WO2024119321A1

WO2024119321A1 - Cell segmentation processing method and apparatus, and electronic device

Info

Publication number: WO2024119321A1
Application number: PCT/CN2022/136657
Authority: WO
Inventors: 黄子睿; 刘伟庆; 李美; 黎宇翔
Original assignee: 深圳华大生命科学研究院
Priority date: 2022-12-05
Filing date: 2022-12-05
Publication date: 2024-06-13
Also published as: CN119836649A

Abstract

The present application relates to the technical field of data processing, and relates to a cell segmentation processing method and apparatus, and an electronic device. The method comprises: first, preprocessing a gene expression map of a cell to obtain a preprocessed map; then, performing binarization processing on the preprocessed map to obtain an initial mask pattern; and then, according to the initial mask pattern and by using a watershed algorithm based on distance transformation, segmenting a connected domain where there is cell adhesion to obtain a segmented mask pattern. By using the technical solution in the present application, multiple image processing methods are combined, such that a relatively reliable cell segmentation result can be provided. Cell segmentation does not rely on an image map, and the additional introduction of a technique for aligning an image map with a gene expression map is not required, thereby eliminating introduced additional errors, and also saving on the overall operation time and technical cost.

Description

Cell segmentation processing method, device and electronic equipment

Technical Field

The present application relates to the field of data processing technology, and in particular to a method, device and electronic equipment for processing cell segmentation.

Background technique

Cell segmentation is an important part of extracting intracellular gene expression in spatiotemporal omics technology. Segmenting cells to obtain the corresponding gene expression at the corresponding spatial position is an indispensable step in the analysis process.

Currently, existing cell segmentation methods usually use microscope photographs or scanned images of corresponding biological tissue sections to perform cell segmentation, supplemented by the registration of images with gene expression maps for subsequent gene extraction and analysis.

However, existing cell segmentation methods have high requirements on the quality of images, and for subsequent analysis, the images need to be aligned with the gene expression maps, which introduces additional technical requirements and possible errors, which in turn affects the efficiency and accuracy of the cell segmentation process and increases the technical costs.

Summary of the invention

In view of this, the present application provides a cell segmentation processing method, device and electronic device, the main purpose of which is to improve the technical problems that the current existing cell segmentation methods will affect the efficiency and accuracy of cell segmentation processing and also increase the technical cost.

In a first aspect, the present application provides a method for processing cell segmentation, comprising:

Obtain gene expression profiles of cells;

Preprocessing the gene expression graph to obtain a preprocessed graph;

Binarizing the preprocessed image to obtain an initial mask image;

According to the initial mask image, a watershed algorithm based on distance transformation is used to segment the connected domain where cell adhesion exists, so as to obtain a segmented mask image.

In a second aspect, the present application provides a cell segmentation processing device, comprising:

an acquisition module, configured to acquire a gene expression profile of a cell;

A processing module is configured to preprocess the gene expression graph to obtain a preprocessed graph;

The segmentation module is configured to perform binarization processing on the pre-processed image to obtain an initial mask image; based on the initial mask image, a watershed algorithm based on distance transformation is used to segment the connected domain with cell adhesion to obtain a segmented mask image.

In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the cell segmentation processing method described in the first aspect.

In a fourth aspect, the present application provides an electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the cell segmentation processing method described in the first aspect when executing the computer program.

By means of the above technical scheme, the present application provides a method, device and electronic device for processing cell segmentation. Compared with the currently available cell segmentation methods, the present application provides a solution for performing cell segmentation directly based on the gene expression map. Specifically, the gene expression map of the cell is first preprocessed to obtain a preprocessed map; then the preprocessed map is binarized to obtain an initial mask map; then, according to the initial mask map, a watershed algorithm based on distance transformation is used to segment the connected domains where cell adhesion exists to obtain a segmented mask map. By applying the technical scheme of the present application and combining multiple image processing methods, a more reliable cell segmentation result can be provided. Cell segmentation does not rely on the image map, and does not require the additional introduction of technology for aligning the image map with the gene expression map, which eliminates the introduction of additional errors, while saving overall operation time and technical costs, and can improve the efficiency and accuracy of cell segmentation processing.

The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

FIG1 is a schematic diagram showing a process flow of a cell segmentation processing method provided in an embodiment of the present application;

FIG2 is a schematic diagram showing a flow chart of another cell segmentation processing method provided in an embodiment of the present application;

FIG3 shows a schematic diagram of an example process flow based on the method of this embodiment provided in an embodiment of the present application;

FIG4 shows an example diagram of the effect of the gene expression graph provided in an embodiment of the present application;

FIG5 shows an example diagram of a sharpened image obtained through sharpening processing provided by an embodiment of the present application;

FIG6 is a schematic diagram showing the effect of an initial mask image provided by an embodiment of the present application;

FIG7 is a schematic diagram showing the effect of an output mask image provided by an embodiment of the present application;

FIG8 shows a schematic structural diagram of a cell segmentation processing device provided in an embodiment of the present application.

Detailed ways

In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the scheme of the present application will be further described below. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

In order to improve the technical problem that the existing cell segmentation method affects the efficiency and accuracy of the cell segmentation process and also increases the technical cost, this embodiment provides a cell segmentation processing method, as shown in FIG1 , the method includes:

Step 101: Obtain a gene expression map of a cell.

A gene expression map, or gene expression atlas, can be obtained based on gene expression data.

Step 102: preprocess the gene expression map of the cell to obtain a preprocessed map.

Since the gene expression map is a scatter plot, it is not convenient for cell segmentation, so preprocessing is required to process the gene expression map of the cell to obtain a preprocessed map with enhanced boundary effect.

Step 103: binarize the preprocessed image to obtain an initial mask image.

In this embodiment, the Otsu method may be used to perform binarization processing on the preprocessed image to obtain an initial mask image.

Step 104: According to the initial mask image, a watershed algorithm based on distance transformation is used to segment the connected domain where the cells are adhered, so as to obtain a segmented mask image.

The watershed algorithm considers image segmentation based on the composition of the watershed.

Compared with the existing cell segmentation methods, this embodiment provides a solution for cell segmentation directly based on gene expression maps, which uses a combination of multiple image processing methods to provide more reliable cell segmentation results. Cell segmentation does not rely on image maps, and does not require the introduction of additional technology to align image maps with gene expression maps, eliminating the introduction of additional errors, while saving overall operation time and technical costs, and can improve the efficiency and accuracy of cell segmentation processing.

Further, as a refinement and extension of the above embodiment, in order to fully illustrate the specific implementation process of the method of this embodiment, this embodiment provides a specific method as shown in FIG. 2, which includes:

Step 201: Obtain a gene expression matrix including spatial positions.

A gene expression matrix including spatial positions is obtained from gene expression data of the cell. The gene expression data may include gene identifiers, coordinate positions and total gene expression amounts of the corresponding coordinate positions of multiple genes.

Step 202: Generate a gene expression map of the cell based on the gene expression matrix.

As an optional method, step 202 may specifically include: first obtaining the coordinate position of the expressed gene in the gene expression matrix and the total gene expression amount at the corresponding coordinate position; then generating a gene expression map based on the coordinate position of the expressed gene and the total gene expression amount at the corresponding coordinate position, wherein the gene expression map is a grayscale map, and the grayscale value of the pixel point in the gene expression map is the total gene expression amount at the coordinate position corresponding to the pixel point. Through this optional method, the gene expression map of the cell can be accurately generated.

For example, in the specific use of cell segmentation, a gene expression matrix containing spatial positions is input, and an expression image is generated based on the coordinate positions of the expressed genes and the total gene expression amounts at the corresponding positions. The specific form of the image is a grayscale image, and the grayscale value of the coordinate point is the total amount of gene expression at the coordinate.

As another optional method, step 202 may specifically include: drawing a gene expression map of the cell according to the gene expression matrix and the segmented mask map, wherein the spatial position in the gene expression matrix corresponds to the spatial position in the segmented mask map. For example, a cell-based expression map is drawn according to the gene expression matrix and the segmented mask map, and the spatial position in the gene expression matrix may correspond to the spatial position in the segmented mask map. Through this optional method, the gene expression map of the cell can be accurately generated.

Since the generated gene expression image is a scatter plot, it is not easy to segment and needs to be processed. Specifically, the process shown in step 203 can be performed.

Step 203: pre-process the gene expression graph to obtain a median graph, and sharpen the median graph to obtain a sharpened graph.

Since the gene expression graph is a scatter plot, it is not convenient to perform cell segmentation, so preprocessing is required. In this embodiment, the gene expression graph of the cell is first processed into a median graph. Then, the median graph can be sharpened using a Laplacian operator to enhance the boundary effect of the median graph, thereby obtaining a sharpened graph.

Optionally, the gene expression map is preprocessed to obtain a median map, which may specifically include: first, using a convolution kernel of a preset size (such as 13*13) to perform a convolution operation on the gene expression map so that the scattered points in the gene expression map are adhered to obtain a first convolution map; then detecting the local maximum point of the first convolution map according to the two-dimensional grayscale peak of the image; obtaining the pth percentile of the local maximum point, wherein p is a preset value, such as the pth percentile may be a 98% percentile value, or a 99% percentile value, etc.; if the pth percentile is within a preset range, using a first median filter to perform median filtering on the first convolution map to obtain a median map, wherein the filter size of the first median filter is determined according to the preset size.

Further optionally, after obtaining the pth percentile in the local maximum point, the method of this embodiment may also include: if the pth percentile is outside the preset range, determining the new size of the convolution kernel based on the pth percentile and the preset size; performing a convolution operation on the gene expression graph using the convolution kernel of the new size, so that the scattered points in the gene expression graph are adhered to obtain a second convolution graph; performing a median filtering on the second convolution graph using a second median filter to obtain a median graph, wherein the filter size of the second median filter is determined based on the new size.

Exemplarily, determining the new size of the convolution kernel based on the pth percentile and the preset size may specifically include: calculating the new size of the convolution kernel according to the formula K=N*(N/R), wherein K*K represents the new size of the convolution kernel, N*N represents the preset size, and R represents the pth percentile.

For example, first use a convolution kernel of size 13*13 (empirical value) to perform a convolution operation on the original image of the gene expression map to make the scatter plot stick together to obtain the first convolution map; then detect the local maximum point of the first convolution map according to the two-dimensional grayscale peak of the image, and take out the 99% quantile value R of all the local maximum points. If the R value is too different from a set empirical threshold (too high or too low, it will affect subsequent processing), it is considered that the 13*13 convolution kernel is not suitable for the original image. At this time, calculate K=13*(13/R), change the original 13*13 convolution kernel to a K*K convolution kernel, reprocess the original image, and obtain the second convolution map; if the R value is within the allowable range, use the first convolution map. The convolution image is in the form of stacked blocks of different grayscales. The grayscale span is quite dramatic, and some images express the phenomenon of hollow centers of clusters. To eliminate this problem, 13*2.7=35 (2.7 can be an empirical value, and the decimal is rounded down) is used as the filter size (if K needs to be calculated, the kernel size here is K*2.7) to perform median filtering on the second convolution image to obtain the median image.

In order to fill the voids in the point clusters, the median filter used above is relatively large, and the grayscale boundary of the median image may be relatively blurred. The laplacian operator is used to sharpen the median image to enhance the grayscale boundary, and a sharpened image is obtained to complete the preprocessing. Next, the cell segmentation process is performed, and specifically, the process shown in steps 204 to 207 can be executed.

Step 204: binarize the sharpened image to obtain an initial mask image.

The sharpened image obtained in the previous step is binarized using the Otsu method to obtain the initial mask image.

Step 205: Filter the connected domains in the initial mask image whose areas do not meet the preset conditions to obtain a filtered mask image.

Optionally, step 205 may specifically include: filtering the connected domains in the initial mask image whose area is greater than a first preset threshold or the connected domains whose area is less than a second preset threshold to obtain a filtered mask image, wherein the first preset threshold is greater than the second preset threshold. For example, an empirical threshold is used to filter the connected domains in the initial mask whose area is too small or too large to obtain a filtered mask image.

Step 206 , traverse each connected domain in the filter mask image, extract the area where the connected domain is located based on the minimum circumscribed rectangle of the connected domain, and use the watershed algorithm to segment the connected domain with cell adhesion to obtain a segmented mask image.

Optionally, step 206 may specifically include: setting the grayscale value of each pixel in each connected domain in the filter mask image to a first value, and setting the grayscale value of each pixel outside the connected domain to a second value; for each target pixel in the connected domain whose grayscale value is the first value, remapping the grayscale value of the target pixel to the distance from the target pixel to the nearest pixel whose grayscale value is the second value, to obtain a distance map of the connected domain; binarizing the distance map of the connected domain to obtain a preset number of pixels in the connected domain that are farthest from the pixel whose grayscale value is the second value; using the preset number of pixels as injection points of the watershed algorithm, and using the watershed function to perform watershed segmentation on the original mask of the connected domain in the filter mask image to obtain a segmented target connected domain, and covering the target connected domain with the filter mask image to obtain a segmented mask image.

For example, traverse each connected domain in the filter mask image, extract the area where the connected domain is located based on the minimum circumscribed rectangle of the connected domain, and use the watershed algorithm based on distance transformation to further segment the connected domain where cell adhesion may exist. The specific steps are as follows:

Step a: For each connected domain area extracted, the grayscale value of the points in the target connected domain is set to 1, and the grayscale value of the points outside the connected domain is set to 0 (including background and non-target connected domains). A distance transformation is performed on each point with a value of 1, and its grayscale value is remapped to the distance from the point to the nearest point with a value of 0 (the distance between adjacent points is 1), thereby obtaining a distance map of the connected domain.

Step b: Use an empirical threshold to binarize the distance map to obtain the farthest points from the point with a distance value of 0 in the connected domain (since each connected domain is different, the number of points is not fixed).

Step c: Use the several points obtained in the previous step as water injection points of the watershed algorithm, use the watershed function in OpenCV to perform watershed segmentation on the original mask of the connected domain, obtain the segmented target connected domain, and overlay the result onto the filter mask image.

Step d: perform the above steps on each connected domain traversed to obtain a segmented mask image.

Step 207: Perform a closure operation on the segmented mask image to obtain a mask image of the cell segmentation result.

In order to make the cell boundaries more regular, the segmented mask is closed to obtain the final mask map. Finally, the final mask map can be output and saved for subsequent biological analysis at the cell level in combination with the original expression matrix.

For example, as shown in FIG3, it is a schematic diagram of an example flow chart based on the method of this embodiment. First, a gene expression matrix can be input, and a gene expression image can be generated based on the matrix, as shown in FIG4. Then, a 13*13 convolution kernel is used to perform a convolution operation on the gene expression map, so that the scattered points in the gene expression map are adhered to obtain a first convolution map. At this time, it is necessary to judge by a threshold, that is, the local maximum point of the first convolution map is detected according to the two-dimensional grayscale peak of the image, and the 99% quantile value R of all the local maximum points is taken out. If the R value is too far from a set empirical threshold (too high or too low, it will affect subsequent processing), it is considered that the 13*13 convolution kernel is not suitable for the original image, and a new convolution kernel size is calculated using a ratio, and then the new convolution kernel is used to process the original image to obtain a second convolution map, and the median filter size is calculated using this ratio, and then the second convolution map is processed with a median filter of this size to obtain a median map. If the R value is within the allowable range, the first convolution map is used, and the first convolution map is processed with a median filter of size 35 to obtain a median map.

After obtaining the median image, the laplacian operator is used to sharpen the median image to obtain a sharpened image, as shown in Figure 5. The sharpened image is binarized using the large law method to obtain the initial mask image, as shown in Figure 6, and then the adhesion cells are segmented by area filtering and watershed algorithm. Finally, the cell mask image is output and saved, as shown in Figure 7.

Furthermore, as a specific implementation of the method shown in FIG. 1 and FIG. 2 , this embodiment provides a cell segmentation processing device, as shown in FIG. 8 , the device includes: an acquisition module 31 , a processing module 32 , and a segmentation module 33 .

An acquisition module 31 is configured to acquire a gene expression profile of a cell;

A processing module 32 is configured to preprocess the gene expression graph to obtain a preprocessed graph;

The segmentation module 33 is configured to perform binarization processing on the pre-processed image to obtain an initial mask image; based on the initial mask image, use a watershed algorithm based on distance transformation to segment the connected domain with cell adhesion to obtain a segmented mask image.

In a specific application scenario, the segmentation module 33 is specifically configured to filter out the connected domains in the initial mask image whose areas do not meet the preset conditions to obtain a filtered mask image; traverse each connected domain in the filtered mask image, extract the area where the connected domain is located based on the minimum circumscribed rectangle of the connected domain, and use the watershed algorithm to segment the connected domain with cell adhesion to obtain a segmented mask image.

In a specific application scenario, the segmentation module 33 is further configured to set the grayscale value of each pixel in each connected domain in the filter mask image to a first value, and set the grayscale value of each pixel outside the connected domain to a second value; for each target pixel in the connected domain whose grayscale value is the first value, remap the grayscale value of the target pixel to the distance from the target pixel to the pixel whose grayscale value is the second value closest to it, to obtain a distance map of the connected domain; binarize the distance map of the connected domain to obtain a preset number of pixels in the connected domain that are farthest from the pixel whose grayscale value is the second value; use the preset number of pixels as injection points of the watershed algorithm, and use the watershed function to perform watershed segmentation on the original mask of the connected domain in the filter mask image to obtain a segmented target connected domain, and cover the target connected domain with the filter mask image to obtain a segmented mask image.

In a specific application scenario, the segmentation module 33 is further configured to filter the connected domains in the initial mask image whose area is greater than a first preset threshold, or the connected domains whose area is less than a second preset threshold, to obtain the filtered mask image, wherein the first preset threshold is greater than the second preset threshold.

In a specific application scenario, the processing module 32 is specifically configured to pre-process the gene expression graph to obtain a median graph; and perform sharpening processing on the median graph to obtain a sharpened graph.

Correspondingly, the segmentation module 33 is specifically configured to perform binarization processing on the sharpening image to obtain an initial mask image.

In a specific application scenario, the processing module 32 is further configured to perform a convolution operation on the gene expression map using a convolution kernel of a preset size, so that the scattered points in the gene expression map are adhered to obtain a first convolution map; detect the local maximum point of the first convolution map according to the two-dimensional grayscale peak of the image; obtain the pth percentile of the local maximum point, wherein p is a preset value; if the pth percentile is within a preset range, perform median filtering on the first convolution map using a first median filter to obtain the median map, wherein the filter size of the first median filter is determined according to the preset size.

In a specific application scenario, the processing module 32 is further configured to, after obtaining the pth percentile in the local maximum point, if the pth percentile is outside a preset range, determine a new size of the convolution kernel according to the pth percentile and the preset size; perform a convolution operation on the gene expression map using the convolution kernel of the new size so that the scattered points in the gene expression map are adhered to obtain a second convolution map; perform a median filtering on the second convolution map using a second median filter to obtain the median map, wherein the filter size of the second median filter is determined based on the new size.

In a specific application scenario, the processing module 32 is further configured to calculate the new size of the convolution kernel according to the formula K=N*(N/R), wherein K*K represents the new size of the convolution kernel, N*N represents the preset size, and R represents the pth percentile.

In a specific application scenario, the acquisition module 31 is specifically configured to acquire a gene expression matrix including spatial positions; and generate the gene expression graph based on the gene expression matrix.

In a specific application scenario, the acquisition module 31 is specifically configured to obtain the coordinate position of the expressed gene in the gene expression matrix and the total gene expression amount of the corresponding coordinate position; based on the coordinate position of the expressed gene and the total gene expression amount of the corresponding coordinate position, generate the gene expression map, wherein the gene expression map is a grayscale map, and the grayscale value of the pixel point in the gene expression map is the total gene expression amount of the coordinate position corresponding to the pixel point.

In a specific application scenario, the acquisition module 31 is further configured to draw a gene expression map of the cell based on the gene expression matrix and the segmented mask map, wherein the spatial position in the gene expression matrix corresponds to the spatial position in the segmented mask map.

In a specific application scenario, the segmentation module 33 is further configured to perform a closure operation on the segmented mask image to obtain a mask image of a cell segmentation result.

It should be noted that, for other corresponding descriptions of the functional units involved in the cell segmentation processing device provided in this embodiment, reference may be made to the corresponding descriptions in FIG. 1 and FIG. 2 , which will not be repeated here.

Based on the above method as shown in FIG. 1 and FIG. 2 , accordingly, this embodiment further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above method as shown in FIG. 1 and FIG. 2 is implemented.

Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of the present application.

Based on the above method shown in Figures 1 and 2, and the virtual device embodiment shown in Figure 8, in order to achieve the above purpose, the embodiment of the present application also provides an electronic device, which can be a personal computer, a laptop computer, etc., and the device includes a storage medium and a processor; the storage medium is used to store computer programs; the processor is used to execute the computer program to implement the above method shown in Figures 1 and 2.

Optionally, the above-mentioned physical device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

Those skilled in the art will appreciate that the above-mentioned physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different arrangements of components.

The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the above-mentioned physical device, and supports the operation of the information processing program and other software and/or programs. The network communication module is used to realize the communication between the components inside the storage medium, and the communication with other hardware and software in the information processing physical device.

Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware. By applying the solution of this embodiment, compared with the currently available cell segmentation method, this embodiment provides a solution for cell segmentation directly based on gene expression maps, using a combination of multiple image processing methods to provide more reliable cell segmentation results. Cell segmentation does not rely on image maps, and does not require the additional introduction of technology for aligning image maps with gene expression maps, eliminating the introduction of additional errors, while saving overall operation time and technical costs, and can improve the efficiency and accuracy of cell segmentation processing.

It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

The above is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features applied for herein.

Claims

A cell segmentation processing method, characterized by comprising:

Obtain gene expression profiles of cells;

Preprocessing the gene expression graph to obtain a preprocessed graph;

Binarizing the preprocessed image to obtain an initial mask image;

According to the initial mask image, a watershed algorithm based on distance transformation is used to segment the connected domain where cell adhesion exists, so as to obtain a segmented mask image.
The method according to claim 1, characterized in that the segmenting of the connected domains where cell adhesion exists using a watershed algorithm based on distance transformation according to the initial mask image to obtain a segmented mask image comprises:

Filtering the connected domains whose areas do not meet the preset conditions in the initial mask image to obtain a filtered mask image;

Each connected domain in the filtering mask image is traversed, and the area where the connected domain is located is extracted based on the minimum circumscribed rectangle of the connected domain. The connected domain with cell adhesion is segmented using the watershed algorithm to obtain a segmented mask image.
The method according to claim 2 is characterized in that the traversing each connected domain in the filter mask image, extracting the area where the connected domain is located based on the minimum circumscribed rectangle of the connected domain, and using the watershed algorithm to segment the connected domain where cell adhesion exists to obtain the segmented mask image, comprises:

The grayscale value of each pixel in each connected domain in the filter mask image is set to a first value, and the grayscale value of each pixel outside the connected domain is set to a second value;

For each target pixel whose grayscale value is the first value in the connected domain, remap the grayscale value of the target pixel to the distance from the target pixel to the nearest pixel whose grayscale value is the second value, to obtain a distance map of the connected domain;

Binarize the distance map of the connected domain to obtain a preset number of pixel points in the connected domain that are farthest from the pixel point whose grayscale value is the second value;

The preset number of pixel points are used as water injection points of the watershed algorithm, and the original mask of the connected domain in the filtering mask image is subjected to watershed segmentation using the watershed function to obtain the segmented target connected domain, and the target connected domain is covered on the filtering mask image to obtain the segmented mask image.
The method according to claim 2, characterized in that filtering the connected domains in the initial mask image whose areas do not meet preset conditions to obtain a filtered mask image comprises:

The connected domains whose areas are greater than a first preset threshold or the connected domains whose areas are less than a second preset threshold in the initial mask image are filtered to obtain the filtered mask image, wherein the first preset threshold is greater than the second preset threshold.
The method according to claim 1, characterized in that the preprocessing of the gene expression graph to obtain the preprocessed graph comprises:

Preprocessing the gene expression graph to obtain a median graph;

Performing sharpening processing on the median image to obtain a sharpened image;

The binarization of the preprocessed image to obtain an initial mask image comprises:

The sharpened image is binarized to obtain an initial mask image.
The method according to claim 5, characterized in that the preprocessing of the gene expression graph to obtain a median graph comprises:

Performing a convolution operation on the gene expression graph using a convolution kernel of a preset size so that scattered points in the gene expression graph are adhered to obtain a first convolution graph;

Detecting the local maximum point of the first convolution image according to the two-dimensional grayscale peak of the image;

Obtaining the pth percentile of the local maximum point, wherein p is a preset value;

If the pth percentile is within a preset range, the first convolution image is median filtered using a first median filter to obtain the median image, wherein the filter size of the first median filter is determined based on the preset size.
The method according to claim 6, characterized in that after obtaining the pth percentile of the local maximum point, the method further comprises:

If the pth percentile is outside a preset range, determining a new size of the convolution kernel according to the pth percentile and the preset size;

Performing a convolution operation on the gene expression graph using the convolution kernel of the new size, so that scattered points in the gene expression graph are adhered to obtain a second convolution graph;

Performing median filtering on the second convolution image using a second median filter to obtain the median image, wherein the filter size of the second median filter is determined according to the new size.
The method according to claim 7, characterized in that the determining the new size of the convolution kernel according to the pth percentile and the preset size comprises:

According to the formula K=N*(N/R), the new size of the convolution kernel is calculated, where K*K represents the new size of the convolution kernel, N*N represents the preset size, and R represents the pth percentile.
The method according to claim 1, characterized in that obtaining the gene expression profile of the cell comprises:

Obtain a gene expression matrix containing spatial locations;

Based on the gene expression matrix, the gene expression map is generated.
The method according to claim 9, characterized in that generating the gene expression graph based on the gene expression matrix comprises:

Obtaining the coordinate positions of the expressed genes in the gene expression matrix and the total gene expression amounts at the corresponding coordinate positions;

The gene expression map is generated according to the coordinate positions of the expressed genes and the total gene expression amounts at the corresponding coordinate positions, wherein the gene expression map is a grayscale map, and the grayscale value of a pixel point in the gene expression map is the total gene expression amount at the coordinate position corresponding to the pixel point.
The method according to claim 9, characterized in that generating the gene expression graph based on the gene expression matrix comprises:

A gene expression map of a cell is drawn according to the gene expression matrix and the segmented mask map, wherein the spatial position in the gene expression matrix corresponds to the spatial position in the segmented mask map.
The method according to any one of claims 1 to 11, characterized in that after segmenting the connected domain with cell adhesion using a watershed algorithm based on distance transformation according to the initial mask image to obtain a segmented mask image, the method further comprises:

A closure operation is performed on the segmented mask image to obtain a mask image of a cell segmentation result.
A cell segmentation processing device, characterized in that it comprises:

an acquisition module, configured to acquire a gene expression profile of a cell;

A processing module is configured to preprocess the gene expression graph to obtain a preprocessed graph;

The segmentation module is configured to perform binarization processing on the pre-processed image to obtain an initial mask image; based on the initial mask image, a watershed algorithm based on distance transformation is used to segment the connected domain with cell adhesion to obtain a segmented mask image.
A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method according to any one of claims 1 to 12.
An electronic device comprises a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 12 when executing the computer program.