Method, device, equipment and readable storage medium for dividing region of interest
By performing clustering and consistency parameters calculations on mass spectrometry imaging data and biological tissue section staining images, the number of regions of interest is automatically determined, which solves the problem of low accuracy in the prior art division of regions of interest and improves the accuracy of mass spectrometry imaging data analysis.
Patent Information
- Application Number
- CN202210539455.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-18
- Publication Date
- 2025-05-02
- Estimated Expiration
- 2042-05-18
AI Technical Summary
When the existing clustering method is used to divide the regions of interest in mass spectrometry imaging data, since the number of regions of interest is artificially set, the accuracy of the division results is low.
By acquiring mass spectrometry imaging data and biological tissue section staining images, dividing image blocks, and clustering the mass spectrometry data and image blocks based on the number of different regions of interest, calculating the consistency parameters of the clustering results, determining the number of regions of interest corresponding to the maximum consistency parameter as the target number, and then dividing the regions of interest.
The accuracy of the division results of the region of interest in mass spectrometry imaging data is improved, and the number of target regions of interest is determined through quantitative methods to ensure the reliability of the division results.
Smart Images
Figure CN114708428B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of mass spectrometry imaging data analysis, and more specifically, to a method, device, equipment and readable storage medium for dividing a region of interest. Background Art
[0002] Mass spectrometry imaging (MSI) is an emerging chemical imaging technology. Mass spectrometry imaging separates various chemical components on the sample surface (such as proteins, peptides, lipids, metabolites, drugs, etc.) and outputs mass spectrometry imaging data. Each pixel in the mass spectrometry imaging data corresponds to a mass spectrum. Mass spectrometry imaging data analysis is generally divided into two types. The first is to find differential compounds between different regions of interest in a sample (such as changes at the molecular level between healthy tissue and diseased tissue); the second is to find differential compounds between the same regions of interest in different samples (such as changes at the molecular level in the same anatomical structure between the control group and the experimental group). Therefore, accurately delineating the region of interest (ROI) of mass spectrometry imaging data is the basic prerequisite for mass spectrometry imaging data analysis.
[0003] At present, clustering methods are often used to divide the regions of interest of mass spectrometry imaging data. Specifically, based on the basic assumption that "different tissue types have different chemical components, and different chemical components generate different mass spectra", cluster analysis is performed according to the similarity of mass spectra between pixels, that is, the pixels in the mass spectrometry imaging data are divided into clusters with the same number as the regions of interest, and each cluster corresponds to a region of interest. However, the existing method of using clustering methods to divide the regions of interest of mass spectrometry imaging data may result in low accuracy in the division results of the regions of interest of mass spectrometry imaging data because the number of regions of interest is artificially set.
[0004] Therefore, how to improve the accuracy of the results of dividing the region of interest of mass spectrometry imaging data has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention
[0005] In view of the above problems, the present application proposes a method, device, equipment and readable storage medium for dividing a region of interest to improve the accuracy of the results of dividing the region of interest of mass spectrometry imaging data.
[0006] The specific plan is as follows:
[0007] A method for dividing a region of interest, the method comprising:
[0008] Acquire mass spectrometry imaging data of a region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest;
[0009] Dividing the biological tissue section staining image into a plurality of image blocks;
[0010] For each number of regions of interest, clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result; clustering each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result; calculating a consistency parameter between the first clustering result and the second clustering result;
[0011] Determine the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest;
[0012] According to the first clustering result and the second clustering result corresponding to the number of the target regions of interest, a division result of the regions of interest of the mass spectrometry imaging data is obtained.
[0013] Optionally, dividing the biological tissue section staining image into a plurality of image blocks comprises:
[0014] Based on the pixel mapping relationship between the mass spectrometry imaging data and the biological tissue section staining image, the biological tissue section staining image is divided into a plurality of image blocks, each image block corresponding to a pixel in the mass spectrometry imaging data.
[0015] Optionally, clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result includes:
[0016] Performing dimensionality reduction on the mass spectrum of each pixel in the mass spectrometry imaging data to obtain a mass spectrum after dimensionality reduction;
[0017] A clustering algorithm is used to cluster the mass spectra after dimensionality reduction based on the number of regions of interest to obtain a first clustering result.
[0018] Optionally, clustering the image blocks in the biological tissue section staining image based on the number of the regions of interest to obtain a second clustering result includes:
[0019] Performing feature extraction on each image block to obtain a tissue morphology feature vector of each image block, wherein each element in the tissue morphology feature vector corresponds to a response intensity of a tissue morphology feature in the image block;
[0020] Based on the number of the regions of interest and the tissue morphology feature vectors of the image blocks, the image blocks in the biological tissue section staining image are clustered to obtain a second clustering result.
[0021] Optionally, the extracting features from each image block to obtain a tissue morphology feature vector of each image block includes:
[0022] A feature extraction network is used to extract features from each image block to be processed, thereby obtaining a tissue morphology feature vector for each image block to be processed. The feature extraction network is obtained by pre-training a convolutional neural network using a natural image data set.
[0023] Optionally, clustering the image blocks in the biological tissue section staining image based on the number of the regions of interest and the tissue morphology feature vectors of the image blocks to obtain a second clustering result includes:
[0024] Acquiring position information of each image block in the biological tissue slice staining image;
[0025] Generate a morphological feature data set based on the position information of each image block in the biological tissue section staining image and the tissue morphological feature vector of each image block;
[0026] Based on the morphological feature data set, a tissue morphological feature spatial distribution map is generated, wherein each spatial position point in the tissue morphological feature spatial distribution map corresponds to a tissue morphological feature spectrum;
[0027] Performing dimension reduction on each tissue morphology characteristic spectrum to obtain a reduced-dimensional tissue morphology characteristic spectrum;
[0028] A clustering algorithm is used to cluster the tissue morphology characteristic spectra after dimensionality reduction based on the number of regions of interest to obtain a second clustering result.
[0029] Optionally, the calculating a consistency parameter of the first clustering result and the second clustering result includes:
[0030] The Cohen's Kappa coefficient of the first clustering result and the second clustering result is calculated as a consistency parameter of the first clustering result and the second clustering result.
[0031] A device for dividing a region of interest, the device comprising:
[0032] An acquisition unit, used for acquiring mass spectrometry imaging data of the region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest;
[0033] An image block division unit, used for dividing the biological tissue section staining image into a plurality of image blocks;
[0034] A clustering processing unit, for clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest for each region of interest to obtain a first clustering result; clustering each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result; and calculating a consistency parameter between the first clustering result and the second clustering result;
[0035] A target region of interest number determining unit, used to determine the number of regions of interest corresponding to the maximum consistency parameter as the target region of interest number;
[0036] The region of interest division result determination unit is used to obtain the region of interest division result of the mass spectrometry imaging data according to the first clustering result and the second clustering result corresponding to the number of target regions of interest.
[0037] A device for dividing a region of interest, comprising a memory and a processor;
[0038] The memory is used to store programs;
[0039] The processor is used to execute the program to implement the various steps of the method for dividing the region of interest as described above.
[0040] A readable storage medium stores a computer program, and when the computer program is executed by a processor, the computer program implements the various steps of the method for dividing a region of interest as described above.
[0041] By means of the above technical scheme, the present application discloses a method, device, equipment and readable storage medium for dividing regions of interest. Obtain mass spectrometry imaging data of regions of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data and a plurality of numbers of regions of interest; divide the biological tissue section staining image into a plurality of image blocks; for each number of regions of interest, cluster each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result; cluster each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result, and calculate the consistency parameter of the first clustering result and the second clustering result; determine the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest; obtain the division result of the regions of interest of the mass spectrometry imaging data according to the first clustering result and the second clustering result corresponding to the target number of regions of interest. Based on the above scheme, different numbers of regions of interest are used to divide the regions of interest from two aspects, namely, the mass spectrometry imaging data and the biological tissue section staining image corresponding to the mass spectrometry imaging data. Based on the consistency of the region of interest division results from the two aspects, the number of target regions of interest is quantitatively determined, and the regions of interest divided for the mass spectrometry imaging data based on the target number of regions of interest have the highest accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0043] Figure 1 A schematic diagram of a flow chart of a method for dividing a region of interest disclosed in an embodiment of the present application;
[0044] Figure 2 A schematic flow chart of a method for clustering pixels in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result disclosed in an embodiment of the present application;
[0045] Figure 3 A schematic flow chart of a method disclosed in an embodiment of the present application for clustering each image block in the biological tissue section staining image based on the number of the regions of interest to obtain a second clustering result;
[0046] Figure 4 This is an example diagram of a method for dividing a region of interest disclosed in an embodiment of the present application;
[0047] Figure 5A schematic diagram of the structure of a device for dividing a region of interest disclosed in an embodiment of the present application;
[0048] Figure 6 This is a hardware structure block diagram of a device for dividing a region of interest disclosed in an embodiment of the present application. DETAILED DESCRIPTION
[0049] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0050] Next, the method for dividing the region of interest provided by the present application is introduced through the following embodiments.
[0051] Reference Figure 1 , Figure 1 This is a flow chart of a method for dividing a region of interest disclosed in an embodiment of the present application. The method may include:
[0052] Step S101: acquiring mass spectrometry imaging data of a region of interest to be divided, a stained image of a biological tissue section corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest.
[0053] In the present application, the biological tissue section staining image corresponding to the mass spectrometry imaging data can be obtained by various biological imaging methods, such as optical microscope imaging, two-photon imaging, etc., and the present application does not impose any limitation on this.
[0054] The number of regions of interest is consistent with the categories in subsequent clustering. For example, if the number of regions of interest is 2, there will be 2 clusters in subsequent clustering, and if the number of regions of interest is 3, there will be 3 clusters in subsequent clustering. In this application, multiple numbers of regions of interest can be preset, for example, four numbers of regions of interest can be preset, namely 2, 3, 4, and 5.
[0055] It should be noted that the mass spectrometry imaging data of the region of interest to be divided can be obtained after preprocessing the original mass spectrometry imaging data. The method of preprocessing the original mass spectrometry imaging data can be: calling the Cardinal2 library to normalize the mass spectrometry imaging data, smooth the spectrum, correct the baseline, and extract, align, filter and bin the spectrum peaks.
[0056] The biological tissue section staining image corresponding to the mass spectrometry imaging data can be obtained by preprocessing the original biological tissue section staining image. The method of preprocessing the original biological tissue section staining image can be: calling the HistomicsTK library to perform sample tissue-background segmentation, color normalization, color deconvolution and other processing on the microscopic data.
[0057] Step S102: dividing the biological tissue section staining image into a plurality of image blocks.
[0058] In the present application, as an implementable embodiment, the biological tissue section staining image can be divided into multiple image blocks based on the pixel mapping relationship between the mass spectrometry imaging data and the biological tissue section staining image, and a pixel in each image block corresponds to a pixel in the mass spectrometry imaging data.
[0059] The pixel mapping relationship between the mass spectrometry imaging data and the biological tissue section staining image can be realized by constructing an affine transformation function using the SimpleITK library.
[0060] Step S103: for each number of regions of interest, cluster each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result; cluster each image block in the biological tissue section stained image based on the number of regions of interest to obtain a second clustering result; calculate the consistency parameter of the first clustering result and the second clustering result.
[0061] It should be noted that the first clustering result is used to indicate the clustering label of each pixel in the mass spectrometry imaging data, and the second clustering result is used to indicate the clustering label of each image block in the biological tissue section staining image. The purpose of calculating the consistency parameter of the first clustering result and the second clustering result is to achieve cross-validation between the clustering results.
[0062] In the present application, as an implementable method, the Cohen's Kappa coefficient of the first clustering result and the second clustering result can be calculated as a consistency parameter of the first clustering result and the second clustering result. As another implementable method, the mutual information (MutualInformation) of the first clustering result and the second clustering result can be calculated as a consistency parameter of the first clustering result and the second clustering result.
[0063] After this step, the first clustering result, the second clustering result and the consistency parameter corresponding to the number of each region of interest can be obtained.
[0064] Step S104: determining the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest.
[0065] The higher the consistency parameter, the higher the consistency between the first clustering result and the second clustering result. The number of regions of interest corresponding to the largest consistency parameter was jointly confirmed by two independent biological imaging modalities, and the accuracy was higher.
[0066] Step S105: obtaining a result of dividing the regions of interest of the mass spectrometry imaging data according to the first clustering result and the second clustering result corresponding to the number of the target regions of interest.
[0067] In the present application, as an implementable method, the first clustering result corresponding to the number of target regions of interest may be used as the result of dividing the regions of interest of the mass spectrometry imaging data.
[0068] As another feasible implementation method, in the first clustering result corresponding to the number of target regions of interest, the clustering labels of pixels in the mass spectrometry imaging data having the same clustering labels as the clustering labels of each image block in the biological tissue section stained image in the second clustering result corresponding to the number of target regions of interest can be output as a result with higher credibility, and in the first clustering result corresponding to the number of target regions of interest, the clustering labels of pixels in the mass spectrometry imaging data having different clustering labels from the clustering labels of each image block in the biological tissue section stained image in the second clustering result corresponding to the number of target regions of interest can be output as a result with low credibility.
[0069] The present embodiment discloses a method for dividing regions of interest. Obtain mass spectrometry imaging data of regions of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and a plurality of numbers of regions of interest; divide the biological tissue section staining image into a plurality of image blocks; for each number of regions of interest, cluster each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result; cluster each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result, and calculate the consistency parameter of the first clustering result and the second clustering result; determine the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest; obtain the division result of the regions of interest of the mass spectrometry imaging data according to the first clustering result and the second clustering result corresponding to the target number of regions of interest. Based on the above scheme, different numbers of regions of interest are used to divide the regions of interest from two aspects, namely, the mass spectrometry imaging data and the biological tissue section staining image corresponding to the mass spectrometry imaging data. Based on the consistency of the region of interest division results from the two aspects, the number of target regions of interest is quantitatively determined, and the regions of interest divided for the mass spectrometry imaging data based on the target number of regions of interest have the highest accuracy.
[0070] In another embodiment of the present application, a process of clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result is described in detail.
[0071] Reference Figure 2 , Figure 2 A schematic flow chart of a method for clustering pixels in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result disclosed in an embodiment of the present application, the method may include:
[0072] Step S201: performing dimensionality reduction on the mass spectrum of each pixel in the mass spectrometry imaging data to obtain a mass spectrum after dimensionality reduction.
[0073] In the present application, a nonlinear dimensionality reduction algorithm or a linear dimensionality reduction algorithm may be used to reduce the dimension of the mass spectrum of each pixel in the mass spectrometry imaging data to obtain a mass spectrum after dimensionality reduction.
[0074] Nonlinear dimensionality reduction algorithms include t-SNE (t-distributed stochastic neighbor embedding) algorithm, UMAP (Uniform Manifold Approximation and Projection) algorithm, etc.
[0075] Linear dimensionality reduction algorithms include PCA (Principal Component Analysis) algorithm, NMF (Non-negative matrix factorization), ICA (independent component analysis), etc.
[0076] Step S202: using a clustering algorithm to cluster the mass spectra after dimensionality reduction based on the number of regions of interest to obtain a first clustering result.
[0077] In the present application, the clustering algorithm includes: any one of Kmeans, hierarchical clustering, density-based clustering method, maximum expectation clustering using Gaussian mixture model, graph group detection, and mean shift clustering.
[0078] In another embodiment of the present application, a specific implementation method of clustering the image blocks in the biological tissue section staining image based on the number of the regions of interest to obtain a second clustering result is introduced.
[0079] Reference Figure 3 , Figure 3A schematic flow chart of a method for clustering the image blocks in the biological tissue section staining image based on the number of the regions of interest disclosed in an embodiment of the present application to obtain a second clustering result, the method may include:
[0080] Step S301: extracting features from each image block to obtain a tissue morphology feature vector for each image block, wherein each element in the tissue morphology feature vector corresponds to a response intensity of a tissue morphology feature in the image block.
[0081] In the present application, the tissue morphological feature vector of each image block can be various computer vision features, such as grayscale co-occurrence matrix, threshold adjacency statistics, intensity statistical features, features of the intermediate layer output of the neural network model, etc. This application does not impose any limitations on this.
[0082] As an implementable method, a feature extraction network can be used to extract features from each image block to obtain a tissue morphology feature vector for each image block. The feature extraction network is obtained by pre-training a convolutional neural network using a natural image data set. For example, a convolutional neural network (such as DenseNet, ResNet, VGG, etc.) pre-trained on a natural image data set (such as ImageNet) is used as a feature extraction network. The image block is used as input after resizing and normalization preprocessing, and forward propagation is performed in the convolutional neural network. In a pre-selected intermediate layer (such as the conv5_block32 layer of DenseNet201), the 3-channel image block is converted into a three-dimensional array with a large number of channels and a small number of pixels, and each channel corresponds to a visual feature, i.e., a tissue morphology feature. Through a pooling operation (such as average global pooling), the three-dimensional array is reorganized into a tissue morphology feature vector, and each element in the tissue morphology feature vector corresponds to the response intensity of a morphological feature in the image block.
[0083] Step S302: obtaining position information of each image block in the biological tissue slice staining image.
[0084] Step S303: generating a morphological feature data set based on the position information of each image block in the biological tissue section staining image and the tissue morphological feature vector of each image block.
[0085] Step S304: generating a tissue morphology feature spatial distribution map based on the morphology feature data set, wherein each spatial position point in the tissue morphology feature spatial distribution map corresponds to a tissue morphology feature spectrum.
[0086] Step S305: performing dimension reduction on each tissue morphology characteristic spectrum to obtain a reduced-dimensional tissue morphology characteristic spectrum.
[0087] In the present application, a nonlinear dimensionality reduction algorithm or a linear dimensionality reduction algorithm may be used to reduce the dimension of each tissue morphology characteristic spectrum to obtain a tissue morphology characteristic spectrum after dimensionality reduction.
[0088] Nonlinear dimensionality reduction algorithms include t-SNE (t-distributed stochastic neighbor embedding) algorithm, UMAP (Uniform Manifold Approximation and Projection) algorithm, etc.
[0089] Linear dimensionality reduction algorithms include PCA (Principal Component Analysis) algorithm, NMF (Non-negative matrix factorization), ICA (independent component analysis), etc.
[0090] Step S306: using a clustering algorithm to cluster the tissue morphology characteristic spectra after dimension reduction based on the number of regions of interest to obtain a second clustering result.
[0091] In the present application, the clustering algorithm may include: any one of Kmeans, hierarchical clustering, density-based clustering method, maximum expectation clustering using Gaussian mixture model, graph group detection, and mean shift clustering.
[0092] It should be noted that the clustering algorithm used in this step needs to be consistent with the clustering algorithm used in step S202.
[0093] For easier understanding, see Figure 4 , Figure 4 FIG. 1 is an example diagram of a method for dividing a region of interest disclosed in an embodiment of the present application. Figure 4 As shown, part (a) is a schematic diagram of clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result, and part (b) is a schematic diagram of clustering each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result.
[0094] The following is a description of a device for dividing a region of interest disclosed in an embodiment of the present application. The device for dividing a region of interest described below and the method for dividing a region of interest described above may refer to each other.
[0095] Reference Figure 5 , Figure 5 FIG. 1 is a schematic diagram of a device for dividing a region of interest disclosed in an embodiment of the present application. Figure 5 As shown, the device for dividing the region of interest may include:
[0096] An acquisition unit 11 is used to acquire mass spectrometry imaging data of a region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest;
[0097] An image block division unit 12, used for dividing the biological tissue section staining image into a plurality of image blocks;
[0098] The clustering processing unit 13 is used to cluster each pixel in the mass spectrometry imaging data based on the number of regions of interest for each region of interest to obtain a first clustering result; cluster each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result; and calculate a consistency parameter between the first clustering result and the second clustering result;
[0099] A target region of interest number determining unit 14, configured to determine the number of regions of interest corresponding to the maximum consistency parameter as the target region of interest number;
[0100] The region of interest division result determination unit 15 is used to obtain the region of interest division result of the mass spectrometry imaging data according to the first clustering result and the second clustering result corresponding to the number of target regions of interest.
[0101] Reference Figure 6 , Figure 6 The hardware structure block diagram of the device for dividing the region of interest provided in the embodiment of the present application is shown in FIG. Figure 6 , the hardware structure of the device for dividing the region of interest may include: at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4;
[0102] In the embodiment of the present application, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 communicate with each other through the communication bus 4;
[0103] The processor 1 may be a central processing unit CPU, or an application-specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention, etc.;
[0104] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory;
[0105] The memory stores a program, and the processor can call the program stored in the memory, wherein the program is used to:
[0106] Acquire mass spectrometry imaging data of a region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest;
[0107] Dividing the biological tissue section staining image into a plurality of image blocks;
[0108] For each number of regions of interest, clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result; clustering each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result; calculating a consistency parameter between the first clustering result and the second clustering result;
[0109] Determine the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest;
[0110] According to the first clustering result and the second clustering result corresponding to the number of the target regions of interest, a division result of the regions of interest of the mass spectrometry imaging data is obtained.
[0111] Optionally, the detailed functions and extended functions of the program may refer to the above description.
[0112] The embodiment of the present application further provides a readable storage medium, which may store a program suitable for execution by a processor, wherein the program is used to:
[0113] Acquire mass spectrometry imaging data of a region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest;
[0114] Dividing the biological tissue section staining image into a plurality of image blocks;
[0115] For each number of regions of interest, clustering each pixel in the mass spectrometry imaging data based on the number of regions of interest to obtain a first clustering result; clustering each image block in the biological tissue section staining image based on the number of regions of interest to obtain a second clustering result; calculating a consistency parameter between the first clustering result and the second clustering result;
[0116] Determine the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest;
[0117] According to the first clustering result and the second clustering result corresponding to the number of the target regions of interest, a division result of the regions of interest of the mass spectrometry imaging data is obtained.
[0118] Optionally, the detailed functions and extended functions of the program may refer to the above description.
[0119] Finally, it should be noted that, in this article, relational terms such as first and second, etc. 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 terms "include", "comprise" or any other variants thereof are 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 "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0120] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.
[0121] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may 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 shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for dividing a region of interest, characterized in that: The method comprises: Acquire mass spectrometry imaging data of a region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest; Dividing the biological tissue section staining image into a plurality of image blocks; For each number of regions of interest, the mass spectra of each pixel in the mass spectrometry imaging data are reduced in dimension to obtain a mass spectra after dimension reduction; a clustering algorithm is used to cluster each mass spectra after dimension reduction based on the number of regions of interest to obtain a first clustering result; feature extraction is performed on each image block to obtain a tissue morphology feature vector of each image block, wherein each element in the tissue morphology feature vector corresponds to the response intensity of a tissue morphology feature in the image block; based on the number of regions of interest and the tissue morphology feature vectors of each image block, each image block in the biological tissue section staining image is clustered to obtain a second clustering result; the Cohen's Kappa coefficient of the first clustering result and the second clustering result is calculated as a consistency parameter of the first clustering result and the second clustering result; Determine the number of regions of interest corresponding to the maximum consistency parameter as the target number of regions of interest; According to the first clustering result and the second clustering result corresponding to the number of the target regions of interest, a division result of the regions of interest of the mass spectrometry imaging data is obtained.
2. The method according to claim 1, characterized in that The step of dividing the biological tissue section staining image into a plurality of image blocks comprises: Based on the pixel mapping relationship between the mass spectrometry imaging data and the biological tissue section staining image, the biological tissue section staining image is divided into a plurality of image blocks, and a pixel in each image block corresponds to a pixel in the mass spectrometry imaging data.
3. The method according to claim 1, characterized in that: The step of extracting features from each image block to obtain a tissue morphology feature vector for each image block includes: A feature extraction network is used to extract features from each image block to be processed, thereby obtaining a tissue morphology feature vector for each image block to be processed. The feature extraction network is obtained by pre-training a convolutional neural network using a natural image data set.
4. The method according to claim 1, characterized in that: The clustering of each image block in the biological tissue section staining image based on the number of the regions of interest and the tissue morphology feature vector of each image block to obtain a second clustering result includes: Acquiring position information of each image block in the biological tissue slice staining image; Generate a morphological feature data set based on the position information of each image block in the biological tissue section staining image and the tissue morphological feature vector of each image block; Based on the morphological feature data set, a tissue morphological feature spatial distribution map is generated, wherein each spatial position point in the tissue morphological feature spatial distribution map corresponds to a tissue morphological feature spectrum; Performing dimension reduction on each tissue morphology characteristic spectrum to obtain a reduced-dimensional tissue morphology characteristic spectrum; A clustering algorithm is used to cluster the tissue morphology characteristic spectra after dimensionality reduction based on the number of regions of interest to obtain a second clustering result.
5. A device for dividing a region of interest, characterized in that: The device comprises: An acquisition unit, used for acquiring mass spectrometry imaging data of the region of interest to be divided, a biological tissue section staining image corresponding to the mass spectrometry imaging data, and the number of multiple regions of interest; An image block division unit, used for dividing the biological tissue section staining image into a plurality of image blocks; A clustering processing unit is used to reduce the dimension of the mass spectrum of each pixel in the mass spectrometry imaging data according to the number of each region of interest to obtain a mass spectrum after dimension reduction; cluster the mass spectrum after dimension reduction based on the number of regions of interest using a clustering algorithm to obtain a first clustering result; extract features from each image block to obtain a tissue morphology feature vector of each image block, wherein each element in the tissue morphology feature vector corresponds to the response intensity of a tissue morphology feature in the image block; cluster each image block in the biological tissue section staining image based on the number of regions of interest and the tissue morphology feature vector of each image block to obtain a second clustering result; calculate the Cohen's Kappa coefficient of the first clustering result and the second clustering result as a consistency parameter of the first clustering result and the second clustering result; A target region of interest number determining unit, used to determine the number of regions of interest corresponding to the maximum consistency parameter as the target region of interest number; The region of interest division result determination unit is used to obtain the region of interest division result of the mass spectrometry imaging data according to the first clustering result and the second clustering result corresponding to the number of target regions of interest.
6. A device for dividing a region of interest, characterized in that: including memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the method for dividing a region of interest as described in any one of claims 1 to 4.
7. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the method for dividing a region of interest as claimed in any one of claims 1 to 4 is implemented.
Citation Information
Patent Citations
Method for eliminating shadow
CN102332157A
Image segmentation method based on immune meme clustering
CN102663751A