A method, device, medium and product for identifying a prostate cancer subtype

Abstract

The application discloses a prostate cancer subtype identification method, device, medium and product, and relates to the technical field of medical image processing. The method comprises the following steps: performing feature extraction, feature dimension reduction and feature clustering on a plurality of sample image sets obtained by imaging a plurality of samples of prostates by using multi-parameter magnetic resonance imaging, obtaining clustering results, one clustering result representing one prostate cancer subtype, further establishing a subtype identification model, and determining the subtype of prostate cancer in a prostate to be identified by using the subtype identification model, so that the identification process of the prostate cancer subtype can be automatically completed based on the multi-parameter magnetic resonance imaging. Since the multi-parameter magnetic resonance imaging provides rich prostate tissue structure and function information, the accuracy and stability of the subtype identification can be improved, and the method can effectively assist doctors in accurately diagnosing prostate cancer.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method, device, medium, and product for identifying prostate cancer subtypes based on multi-parameter magnetic resonance imaging. Background Technology

[0002] Prostate cancer is one of the most common malignant tumors in men, and early and accurate diagnosis is crucial for its treatment and prognosis. Accurate diagnosis requires identifying the prostate cancer subtype, which assists doctors in making a precise diagnosis. However, traditional methods for identifying prostate cancer subtypes rely heavily on doctors' visual interpretation and experience, resulting in high subjectivity and low accuracy. Multiparameter magnetic resonance imaging (mpMRI), due to its ability to provide rich information on prostate tissue structure and function, has become an important diagnostic tool for prostate cancer. Therefore, developing an automated prostate cancer subtype identification method based on mpMRI is of great significance for achieving accurate diagnosis of prostate cancer. Summary of the Invention

[0003] The purpose of this invention is to provide a method, device, medium, and product for identifying prostate cancer subtypes, which can automatically identify prostate cancer subtypes and assist doctors in accurately diagnosing prostate cancer.

[0004] To achieve the above objectives, the present invention provides the following solution:

[0005] A method for identifying prostate cancer subtypes includes:

[0006] Multiple sample image sets are obtained by imaging multiple prostate samples using multi-parameter magnetic resonance imaging; the prostate samples have prostate cancer; each prostate sample corresponds to one sample image set, and the sample image set includes sample T1 weighted images, sample T2 weighted images, and sample diffusion weighted images;

[0007] Feature extraction, feature dimensionality reduction, and feature clustering are performed on all the sample image sets to obtain clustering results; each clustering result represents a prostate cancer subtype.

[0008] A subtype identification model was established based on the dimensionality-reduced sample features of all the prostate samples and the prostate cancer subtypes.

[0009] A set of images to be identified is obtained by imaging the prostate gland using multi-parameter magnetic resonance imaging; feature extraction and dimensionality reduction are performed on the set of images to be identified to obtain features to be identified; using the features to be identified as input, the subtype of prostate cancer in the prostate gland is determined by the subtype identification model; the set of images to be identified includes T1-weighted images, T2-weighted images, and diffusion-weighted images.

[0010] In some embodiments, feature extraction, feature dimensionality reduction, and feature clustering are performed on all the sample image sets to obtain clustering results, specifically including:

[0011] For each of the sample image sets, the sample T1-weighted image, the sample T2-weighted image, and the sample diffusion-weighted image are registered to obtain registered sample T1-weighted images, registered sample T2-weighted images, and registered sample diffusion-weighted images; lesion segmentation is performed on the registered sample T2-weighted image to obtain the prostate cancer contour; feature extraction is performed based on the registered sample T1-weighted image, the registered sample T2-weighted image, the registered sample diffusion-weighted image, and the prostate cancer contour to obtain the sample features of the sample image set; the sample features include the radiomics features of the registered sample T1-weighted image, the registered sample T2-weighted image, and the registered sample diffusion-weighted image;

[0012] The sample features of all the sample image sets are subjected to feature dimensionality reduction to obtain the dimensionality-reduced sample features of the sample image sets;

[0013] Feature clustering is performed on the dimensionality-reduced sample features of all the sample image sets to obtain the clustering results.

[0014] In some embodiments, feature extraction is performed based on the registered sample T1-weighted image, the registered sample T2-weighted image, the registered sample diffusion-weighted image, and the prostate cancer contour to obtain sample features of the sample image set, specifically including:

[0015] Using the registered sample T1 weighted image, the registered sample T2 weighted image, the registered sample diffusion weighted image, and the prostate cancer contour as input, feature extraction is performed using Pyradiomics software to obtain the sample features of the sample image set.

[0016] In some embodiments, the radiomics features include 3D shape features, first-order histogram features, second-order texture features, and higher-order features obtained through Laplacian Gaussian transform and wavelet transform.

[0017] In some embodiments, feature dimensionality reduction is performed on the sample features of all the sample image sets to obtain the dimensionality-reduced sample features of the sample image sets, specifically including:

[0018] Using the sample features of all the sample image sets as input, the UMAP dimensionality reduction algorithm is used to perform feature dimensionality reduction to obtain the dimensionality-reduced sample features of the sample image sets.

[0019] In some embodiments, feature clustering is performed on the dimensionality-reduced sample features of all the sample image sets to obtain clustering results, specifically including:

[0020] Using the dimensionality-reduced sample features of all the sample image sets as input, various multi-omics clustering algorithms are used to perform feature clustering to obtain the preliminary clustering results corresponding to each of the multi-omics clustering algorithms;

[0021] A consensus matrix is ​​constructed based on the preliminary clustering results corresponding to all the aforementioned multi-omics clustering algorithms; the consensus matrix includes the probability that any two prostate samples are assigned to the same cluster.

[0022] Based on the consistency matrix, the clustering results are determined using a hierarchical clustering algorithm.

[0023] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for identifying prostate cancer subtypes.

[0024] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method for identifying prostate cancer subtypes.

[0025] A computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method for identifying prostate cancer subtypes.

[0026] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0027] This invention provides a method, device, medium, and product for identifying prostate cancer subtypes. It involves feature extraction, feature dimensionality reduction, and feature clustering of multiple sample image sets obtained by imaging multiple prostate samples using multi-parameter magnetic resonance imaging (MRI). Each clustering result represents a prostate cancer subtype. A subtype identification model is then established to determine the subtype of prostate cancer within the prostate to be identified. This allows for the automatic identification of prostate cancer subtypes based on multi-parameter MRI. Since multi-parameter MRI provides rich information on prostate tissue structure and function, it improves the accuracy and stability of subtype identification, effectively assisting doctors in accurately diagnosing prostate cancer. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic flowchart of a method for identifying prostate cancer subtypes provided in Embodiment 1 of the present invention;

[0030] Figure 2 This is a schematic diagram illustrating the principle of a prostate cancer subtype identification method provided in Embodiment 1 of the present invention;

[0031] Figure 3 This is a diagram of the internal structure of a computer device. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] The purpose of this invention is to provide a method, device, medium, and product for identifying prostate cancer subtypes, which can automatically identify prostate cancer subtypes and assist doctors in accurately diagnosing prostate cancer.

[0034] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] Example 1

[0036] like Figure 1and Figure 2 As shown in this embodiment, a method for identifying prostate cancer subtypes includes:

[0037] S1: Obtain multiple sample image sets obtained by imaging multiple prostate samples using multi-parameter magnetic resonance imaging; the prostate samples have prostate cancer; each prostate sample corresponds to one sample image set, and the sample image set includes sample T1 weighted image, sample T2 weighted image and sample diffusion weighted image.

[0038] This embodiment first uses multi-parameter magnetic resonance imaging (MRI) to image the prostates of multiple prostate cancer patients, acquiring MRI images of the prostates. These MRI images include: T1-enhanced images (T1-weighted images), T2-weighted images (T2-weighted images), and diffusion-weighted images. Multi-parameter MRI includes: T1-weighted imaging (T1WI) for acquiring T1-weighted images, which focuses on comparing T1 relaxation times between tissues; T2-weighted imaging (T2WI) for acquiring T2-weighted images, which focuses on comparing T2 relaxation times between tissues; and diffusion-weighted imaging (DWI) for acquiring diffusion-weighted images, which focuses on comparing water molecule diffusion between tissues. In this embodiment, there may be 200 prostate cancer patients. Thus, multiple sample image sets can be obtained by imaging multiple prostate samples with prostate cancer (also known as prostate tumors) using multi-parameter magnetic resonance imaging. Each prostate sample corresponds to one sample image set, which includes sample T1-weighted images, sample T2-weighted images, and sample diffusion-weighted images.

[0039] S2: Perform feature extraction, feature dimensionality reduction, and feature clustering on all the sample image sets to obtain clustering results; one of the clustering results represents a prostate cancer subtype.

[0040] In this embodiment, feature extraction, feature dimensionality reduction, and feature clustering are performed on all sample image sets to obtain clustering results, specifically including:

[0041] (1) For each sample image set, the sample T1 weighted image, sample T2 weighted image and sample diffusion weighted image are registered to obtain the registered sample T1 weighted image, the registered sample T2 weighted image and the registered sample diffusion weighted image; the registered sample T2 weighted image is segmented to obtain the prostate cancer contour; based on the registered sample T1 weighted image, the registered sample T2 weighted image, the registered sample diffusion weighted image and the prostate cancer contour, feature extraction is performed to obtain the sample features of the sample image set; the sample features include the radiomics features of the registered sample T1 weighted image, the radiomics features of the registered sample T2 weighted image and the radiomics features of the registered sample diffusion weighted image.

[0042] The registration process can be affine transformation registration, and the registration algorithm can be any existing algorithm, such as feature-based image registration algorithms. Features can be point features, line features, or surface features. Through the registration process, the registered sample T1-weighted image, the registered sample T2-weighted image, and the registered sample diffusion-weighted image can have the same resolution. Furthermore, the anatomical location of the prostate gland in these three images can correspond. Subsequently, by segmenting the lesion in one registered image, the prostate cancer contour in all three registered images can be determined simultaneously.

[0043] Because T2-weighted images provide clearer tumor boundaries, this embodiment selects the registered T2-weighted images for lesion segmentation. When segmenting the registered T2-weighted images to obtain the prostate cancer contour, the software used can be ITK-SNAP, an open-source medical image processing software obtained from a third-party platform. Other software can also be used, as long as it can perform contour drawing. When using ITK-SNAP for lesion segmentation, the polygon drawing tool is used to manually draw the three-dimensional contour of the prostate cancer. The specific drawing process includes: the operator first locates the layer containing the prostate cancer along the axial direction, and then uses the polygon drawing tool to draw the contour of the prostate cancer along that layer. After each layer is drawn, the complete three-dimensional contour of the prostate cancer is obtained.

[0044] In this embodiment, during feature extraction, the three registered images and the outlined prostate cancer contour can be imported into the third-party software Pyradiomics for radiomics feature extraction. The feature extraction process includes: inputting the registered T1-weighted image, the registered T2-weighted image, the registered diffusion-weighted image, and the outlined prostate cancer contour; and through analysis and calculation by Pyradiomics, outputting the radiomics features of the three registered images. The radiomics features include: 3D shape features, first-order histogram features, second-order texture features, and higher-order features obtained through Laplacian Gaussian transform and wavelet transform.

[0045] Specifically, feature extraction is performed on the registered sample T1-weighted image, registered sample T2-weighted image, registered sample diffusion-weighted image, and prostate cancer contour to obtain sample features of the sample image set. This includes using the registered sample T1-weighted image, registered sample T2-weighted image, registered sample diffusion-weighted image, and prostate cancer contour as input, and performing feature extraction using Pyradiomics software to obtain sample features of the sample image set. These sample features include radiomics features of the registered sample T1-weighted image, radiomics features of the registered sample T2-weighted image, and radiomics features of the registered sample diffusion-weighted image. The radiomics features include 3D shape features, first-order histogram features (i.e., first-order features), and second-order texture features (including gray-level co-occurrence matrix, gray-level...). The high-order features obtained by Laplacian Gaussian transform and wavelet transform include the 3D shape features, first-order histogram features, and second-order texture features of the three registered images obtained by performing Laplacian Gaussian transform on the registered sample T1 weighted image, the registered sample T2 weighted image, the registered sample diffusion weighted image, and the prostate cancer contour, and then extracting features. The high-order features obtained by wavelet transform include the 3D shape features, first-order histogram features, and second-order texture features of the three registered images obtained by performing wavelet transform on the registered sample T1 weighted image, the registered sample T2 weighted image, the registered sample diffusion weighted image, and the prostate cancer contour, and then extracting features.

[0046] (2) Perform feature dimensionality reduction on the sample features of all sample image sets to obtain the dimensionality-reduced sample features of the sample image sets.

[0047] This embodiment performs dimensionality reduction on the multi-parameter features obtained after feature extraction. The dimensionality reduction method used can be the UMAP (Uniform Manifold Approximation and Projection) algorithm. After feature dimensionality reduction, 10 features are retained for each registered image. The feature dimensionality reduction process includes: using the umap package in R statistical analysis software, importing the features of each registered image into the umap package, setting the number of features to be retained to 10, and starting the process to obtain the 10 features retained for each registered image. Of course, this embodiment can also use other feature dimensionality reduction algorithms, and the number of features retained can be greater than 10 according to user needs.

[0048] Specifically, feature dimensionality reduction is performed on the sample features of all sample image sets to obtain the dimensionality-reduced sample features of the sample image sets. This includes: using the sample features of all sample image sets as input, performing feature dimensionality reduction using the UMAP dimensionality reduction algorithm to obtain the dimensionality-reduced sample features of each sample image set. The dimensionality-reduced sample features include the dimensionality-reduced features of the registered sample T1 weighted image, the registered sample T2 weighted image, and the registered sample diffusion weighted image. The types and number of dimensionality-reduced features of the registered sample T1 weighted image, the registered sample T2 weighted image, and the registered sample diffusion weighted image are the same.

[0049] (3) Perform feature clustering on the dimensionality-reduced sample features of all sample image sets to obtain the clustering results.

[0050] This embodiment uses 10 multi-omics clustering algorithms to perform cluster analysis on the dimensionality-reduced features obtained after feature dimensionality reduction. These 10 algorithms include CIMLR, iClusterBayes, MoCluster, COCA, ConsensusClustering, IntNMF, LRAcluster, NEMO, PINSPlus, and SNF. Each of these 10 algorithms yields 10 clustering results. Based on these results, the probability of any pair of prostate cancer samples belonging to the same cluster can be calculated. For example, if 9 clustering results assign the pair to the same cluster, the probability is 0.9. This generates a consensus matrix, which includes the probability that any two prostate cancer samples are assigned to the same cluster. Based on this consensus matrix, a hierarchical clustering algorithm is used to obtain the final clustering results, further dividing prostate cancer into radiomics subtype 1, radiomics subtype 2, ..., radiomics subtype N, where the number of N is the same as the number of final clustering results. It should be noted that this embodiment can also select other numbers of multi-omics clustering algorithms according to user needs.

[0051] Specifically, feature clustering is performed on the dimensionality-reduced sample features of all sample image sets to obtain clustering results. This includes: using the dimensionality-reduced sample features of all sample image sets as input, performing feature clustering using multiple multi-omics clustering algorithms to obtain preliminary clustering results for each multi-omics clustering algorithm; constructing a consistency matrix based on the preliminary clustering results of all multi-omics clustering algorithms, the consistency matrix including the probability that any two sample prostates are assigned to the same cluster; and determining the clustering results using a hierarchical clustering algorithm based on the consistency matrix.

[0052] S3: Establish a subtype identification model based on the dimensionality-reduced sample features of all the prostate samples and the prostate cancer subtypes.

[0053] The dimensionality-reduced sample features of the prostate sample refer to the dimensionality-reduced sample features obtained after feature extraction and dimensionality reduction of the sample image set corresponding to the prostate sample. The prostate cancer subtype of the prostate sample refers to the prostate cancer subtype corresponding to the cluster to which the sample image set to which the prostate sample belongs, determined based on the clustering results.

[0054] In this embodiment, the subtype identification model uses the dimensionality-reduced sample features as predictors (i.e., inputs) and the clustering results (i.e., prostate cancer subtypes) as labels, and is trained based on the traditional K-nearest neighbor machine learning method.

[0055] S4: Obtain the image set to be identified by imaging the prostate gland to be identified using multi-parameter magnetic resonance imaging; perform feature extraction and feature dimensionality reduction on the image set to be identified to obtain the features to be identified; use the features to be identified as input to determine the subtype of prostate cancer in the prostate gland to be identified using the subtype identification model; the image set to be identified includes T1-weighted images, T2-weighted images and diffusion-weighted images.

[0056] The process of feature extraction and dimensionality reduction for the image set to be recognized is the same as that for the sample image set, and will not be repeated here.

[0057] This embodiment provides a novel method for identifying prostate cancer subtypes from an imaging perspective. Previous methods for prostate cancer subtype identification were mostly based on pathological and molecular classifications. This embodiment applies unsupervised clustering analysis to the multi-parameter features of prostate cancer, obtaining radiomics subtypes. This allows for the further construction of a subtype identification model, automatically identifying prostate cancer subtypes—essentially performing automatic classification of prostate cancer images. This can assist physicians in diagnosing prostate cancer and has significant clinical value. This method comprehensively utilizes magnetic resonance imaging information from different modalities, performing multimodal feature fusion to improve the accuracy and stability of subtype identification. Furthermore, this method is applicable to magnetic resonance imaging equipment of different models and parameters, exhibiting strong versatility, practicality, and adaptability.

[0058] Example 2

[0059] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the prostate cancer subtype identification method described in Embodiment 1.

[0060] Example 3

[0061] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the prostate cancer subtype identification method described in Embodiment 1.

[0062] Example 4

[0063] A computer program product includes a computer program that, when executed by a processor, implements the steps of the prostate cancer subtype identification method described in Example 1.

[0064] Example 5

[0065] A computer device, which may be a database, may have an internal structure diagram as shown below. Figure 3As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores pending transactions. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a prostate cancer subtype identification method as described in Embodiment 1.

[0066] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this invention are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0067] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided by this invention may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided by this invention may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0068] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0069] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for identifying prostate cancer subtypes, characterized in that, include: To obtain multiple sample image sets obtained by imaging multiple prostate samples using multi-parameter magnetic resonance imaging; The prostate sample contained prostate cancer. One sample prostate corresponds to one sample image set, the sample image set including sample T1 weighted image, sample T2 weighted image and sample diffusion weighted image; Feature extraction, feature dimensionality reduction, and feature clustering are performed on all the sample image sets to obtain clustering results; each clustering result represents a prostate cancer subtype. A subtype identification model was established based on the dimensionality-reduced sample features of all the prostate samples and the prostate cancer subtypes. A set of images to be identified is obtained by imaging the prostate gland to be identified using multi-parameter magnetic resonance imaging; feature extraction and feature dimensionality reduction are performed on the set of images to be identified to obtain the features to be identified; Using the features to be identified as input, the subtype identification model is used to determine the subtype of prostate cancer in the prostate to be identified; the image set to be identified includes T1-weighted images, T2-weighted images, and diffusion-weighted images.

2. The method for identifying prostate cancer subtypes according to claim 1, characterized in that, Feature extraction, feature dimensionality reduction, and feature clustering are performed on all the sample image sets to obtain the clustering results, specifically including: For each of the sample image sets, the sample T1-weighted image, the sample T2-weighted image, and the sample diffusion-weighted image are registered to obtain registered sample T1-weighted images, registered sample T2-weighted images, and registered sample diffusion-weighted images; lesion segmentation is performed on the registered sample T2-weighted image to obtain the prostate cancer contour; feature extraction is performed based on the registered sample T1-weighted image, the registered sample T2-weighted image, the registered sample diffusion-weighted image, and the prostate cancer contour to obtain the sample features of the sample image set; the sample features include the radiomics features of the registered sample T1-weighted image, the registered sample T2-weighted image, and the registered sample diffusion-weighted image; The sample features of all the sample image sets are subjected to feature dimensionality reduction to obtain the dimensionality-reduced sample features of the sample image sets; Feature clustering is performed on the dimensionality-reduced sample features of all the sample image sets to obtain the clustering results.

3. The method for identifying prostate cancer subtypes according to claim 2, characterized in that, Feature extraction is performed based on the registered sample T1-weighted image, the registered sample T2-weighted image, the registered sample diffusion-weighted image, and the prostate cancer contour to obtain the sample features of the sample image set, specifically including: Using the registered sample T1 weighted image, the registered sample T2 weighted image, the registered sample diffusion weighted image, and the prostate cancer contour as input, feature extraction is performed using Pyradiomics software to obtain the sample features of the sample image set.

4. The method for identifying prostate cancer subtypes according to claim 2, characterized in that, The image omics features include 3D shape features, first-order histogram features, second-order texture features, and higher-order features obtained through Laplacian Gaussian transform and wavelet transform.

5. The method for identifying prostate cancer subtypes according to claim 2, characterized in that, The sample features of all the sample image sets are subjected to feature dimensionality reduction to obtain the dimensionality-reduced sample features of the sample image sets, specifically including: Using the sample features of all the sample image sets as input, the UMAP dimensionality reduction algorithm is used to perform feature dimensionality reduction to obtain the dimensionality-reduced sample features of the sample image sets.

6. The method for identifying prostate cancer subtypes according to claim 2, characterized in that, Feature clustering is performed on the dimensionality-reduced sample features of all the sample image sets to obtain the clustering results, specifically including: Using the dimensionality-reduced sample features of all the sample image sets as input, various multi-omics clustering algorithms are used to perform feature clustering to obtain the preliminary clustering results corresponding to each of the multi-omics clustering algorithms; A consensus matrix is ​​constructed based on the preliminary clustering results corresponding to all the aforementioned multi-omics clustering algorithms; the consensus matrix includes the probability that any two prostate samples are assigned to the same cluster. Based on the consistency matrix, the clustering results are determined using a hierarchical clustering algorithm.

7. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of a method for identifying prostate cancer subtypes according to any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method for identifying prostate cancer subtypes as described in any one of claims 1-6.

9. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method for identifying prostate cancer subtypes as described in any one of claims 1-6.

Images