Cell classification method and device for brain image, equipment and storage medium

Abstract

The invention discloses a brain image cell classification method and device, equipment and a storage medium. The method comprises the following steps: acquiring at least two to-be-classified cell images in a target brain image; for each to-be-classified cell image, carrying out structural feature extraction on the to-be-classified cell image to obtain a structural feature matrix, and determining a cell feature vector of the to-be-classified cell image according to the structural feature matrix; and according to the cell feature vectors corresponding to the to-be-classified cell images, determining a cell classification result of each to-be-classified cell image in the target brain image. The embodiment of the invention solves the problem that a traditional cell classification method is poor in brain cell system classification effect; and the cell classification accuracy and classification efficiency of the brain image are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a method, apparatus, device and storage medium for cell classification of brain images. Background Art

[0002] The brain is the source of all human behavior and consciousness. Its complexity has led researchers to develop various imaging technologies in order to achieve accurate observation of the brain.

[0003] At present, most traditional cell classification methods use machine learning models. The principle is to use images containing cells or shallow appearance features of cells as input data of the machine learning model. For example, shallow appearance features may be the imaging color, imaging texture and imaging shape of cells. Therefore, traditional cell classification methods are not suitable for the complex cell system of the brain, and the classification effect is poor. Summary of the Invention

[0004] Embodiments of the present invention provide a method, apparatus, device, and storage medium for cell classification of brain images to address the problem that traditional cell classification methods have poor classification effects on brain cell systems and to improve the accuracy and efficiency of cell classification of brain images.

[0005] According to one embodiment of the present invention, a method for cell classification of a brain image is provided, the method comprising:

[0006] Acquire at least two cell images to be classified in a target brain image;

[0007] For each cell image to be classified, extracting structural features of the cell image to be classified to obtain a structural feature matrix, and determining a cell feature vector of the cell image to be classified based on the structural feature matrix;

[0008] determining a cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors respectively corresponding to each cell image to be classified;

[0009] The structural feature matrix represents the structural information of cells in the cell image to be classified, and the cell feature vector represents the matrix features corresponding to the structural feature matrix.

[0010] According to another embodiment of the present invention, a cell classification device for a brain image is provided, the device comprising:

[0011] A module for acquiring images of cells to be classified, used for acquiring at least two images of cells to be classified in a target brain image;

[0012] a cell feature vector determination module, configured to extract structural features of each cell image to be classified, obtain a structural feature matrix, and determine the cell feature vector of the cell image to be classified based on the structural feature matrix;

[0013] a cell classification result determination module, configured to determine a cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors corresponding to each cell image to be classified;

[0014] The structural feature matrix represents the structural information of cells in the cell image to be classified, and the cell feature vector represents the matrix features corresponding to the structural feature matrix.

[0015] According to another embodiment of the present invention, an electronic device is provided, the electronic device including:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor so that the at least one processor can perform the cell classification method for brain images according to any embodiment of the present invention.

[0019] According to another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium stores computer instructions. The computer instructions are used to enable a processor to implement the cell classification method for brain images according to any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention extracts structural features of each cell image to be classified in the target brain image to obtain a structural feature matrix, and determines the cell feature vector of the cell image to be classified based on the structural feature matrix. According to the cell feature vector corresponding to each cell image to be classified, the cell classification result of each cell image to be classified in the target brain image is determined. The structural information of brain cells can reflect the characteristics of the category to which the brain cells belong, thereby solving the problem that traditional cell classification methods have poor classification effect on brain cell systems and improving the cell classification accuracy and classification efficiency of brain images.

[0021] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A flowchart of a method for cell classification of brain images provided by one embodiment of the present invention;

[0024] Figure 2 A schematic diagram of a target brain image before and after segmentation provided by one embodiment of the present invention;

[0025] Figure 3 A schematic diagram of a binary image and a distance transformation matrix provided by one embodiment of the present invention;

[0026] Figure 4 A schematic diagram of a cell image to be classified provided by one embodiment of the present invention;

[0027] Figure 5 A schematic diagram of a Gaussian fuzzy matrix provided by one embodiment of the present invention;

[0028] Figure 6 A flowchart of another method for cell classification of brain images provided by one embodiment of the present invention;

[0029] Figure 7 A schematic diagram of a pseudo color table provided by one embodiment of the present invention;

[0030] Figure 8 A schematic diagram of a pseudo-color brain image provided by one embodiment of the present invention;

[0031] Figure 9 A schematic structural diagram of a cell classification device for brain images provided by one embodiment of the present invention;

[0032] Figure 10 The present invention provides a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "to be classified", "reference", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] Figure 1 This is a flowchart of a method for classifying cells in brain images provided by one embodiment of the present invention. This embodiment is applicable to the case of classifying brain cells in brain images. The method can be performed by a cell classification device for brain images. The cell classification device for brain images can be implemented in the form of hardware and / or software. The cell classification device for brain images can be configured in a terminal device. Figure 1 As shown, the method includes:

[0036] S110 , obtaining at least two images of cells to be classified in the target brain image.

[0037] Specifically, the target brain image is used to represent a brain image containing brain cells. For example, the target brain image can be read from a storage device or collected using microscopic imaging technology, wherein the microscopic imaging technology can be optical microscopic imaging technology, electron microscopic imaging technology, fluorescence microscopic imaging technology, etc. There is no limitation on the method of obtaining the target brain image here, and the specific settings can be customized according to actual needs.

[0038] In an optional embodiment, the target brain image is acquired using mesoscopic imaging technology. Mesoscopic imaging is a scale between the microscopic and macroscopic, with observation ranges ranging from micrometers to centimeters. It offers advantages such as a wide field of view, high resolution, and high imaging speed. Mesoscopic imaging technology typically uses nuclear or cytoplasmic staining, and the imaging results of these two staining methods are displayed as grayscale images.

[0039] For example, from the perspective of imaging dimension, the target brain image can be a three-dimensional image or a two-dimensional image, and the imaging dimension of the target brain image is not limited here.

[0040] Specifically, the cell image to be classified is used to represent an image in the target brain image that contains only one cell. In an optional embodiment, the cell image to be classified is obtained by manually annotating the target brain image.

[0041] In another optional embodiment, obtaining at least two cell images to be classified from a target brain image includes performing cell segmentation on the target brain image to obtain the at least two cell images to be classified. Exemplary cell segmentation algorithms include, but are not limited to, the Cellpose algorithm, the Omnipose algorithm, a deformable model, a genetic algorithm, or the Cell-DETR algorithm. The cell segmentation algorithm is not limited herein and can be customized based on actual needs.

[0042] Figure 2 This is a schematic diagram of a target brain image before and after segmentation provided by one embodiment of the present invention. Specifically, Figure 2 The left image is the target brain image before segmentation, and the right image is the image obtained after cell segmentation of the target brain image using the Cellpose algorithm. Each color area in the right image represents a cell, and pixels with the same value belong to the same cell and together constitute the image area where the cell image is located.

[0043] Specifically, the cell size required by the Cellpose algorithm can be determined based on any brain image in the three-dimensional target brain image. In an alternative embodiment, the cell size is determined based on an intermediate brain image in the target brain image. For example, assuming the target brain image size is 1024*1024*128, it can be understood that 128 1024*1024 pixel brain images are stacked together, with the 64th brain image serving as the intermediate brain image.

[0044] Based on the above embodiment, optionally, before acquiring at least two to-be-classified cell images in the target brain image, the method further includes: performing a data format check on the target brain image. Specifically, using a Python tool, the target brain image is read as a NumPy array and then stored as 16-bit data. The target brain image is then checked to ensure that pixel values ​​within the target brain image are integers between 0 and 65535.

[0045] S120 , for each cell image to be classified, extract structural features of the cell image to be classified to obtain a structural feature matrix, and determine a cell feature vector of the cell image to be classified based on the structural feature matrix.

[0046] In this embodiment, the structural feature matrix represents the structural information of cells in the cell image to be classified, and the cell feature vector represents the matrix features corresponding to the structural feature matrix.

[0047] In an optional embodiment, structural feature extraction is performed on the cell image to be classified to obtain a structural feature matrix, including: using a distance transformation algorithm to determine a distance transformation matrix based on the cell image to be classified; determining a reference feature matrix based on the distance transformation matrix and the cell image to be classified; and determining a structural feature matrix of the cell image to be classified based on the reference feature matrix.

[0048] Specifically, assuming that cells are approximately spherical or round, the image of the cell to be classified is converted into a binary image. A distance transformation algorithm is then used to determine a distance transformation matrix corresponding to the binary image. In this embodiment, the distance matrix parameters in the distance transformation matrix represent the distance from a pixel in the cell image to be classified to the nearest background pixel, where a background pixel is defined as a background pixel in the cell image to be classified.

[0049] Exemplarily, the distance transformation algorithm includes but is not limited to the Euclidean distance transform (EDT) algorithm, the Manhattan distance algorithm, the Chebyshev distance algorithm, and the chessboard distance algorithm. The distance transformation algorithm used is not limited here, and can be defined and set according to actual needs.

[0050] Figure 3 A schematic diagram of a binary image and a distance transformation matrix provided by an embodiment of the present invention. Specifically, Figure 3 The left image in the figure represents a binary image, which contains only pixels with pixel values ​​of 0 and pixels with pixel values ​​of 1. The right image represents the distance transformation matrix corresponding to the binary image obtained by using the chessboard distance algorithm.

[0051] In an optional embodiment, the matrix rows and columns of the reference feature matrix respectively represent each distance matrix parameter in the distance transformation matrix and each pixel value in the cell image to be classified, and the structure matrix parameters in the reference feature matrix represent the number of pixels in the cell image to be classified corresponding to the distance matrix parameters and pixel values.

[0052] by Figure 3 For example, the reference feature matrix contains a matrix row corresponding to distance 0, a matrix row corresponding to distance 1, a matrix row corresponding to distance 2, and a matrix row corresponding to distance 3. Assume that the cell image to be classified contains pixel values ​​0-pixel values ​​5, where the 8 pixels corresponding to distance 0 in the image to be classified include 1 pixel with a pixel value of 3, 5 pixels with a pixel value of 4, and 2 pixels with a pixel value of 5, and do not contain pixels with other pixel values. Then the matrix row corresponding to distance 0 in the reference feature matrix is ​​[0 0 0 1 5 2].

[0053] In an optional embodiment, determining the structural feature matrix of the cell image to be classified based on the reference feature matrix includes: using the reference feature matrix as the structural feature matrix of the cell image to be classified.

[0054] In another optional embodiment, the structural feature matrix of the cell image to be classified is determined based on the reference feature matrix, including: obtaining target matrix parameters corresponding to the matrix rows corresponding to the distance matrix parameters being zero and the matrix columns corresponding to the pixel values ​​being preset values ​​in the reference feature matrix; deleting the target matrix parameters from the reference feature matrix to obtain the structural feature matrix of the cell image to be classified.

[0055] It is understood that the meanings of the matrix rows and columns of the reference feature matrix are interchangeable. That is, in another optional embodiment, the matrix rows and columns of the reference feature matrix represent each pixel value in the cell image to be classified and each matrix parameter in the distance transformation matrix. Accordingly, in this embodiment, determining the structural feature matrix of the cell image to be classified based on the reference feature matrix includes: obtaining target matrix parameters corresponding to matrix rows corresponding to pixel values ​​of preset values ​​and matrix columns corresponding to distance matrix parameters of zero in the reference feature matrix; and deleting the target matrix parameters from the reference feature matrix to obtain the structural feature matrix of the cell image to be classified.

[0056] Specifically, the preset value is used to represent the background pixel value of the target brain image. For example, the preset value may be 0. There is no limitation on the preset value here and it can be customized according to actual needs.

[0057] Specifically, the target matrix parameter is used to represent the number of pixels whose pixel values ​​in the cell image to be classified are preset values ​​and are defined as background pixels in the binary image.

[0058] Affected by the segmentation effect, there are large differences in the target matrix parameters in different cell images to be classified. However, the target matrix parameters as image background information do not constitute the structural information of the cells. The benefit of deleting the target matrix parameters is that it can improve the accuracy of the structural information of the cells represented by the structural feature matrix and reduce the noise interference caused by the target matrix parameters to the structural feature matrix, thereby further improving the accuracy of cell classification.

[0059] In an optional embodiment, determining a cell feature vector of a cell image to be classified based on a structural feature matrix includes: performing Gaussian blur processing on the structural feature matrix to obtain a Gaussian blur matrix; performing feature extraction on the Gaussian blur matrix to obtain a cell feature vector of the cell image to be classified; wherein the cell feature vector includes at least two cell features selected from the group consisting of the mean, barycentric coordinates, and standard deviation of the Gaussian blur matrix, the mean, barycentric coordinates, and standard deviation of the X-partial derivative matrix of the Gaussian blur matrix, and the mean, barycentric coordinates, and standard deviation of the Y-partial derivative matrix of the Gaussian blur matrix.

[0060] For example, the standard deviation used in the Gaussian blur processing may be 2, which is only used as an example and is not intended to be limiting.

[0061] Figure 4 A schematic diagram of a cell image to be classified provided by one embodiment of the present invention. Specifically, Figure 4 The middle arrows point to the cell images to be classified of four cells with similar structural information in the target brain image. Figure 5 A schematic diagram of a Gaussian fuzzy matrix provided by an embodiment of the present invention. Specifically, Figure 5 The four Gaussian blur matrices in are based on Figure 4 The four cells shown correspond to the cell images to be classified, Figure 5 The Gaussian fuzzy matrix in is obtained by performing Gaussian fuzzy processing on the structural feature matrix without deleting the target matrix parameters. Figure 5 It can be seen that Figure 4 The Gaussian fuzzy matrices corresponding to the four cells pointed out in FIG also have similar distribution characteristics.

[0062] S130 , determining a cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors corresponding to each cell image to be classified.

[0063] Specifically, the cell classification result is used to represent the relative classification result of the cell image to be classified in the target brain image, which is different from other cell images to be classified.

[0064] In an optional embodiment, the cell classification result of each cell image to be classified in the target brain image is determined based on the cell feature vectors corresponding to each cell image to be classified, including: inputting the cell feature vectors corresponding to each cell image to be classified into a pre-trained cell classification model, and obtaining the output cell classification result of each cell image to be classified in the target brain image.

[0065] Exemplarily, the model architecture of the cell classification model includes but is not limited to Transformer networks, CNN networks (Convolutional Neural Networks), DBN networks (Deep Belief Networks), RNN networks (Recursive Neural Networks) or generative adversarial networks, etc. The model architecture of the cell classification model is not limited here and can be customized according to actual needs.

[0066] The technical solution of this embodiment extracts structural features of each cell image to be classified in the target brain image to obtain a structural feature matrix, and determines the cell feature vector of the cell image to be classified based on the structural feature matrix. According to the cell feature vector corresponding to each cell image to be classified, the cell classification result of each cell image to be classified in the target brain image is determined. This utilizes the fact that the structural information of brain cells can reflect the characteristics of the category to which the brain cells belong, solves the problem that traditional cell classification methods have poor classification effect on brain cell systems, and improves the cell classification accuracy and classification efficiency of brain images.

[0067] Figure 6 This is a flowchart of another method for cell classification of brain images provided by one embodiment of the present invention. This embodiment further refines the above embodiment's "determining the cell classification result of each cell image to be classified in the target brain image based on the cell feature vectors corresponding to each cell image to be classified." Figure 6 As shown, the method includes:

[0068] S210 , obtaining at least two images of cells to be classified in a target brain image.

[0069] S220 , for each cell image to be classified, extract structural features of the cell image to be classified to obtain a structural feature matrix, and determine a cell feature vector of the cell image to be classified based on the structural feature matrix.

[0070] S210-S220 in this embodiment are similar to those in the above embodiment. Figure 1 S110 - S120 shown correspond to the same or similar steps and are not described again in detail in this embodiment.

[0071] S230 , using a feature dimensionality reduction algorithm, and determining the dimensionality reduction features corresponding to each cell image to be classified according to the cell feature vectors corresponding to each cell image to be classified.

[0072] Exemplary feature dimensionality reduction algorithms include, but are not limited to, principal component analysis algorithm, linear discriminant algorithm, and UMAP (Uniform Manifold Approximation and Projection) algorithm, etc. The feature dimensionality reduction algorithm used is not limited here, and can be customized according to actual needs.

[0073] Specifically, the cell feature vectors corresponding to each cell image to be classified are used as input data for the feature dimensionality reduction algorithm, and the feature dimensionality reduction algorithm outputs reduced dimensionality features corresponding to each cell image to be classified. For example, the reduced dimensionality features may be one-dimensional, two-dimensional, or three-dimensional features. The reduced dimensionality of the reduced dimensionality features is not limited here and can be customized according to actual needs.

[0074] S240: Determine a cell classification result of each cell image to be classified in the target brain image based on at least two dimensionality reduction features.

[0075] In an optional embodiment, the cell classification result of each cell image to be classified in the target brain image is determined based on at least two dimensionality reduction features, including: for each cell image to be classified, the cell classification result of the cell image to be classified in the target brain image is determined based on the dimensionality reduction feature of the cell image to be classified and at least one preset feature range.

[0076] In a specific embodiment, if the dimensionality reduction features of the cell image to be classified satisfy a first preset feature range, the cell classification result of the cell image to be classified in the target brain image is set to the first type of cell corresponding to the first preset feature range. If the dimensionality reduction features of the cell image to be classified satisfy a second preset feature range, the cell classification result of the cell image to be classified in the target brain image is set to the second type of cell corresponding to the second preset feature range. Similarly, if the dimensionality reduction features of the cell image to be classified do not satisfy all preset feature ranges, the cell classification result of the cell image to be classified in the target brain image is set to other types of cells other than the above-mentioned types of cells. Specifically, the preset feature range is used to represent the feature range corresponding to a type of cell, and there is no overlapping feature value between each preset feature range.

[0077] In another optional embodiment, determining the cell classification result of each cell image to be classified in the target brain image based on at least two dimensionality reduction features includes: performing a clustering operation on the at least two dimensionality reduction features to obtain the cell classification result of each cell image to be classified in the target brain image.

[0078] Exemplarily, the clustering algorithms used in the clustering operation include but are not limited to the K-means clustering algorithm, the Mean-Shift clustering algorithm, the density-based noise application space clustering algorithm, and the Gaussian mixture model expectation maximization clustering algorithm, etc. The clustering algorithm is not limited here and can be defined and set according to actual needs.

[0079] Based on the above embodiment, optionally, the method further includes: if there is a dimensionality reduction feature less than zero among the at least two dimensionality reduction features, determining a bias value based on the minimum dimensionality reduction feature among the at least two dimensionality reduction features, and determining at least two corrected dimensionality reduction features based on the at least two dimensionality reduction features and the bias value; for each cell image to be classified, determining a pseudo-color value corresponding to the cell image to be classified based on the corrected dimensionality reduction feature corresponding to the cell image to be classified; and determining a pseudo-color brain image corresponding to the target brain image based on the at least two pseudo-color values.

[0080] Specifically, the determined bias value makes the sum of the features corresponding to the minimum dimensionality reduction feature greater than or equal to 0. For example, assuming that the minimum dimensionality reduction feature is -10, the determined bias value may be 10 or 20.

[0081] In an optional embodiment, determining at least two modified dimensionality reduction features based on at least two dimensionality reduction features and a bias value includes: if the dimensionality reduction feature is a three-dimensional feature, using the sum of the features corresponding to each dimensionality reduction feature and the bias value as the modified dimensionality reduction feature. Accordingly, determining a pseudo-color value corresponding to the cell image to be classified based on the modified dimensionality reduction features corresponding to the cell image to be classified includes: using the modified dimensionality reduction features corresponding to the cell image to be classified as the pseudo-color value corresponding to the cell image to be classified.

[0082] In another specific embodiment, at least two revised dimensionality reduction features are determined based on at least two dimensionality reduction features and a bias value, including: if the revised dimensionality reduction feature is not a three-dimensional feature, then the feature sum corresponding to each dimensionality reduction feature and the bias value is used as the biased dimensionality reduction feature, and the at least two biased dimensionality reduction features are normalized to obtain at least two revised dimensionality reduction features.

[0083] Accordingly, determining the pseudo-color value corresponding to the cell image to be classified based on the corrected dimensionality reduction features corresponding to the cell image to be classified includes: searching a pseudo-color table to obtain the pseudo-color value corresponding to the cell image to be classified based on the corrected dimensionality reduction features corresponding to the cell image to be classified. In this embodiment, the normalized feature range is the same as the amount of color mapping data contained in the pseudo-color table.

[0084] Figure 7This is a schematic diagram of a pseudo color table provided by an embodiment of the present invention. Specifically, the pseudo color table contains pseudo color values ​​corresponding to 256 characteristic values ​​from 0 to 255.

[0085] Figure 8 A schematic diagram of a pseudo-color brain image provided by one embodiment of the present invention. Figure 8 The upper image in the figure represents the original brain image, and the image in the box in the original brain image represents the target brain image cropped from the original brain image. Figure 8 The lower left image in the figure shows a two-dimensional pseudo-color brain image, and the lower right image shows a three-dimensional pseudo-color brain image.

[0086] The advantage of this setting is that the pseudo-color value is used as the unique identifier of the cell classification result, which realizes the visualization of the cell classification results of the target brain image and reflects the distribution of different types of cells in the target brain image.

[0087] The technical solution of this embodiment adopts a feature dimensionality reduction algorithm to determine the dimensionality reduction features corresponding to each cell image to be classified based on the cell feature vectors corresponding to each cell image to be classified. Based on at least two dimensionality reduction features, the cell classification result of each cell image to be classified in the target brain image is determined. Compared with the machine learning model, this solves the problem of poor interpretability of cell classification results, avoids relying on a large number of annotated images, and further improves the classification efficiency of cell classification while ensuring the accuracy of cell classification.

[0088] It should be noted that in the technical solution of the present invention, the collection, use, storage, sharing and transfer of user personal information involved are in compliance with the provisions of relevant laws and regulations, and it is necessary to inform the user and obtain the user's consent or authorization. When applicable, the user's personal information is de-identified and / or anonymized and / or encrypted.

[0089] The following is an embodiment of a cell classification device for brain images provided by an embodiment of the present invention. This device and the cell classification method for brain images of the above-mentioned embodiment belong to the same inventive concept. For details not fully described in the embodiment of the cell classification device for brain images, reference can be made to the content regarding the cell classification method for brain images in the above-mentioned embodiment.

[0090] Figure 9 This is a schematic diagram of the structure of a cell classification device for brain images provided by one embodiment of the present invention. Figure 9 As shown, the device includes: a cell image acquisition module 310 for to-be-classified cells, a cell feature vector determination module 320 and a cell classification result determination module 330 .

[0091] The to-be-classified cell image acquisition module 310 is configured to acquire at least two to-be-classified cell images in the target brain image;

[0092] The cell feature vector determination module 320 is used to extract the structural features of each cell image to be classified, obtain a structural feature matrix, and determine the cell feature vector of the cell image to be classified based on the structural feature matrix;

[0093] The cell classification result determination module 330 is used to determine the cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors corresponding to each cell image to be classified;

[0094] The structural feature matrix represents the structural information of cells in the cell image to be classified, and the cell feature vector represents the matrix features corresponding to the structural feature matrix.

[0095] The technical solution of this embodiment extracts structural features of each cell image to be classified in the target brain image to obtain a structural feature matrix, and determines the cell feature vector of the cell image to be classified based on the structural feature matrix. According to the cell feature vector corresponding to each cell image to be classified, the cell classification result of each cell image to be classified in the target brain image is determined. This utilizes the fact that the structural information of brain cells can reflect the characteristics of the category to which the brain cells belong, solves the problem that traditional cell classification methods have poor classification effect on brain cell systems, and improves the cell classification accuracy and classification efficiency of brain images.

[0096] In an optional embodiment, the cell feature vector determination module 320 includes:

[0097] A distance transformation matrix determination unit is used to determine a distance transformation matrix based on the cell image to be classified using a distance transformation algorithm; wherein the distance matrix parameters in the distance transformation matrix represent the distance between a pixel point in the cell image to be classified and the nearest background pixel point;

[0098] a reference feature matrix determination unit, configured to determine a reference feature matrix based on the distance transformation matrix and the cell image to be classified; wherein the matrix rows and columns of the reference feature matrix respectively represent each distance matrix parameter in the distance transformation matrix and each pixel value in the cell image to be classified, and the structure matrix parameters in the reference feature matrix represent the number of pixels in the cell image to be classified corresponding to the distance matrix parameters and pixel values;

[0099] The structural feature matrix determining unit is used to determine the structural feature matrix of the cell image to be classified according to the reference feature matrix.

[0100] In an optional embodiment, the structural feature matrix determination unit is specifically configured to:

[0101] Obtain target matrix parameters corresponding to matrix rows corresponding to distance matrix parameters being zero and matrix columns corresponding to pixel values ​​corresponding to preset values ​​in the reference feature matrix;

[0102] The target matrix parameters are deleted from the reference feature matrix to obtain the structural feature matrix of the cell image to be classified.

[0103] In an optional embodiment, the cell feature vector determination module 320 includes:

[0104] A cell feature vector determination unit is used to perform Gaussian fuzzy processing on the structural feature matrix to obtain a Gaussian fuzzy matrix;

[0105] Perform feature extraction on the Gaussian fuzzy matrix to obtain the cell feature vector of the cell image to be classified;

[0106] The cell feature vector includes at least two cell features among the mean, barycentric coordinates, and standard deviation of the Gaussian fuzzy matrix, the mean, barycentric coordinates, and standard deviation of the X-partial derivative matrix of the Gaussian fuzzy matrix, and the mean, barycentric coordinates, and standard deviation of the Y-partial derivative matrix of the Gaussian fuzzy matrix.

[0107] In an optional embodiment, the cell classification result determination module 330 includes:

[0108] A dimensionality reduction feature determination unit is used to determine the dimensionality reduction features corresponding to each cell image to be classified based on the cell feature vectors corresponding to each cell image to be classified using a feature dimensionality reduction algorithm;

[0109] The cell classification result determination unit is used to determine the cell classification result of each cell image to be classified in the target brain image based on at least two dimensionality reduction features.

[0110] In an optional embodiment, the cell classification result determination unit is specifically configured to:

[0111] For each cell image to be classified, determining a cell classification result of the cell image to be classified in the target brain image based on the dimensionality reduction feature of the cell image to be classified and at least one preset feature range; or,

[0112] When the number of the dimensionality reduction features is at least two, a clustering operation is performed on the at least two dimensionality reduction features to obtain a cell classification result of each cell image to be classified in the target brain image.

[0113] In an optional embodiment, the device further comprises:

[0114] a pseudo-color brain image determination module, configured to determine a bias value based on a minimum of the at least two dimensionality reduction features if a dimensionality reduction feature is less than zero among the at least two dimensionality reduction features, and determine at least two corrected dimensionality reduction features based on the at least two dimensionality reduction features and the bias value;

[0115] For each cell image to be classified, determining a pseudo color value corresponding to the cell image to be classified according to the corrected dimensionality reduction features corresponding to the cell image to be classified;

[0116] A pseudo-color brain image corresponding to the target brain image is determined according to the at least two pseudo-color values.

[0117] The cell classification device for brain images provided in the embodiments of the present invention can execute the cell classification method for brain images provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0118] Figure 10 A schematic diagram of the structure of an electronic device provided for one embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0119] like Figure 10 As shown, the electronic device 10 includes at least one processor 11 and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12 and a random access memory (RAM) 13. The memory stores a computer program that can be executed by the at least one processor 11. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0120] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information or data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0121] The processor 11 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the cell classification method for brain images provided in the above embodiments.

[0122] In some embodiments, the cell classification method for brain images provided in the above embodiments can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps in the cell classification method for brain images described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the cell classification method for brain images in any other appropriate manner (for example, by means of firmware).

[0123] Various embodiments of the systems and techniques described herein can be implemented in the following systems or combinations thereof: digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard parts (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0124] Computer programs for implementing the brain image cell classification methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that, when executed by the processor, the computer programs implement the functions / operations specified in the flowcharts and / or block diagrams. The computer programs can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0125] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable storage medium. Examples of machine-readable storage media can include an electrical connection based on at least one line, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0126] To provide interaction with a user, the systems and techniques described herein can be implemented on a terminal device having: a display device (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the user can provide input to the terminal device. Other types of devices can also provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0127] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0128] A computing system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. The client-server relationship arises through computer programs running on the respective computers and establishing a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within a cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and virtual private server (VPS) services.

[0129] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0130] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for cell classification of brain images, characterized in that: include: Acquire at least two cell images to be classified in a target brain image; For each cell image to be classified, extracting structural features of the cell image to be classified to obtain a structural feature matrix, and determining a cell feature vector of the cell image to be classified based on the structural feature matrix; determining a cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors respectively corresponding to each cell image to be classified; The structural feature matrix represents the structural information of cells in the cell image to be classified, and the cell feature vector represents the matrix features corresponding to the structural feature matrix.

2. The method according to claim 1, characterized in that The step of extracting structural features from the cell image to be classified to obtain a structural feature matrix includes: A distance transformation algorithm is used to determine a distance transformation matrix based on the cell image to be classified; wherein the distance matrix parameters in the distance transformation matrix represent the distance between a pixel point in the cell image to be classified and the nearest background pixel point; Determine a reference feature matrix based on the distance transformation matrix and the cell image to be classified; wherein the matrix rows and columns of the reference feature matrix respectively represent each distance matrix parameter in the distance transformation matrix and each pixel value in the cell image to be classified, and the structure matrix parameters in the reference feature matrix represent the number of pixels in the cell image to be classified corresponding to the distance matrix parameters and the pixel values; The structural feature matrix of the cell image to be classified is determined according to the reference feature matrix.

3. The method according to claim 2, characterized in that Determining the structural feature matrix of the cell image to be classified according to the reference feature matrix includes: Obtain target matrix parameters corresponding to matrix rows corresponding to distance matrix parameters being zero and matrix columns corresponding to pixel values ​​being preset values ​​in the reference feature matrix; The target matrix parameters are deleted from the reference feature matrix to obtain the structural feature matrix of the cell image to be classified.

4. The method according to claim 1, wherein Determining the cell feature vector of the cell image to be classified according to the structural feature matrix includes: Performing Gaussian fuzzy processing on the structural feature matrix to obtain a Gaussian fuzzy matrix; Performing feature extraction on the Gaussian fuzzy matrix to obtain a cell feature vector of the cell image to be classified; The cell feature vector includes at least two cell features among the mean, barycentric coordinates, and standard deviation of the Gaussian fuzzy matrix, the mean, barycentric coordinates, and standard deviation of the X partial derivative matrix of the Gaussian fuzzy matrix, and the mean, barycentric coordinates, and standard deviation of the Y partial derivative matrix of the Gaussian fuzzy matrix.

5. The method according to claim 1, wherein Determining the cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors respectively corresponding to each cell image to be classified includes: Using a feature dimensionality reduction algorithm, according to the cell feature vectors corresponding to each of the cell images to be classified, the dimensionality reduction features corresponding to each of the cell images to be classified are determined; A cell classification result of each to-be-classified cell image in the target brain image is determined based on at least two dimensionality reduction features.

6. The method according to claim 5, characterized in that Determining the cell classification result of each cell image to be classified in the target brain image based on at least two dimensionality reduction features includes: For each cell image to be classified, determining a cell classification result of the cell image to be classified in the target brain image according to the dimensionality reduction feature of the cell image to be classified and at least one preset feature range; or, A clustering operation is performed on the at least two dimensionality reduction features to obtain a cell classification result of each cell image to be classified in the target brain image.

7. The method according to claim 5, characterized in that The method further comprises: If there is a dimensionality reduction feature less than zero among the at least two dimensionality reduction features, determining a bias value according to the minimum dimensionality reduction feature among the at least two dimensionality reduction features, and determining at least two revised dimensionality reduction features according to the at least two dimensionality reduction features and the bias value; For each cell image to be classified, determining a pseudo color value corresponding to the cell image to be classified according to the corrected dimensionality reduction feature corresponding to the cell image to be classified; A pseudo-color brain image corresponding to the target brain image is determined according to the at least two pseudo-color values.

8. A cell classification device for brain images, characterized in that: include: A module for acquiring images of cells to be classified, used for acquiring at least two images of cells to be classified in a target brain image; a cell feature vector determination module, configured to extract structural features of each cell image to be classified, obtain a structural feature matrix, and determine the cell feature vector of the cell image to be classified based on the structural feature matrix; a cell classification result determination module, configured to determine a cell classification result of each cell image to be classified in the target brain image according to the cell feature vectors corresponding to each cell image to be classified; The structural feature matrix represents the structural information of cells in the cell image to be classified, and the cell feature vector represents the matrix features corresponding to the structural feature matrix.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the cell classification method for brain images according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the cell classification method for brain images according to any one of claims 1 to 7 when executed.

Images