgraph divided into a target information area.
6. The apparatus of claim 5, further comprising:
and the color adjusting module of the pathological subgraph to be processed is used for adjusting the color of each pathological subgraph to be processed into a picture with the color consistent with that of the standard picture.
7. The apparatus of claim 5, wherein the target to-be-processed pathology subgraph determination module comprises:
the target information judgment sub-module is used for judging whether each pathological subgraph to be processed contains target information or not through the classifier aiming at each pathological subgraph to be processed;
and the target pathological subgraph to be processed determining submodule is used for determining the pathological subgraph to be processed as the target pathological subgraph to be processed if the target information is contained in the pathological subgraph to be processed.
8. An electronic device is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing mutual communication by the memory through the communication bus;
a memory for storing a computer program;
a processor for implementing the method steps of any of claims 1 to 4 when executing a program stored in the memory.
9. A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, which computer program, when being executed by a processor, carries out the method steps of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910138631.2A CN109859218B (en) | 2019-02-25 | 2019-02-25 | Pathological graph key area determination method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910138631.2A CN109859218B (en) | 2019-02-25 | 2019-02-25 | Pathological graph key area determination method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109859218A CN109859218A (en) | 2019-06-07 |
| CN109859218B true CN109859218B (en) | 2020-12-04 |
Family
ID=66898930
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910138631.2A Active CN109859218B (en) | 2019-02-25 | 2019-02-25 | Pathological graph key area determination method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109859218B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110619639A (en) * | 2019-08-26 | 2019-12-27 | 苏州同调医学科技有限公司 | Method for segmenting radiotherapy image by combining deep neural network and probability map model |
| CN111882521A (en) * | 2020-06-21 | 2020-11-03 | 浙江大学山东工业技术研究院 | Image processing method of cell smear |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102007029886B4 (en) * | 2007-06-28 | 2016-10-06 | Siemens Healthcare Gmbh | A method of segmenting a myocardial wall and a device for detecting a pathologically altered coronary artery |
| US9031294B2 (en) * | 2010-07-23 | 2015-05-12 | National Institute Of Advanced Industrial Science And Technology | Region segmented image data creating system and feature extracting system for histopathological images |
| CN102663728B (en) * | 2012-03-11 | 2015-04-08 | 西安电子科技大学 | Dictionary learning-based medical image interactive joint segmentation |
| US10453195B2 (en) * | 2016-08-19 | 2019-10-22 | Optrascan, Inc. | Method of detecting tissue area of interest in digital pathology imaging by executing computer-executable instructions stored on a non-transitory computer-readable medium |
| CN107247949B (en) * | 2017-08-02 | 2020-06-19 | 智慧眼科技股份有限公司 | Face recognition method and device based on deep learning and electronic equipment |
| CN107945200B (en) * | 2017-12-14 | 2021-08-03 | 中南大学 | Image binarization segmentation method |
| CN108229344A (en) * | 2017-12-19 | 2018-06-29 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment, computer program and storage medium |
| CN108765386A (en) * | 2018-05-16 | 2018-11-06 | 中铁科学技术开发公司 | A kind of tunnel slot detection method, device, electronic equipment and storage medium |
| CN109040588A (en) * | 2018-08-16 | 2018-12-18 | Oppo广东移动通信有限公司 | Photographic method, device, storage medium and the terminal of facial image |
-
2019
- 2019-02-25 CN CN201910138631.2A patent/CN109859218B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109859218A (en) | 2019-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7146886B2 (en) | Systems and methods for comprehensive multi-assay tissue analysis | |
| WO2021139258A1 (en) | Image recognition based cell recognition and counting method and apparatus, and computer device | |
| US8600143B1 (en) | Method and system for hierarchical tissue analysis and classification | |
| US9684960B2 (en) | Automated histological diagnosis of bacterial infection using image analysis | |
| CN110245657B (en) | Pathological image similarity detection method and detection device | |
| JP2021166062A (en) | Focal point weighting machine learning classifier error prediction for microscope slide image | |
| US10861156B2 (en) | Quality control for digital pathology slides | |
| CN108682008B (en) | A kind of Leukocyte Image clarity evaluation method and device | |
| CN114972548A (en) | Image processing system and method for displaying multiple images of a biological specimen | |
| WO2021189771A1 (en) | Slide digitization information quality testing method and apparatus, and device and medium | |
| US12008750B2 (en) | Generating annotation data of tissue images | |
| CN112750121B (en) | System and method for detecting digital image quality of pathological slide | |
| CN112464802B (en) | Automatic identification method and device for slide sample information and computer equipment | |
| CN112215790A (en) | KI67 index analysis method based on deep learning | |
| CN109859218B (en) | Pathological graph key area determination method and device, electronic equipment and storage medium | |
| US20220318979A1 (en) | Image Quality Analysis for Artifact Detection in Pathology Slide Images | |
| US12020477B2 (en) | Systems and methods for generating encoded representations for multiple magnifications of image data | |
| CN113393454A (en) | Method and device for segmenting pathological target examples in biopsy tissues | |
| CN113066080A (en) | Method and device for identifying slice tissue, cell identification model and tissue segmentation model | |
| CN109978849B (en) | Baseline cell determination method, device and readable medium based on digital pathology image | |
| Jiménez et al. | Image quality metrics applied to digital pathology | |
| CN114170415A (en) | TMB classification method and system based on histopathology image depth domain adaptation | |
| Paulik et al. | Tissue Block and Digital Slide Registration Method for Microscopic Samples | |
| Modi et al. | Multi-Stain Multi-Level Convolutional Network for Multi-Tissue Breast Cancer Image Segmentation | |
| CN115468893A (en) | Liquid-based cell analysis method, apparatus 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 |