Radiological image multi-region of interest based tumor heterogeneity assessment method, system, device, and medium

Abstract

The application discloses a kind of based on radiological image multi-interest region's tumor heterogeneity evaluation method, system, equipment and medium, method includes: interest region subvolume presegmentation, extract Radiomics imageomics feature from the subvolume region of presegmentation, using Gaussian mixture model to carry out group level's unsupervised clustering to all the characteristic values obtained, and using Bayesian information criterion BIC to evaluate clustering effect, based on optimal subvolume class number of intratumor, peritumoral and tumor ring, maximum subvolume volume in interest region and the volume of interest region Calculation of tumor heterogeneity index, utilize the tumor heterogeneity index to evaluate tumor heterogeneity.The application is positioned by subvolume segmentation in tumor, peritumoral and tumor ring, extracts the local imageomics feature of multiple interest regions, calculates tumor heterogeneity index based on machine learning unsupervised clustering algorithm on global level, to realize noninvasive, convenient, accurate tumor heterogeneity evaluation.

Description

Technical Field

[0001] This invention belongs to the technical field of artificial intelligence, specifically relating to a method, system, device, and medium for assessing tumor heterogeneity based on multiple regions of interest in radiographic images. Background Technology

[0002] Radiographic imaging, such as computed tomography (CT) or magnetic resonance imaging (MRI), is the most commonly used non-invasive diagnostic tool in clinical practice, used for the diagnosis, staging, monitoring, and prognostic prediction of various cancers. Radiographic images contain morphological and textural information of tumors, providing a comprehensive view for assessing the occurrence and evolution of tumor tissue. Genomic heterogeneity within a tumor can lead to differences in cellular composition or properties, which can be reflected in local or global differences in tumor appearance on radiographic images. Therefore, quantitative features in radiographic images can be used to assess tumor heterogeneity.

[0003] Previous studies have shown that radiomics features or deep learning features based on radiographic images can be used to assess tumor heterogeneity. Current research methods are mainly divided into the following three categories: (1) Based on the assumption that the tumor is uniformly distributed in images, the tumor as a whole is analyzed, and then one or more radiomics features are incorporated into the final model for heterogeneity assessment. (2) The tumor region is divided into high-risk and low-risk sub-regions, and then features are extracted within the sub-regions. A model is built based on the feature set of these sub-regions. (3) First, a 2D image of the largest tumor layer is selected, and the image is divided into several sub-regions based on a sliding window; then, radiomics features are extracted, and tumor sub-regions are generated based on a clustering algorithm; finally, the tumor heterogeneity score is calculated by the area of ​​the largest connected region and the area of ​​the tumor at the largest layer, thereby assessing tumor heterogeneity.

[0004] However, in practice, it has been found that the first type of research method fails to consider the role of local tumor information in assessing tumor heterogeneity because tumors exhibit diverse biological characteristics and phenotypes that vary in their radiographic appearance. The second type of research method assesses heterogeneity based on subregional features within the tumor, considering only local imaging features and lacking global information for overall tumor assessment. The third type of research method, although it combines global and local information to assess tumor heterogeneity, only uses 2D images of the largest tumor layer, failing to fully utilize the rich spatial information of three-dimensional radiographic data and the role of tumor microenvironment information in assessing tumor heterogeneity.

[0005] Therefore, how to provide a tumor heterogeneity assessment technique that considers both the local features of the region of interest and the complexity of the tumor microenvironment and the global features of the spatial information of the region of interest is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] The main objective of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a method, system, device and medium for assessing tumor heterogeneity based on multiple regions of interest in radiographic imaging. By segmenting the tumor into subvolumes, around the tumor and around the tumor ring, local radiomics features of multiple regions of interest are extracted. Based on machine learning unsupervised clustering algorithm, the tumor heterogeneity index is calculated at the global level, thereby achieving non-invasive, convenient and accurate assessment of tumor heterogeneity.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a method for assessing tumor heterogeneity based on multiple regions of interest in radiographic imaging, comprising the following steps:

[0009] After preprocessing the radiographic image containing the tumor region mask, a linear iterative clustering algorithm is used within the subvolume pre-segmentation region to pre-segment the region of interest (MOI) containing pixels into multiple supervoxels. Subvolume pre-segmentation is completed through iterative calculation. The subvolume pre-segmentation region is formed by determining the bounding box based on the size and location of the tumor in the radiographic image. The MOI includes intratumoral, peritumoral, and tumor ring regions.

[0010] As a preferred technical solution, the preprocessing of the radiographic image containing the tumor region mask further includes the following steps:

[0011] Different preprocessing methods are selected based on the modality of the original radiographic images; the original image modalities include CT and MRI; MRI images are preprocessed using standard normalization calculations, as shown in the following formula:

[0012]

[0013] Where x is the pixel intensity value of the original image, x scaled σ represents the pixel intensity value of the MRI image after standard normalization, μ represents the mean pixel intensity value of the original image, and σ represents the standard deviation of the pixel intensity value of the original image.

[0014] The CT images are preprocessed using maximum and minimum value normalization calculations, as shown in the following formula:

[0015]

[0016] Where x s ′ caled x represents the pixel intensity value of the CT image after normalization of the maximum and minimum values. min x represents the minimum pixel intensity value in the original image. max This represents the maximum intensity value of each pixel in the original image.

[0017] Radiomics features are extracted from pre-segmented subvolume regions to obtain feature values; the radiomics features include first-order histogram features and texture features.

[0018] As a preferred technical solution, the tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging extracts Radiomics features from pre-segmented subvolume regions using the open-source radiomics toolkit PyRadiomics, guided by prior clinical knowledge.

[0019] For all the obtained feature values, Gaussian mixture model is used to perform unsupervised clustering at the population level, and the Bayesian information criterion (BIC) is used to evaluate the clustering effect. K is taken as the value with the smallest BIC value. best The optimal number of subvolume categories for tumor subvolume segmentation; where parameter K... best This represents the optimal number of cluster centers;

[0020] As a preferred technical solution, the sub-volume pre-segmentation specifically includes:

[0021] Subvolume pre-segmentation regions are determined using geometric bounding boxes based on the size and location of the tumor in the radiographic images;

[0022] Within the tumor bounding box, a simple linear iterative clustering algorithm is used to segment the region of interest (ROI) with a total of N pixels into n hypervoxels. The specific steps include:

[0023] Initialize k cluster centers, with the distance between adjacent centers being...

[0024] Calculate the gradient value within a cube neighborhood with a side length of 2S, centered on the cluster center. Select the point with the smallest gradient change as the new cluster center. Calculate the distance D between each pixel and the cluster center and re-divide the area into k cluster centers, as shown in the following formula:

[0025]

[0026]

[0027]

[0028] Where d c Representing color distance, expressed as the intensity value I between two pixels. j and I i Difference representation, d s This represents the spatial distance, which is the spatial distance between two pixels in the image. The spatial coordinates of the two pixels are j(a, b). j ,bj ,c j ) and i(a i ,b i ,c i ), a j ,a i ,b j ,b i ,c j ,c i These are the coordinate parameters of the pixel, and the distance D is determined by d. c and d s It consists of two parts, where m is the maximum color distance;

[0029] After an iterative process, the subvolume pre-segmentation is finally completed;

[0030] As a preferred technical solution, the generation of sub-volume specifically includes:

[0031] For all the obtained feature values, Gaussian mixture model is used to perform unsupervised clustering at the population level; when performing unsupervised clustering at the population level, the category K is clustered according to a set value;

[0032] After clustering is completed, the Bayesian Information Criterion (BIC) is used to evaluate the clustering effect. The calculation formula is as follows:

[0033] BIC = -2ln L + a ln n1

[0034] Where a is the total number of parameters in the Gaussian mixture model, n1 is the sample size, L is the likelihood function of the data under the Gaussian mixture model, and K is the value of the minimum BIC value. best The final number of sub-volume categories is used to determine the sub-volume categories.

[0035] Each patient's tumor mask is determined by It consists of one or more categories combined.

[0036] The number of optimal subvolume categories within the tumor, around the tumor, and around the tumor ring is obtained respectively. The number of optimal subvolume categories within the tumor, around the tumor, and around the tumor ring reflects the complexity of the region of interest. The tumor heterogeneity index is calculated based on the number of optimal subvolume categories within the tumor, around the tumor, and around the tumor ring, the maximum subvolume volume in the region of interest, and the volume of the region of interest. The tumor heterogeneity is evaluated using the tumor heterogeneity index.

[0037] As a preferred technical solution, the tumor heterogeneity index is calculated using the following formula:

[0038]

[0039] Where ITSH_index is the tumor heterogeneity index, Ctumor C peri and C ring SubV represents the complexity of the tumor interior, peritumoral region, and tumor ring on radiographic images, respectively; tumor_max SubV peri_max and SubV ring_max V represents the maximum subvolume volume in the region of interest; tumor_total V Peri_total and V ring_total These represent the volumes of the region of interest.

[0040] Secondly, the present invention also provides a tumor heterogeneity assessment system based on multiple regions of interest in radiographic images, which is applied to a tumor heterogeneity assessment method based on multiple regions of interest in radiographic images, including a subvolume pre-segmentation module, a feature extraction module, an unsupervised clustering module, and a computation execution module;

[0041] The subvolume pre-segmentation module is used to process a given radiographic image containing a tumor region mask. After preprocessing the image, the simple linear iterative clustering algorithm SLIC is used to pre-segment the region of interest containing pixels into supervoxels within the subvolume pre-segmentation area. After iterative calculation, the subvolume pre-segmentation is completed.

[0042] The feature extraction module extracts Radiomics features from the pre-segmented sub-volume regions to obtain feature values;

[0043] The unsupervised clustering module performs population-level unsupervised clustering using a Gaussian mixture model on all obtained feature values, and evaluates the clustering effect using the Bayesian Information Criterion (BIC) to generate subvolumes.

[0044] The computation execution module obtains the optimal number of subvolume categories for each of the multiple regions of interest, calculates the tumor heterogeneity index, and obtains an assessment of tumor heterogeneity.

[0045] Thirdly, the present invention provides an electronic device, the electronic device comprising:

[0046] At least one processor; and,

[0047] A memory communicatively connected to the at least one processor; wherein,

[0048] The memory stores computer program instructions that can be executed by the at least one processor to enable the at least one processor to perform the tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging.

[0049] Fourthly, the present invention provides a computer-readable storage medium storing a program that, when executed by a processor, implements the tumor heterogeneity assessment method based on multiple regions of interest in radiographic images.

[0050] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0051] (1) This invention provides a novel method for cutting tumor radiographic images by using a tumor subvolume segmentation method and extracting features from multiple regions within the tumor, around the tumor, and around the tumor ring. This method fully utilizes the information of the tumor and its microenvironment in the radiographic images, thus providing a global assessment of the overall heterogeneity of the tumor.

[0052] (2) This invention uses a Gaussian mixture model to perform unsupervised clustering at the population level and uses the Bayesian Information Criterion (BIC) to evaluate the clustering effect. The complexity of the region of interest is reflected by obtaining the number of optimal sub-volume categories within the tumor, around the tumor, and around the tumor ring. Then, the tumor heterogeneity index is calculated based on the number of optimal sub-volume categories within the tumor, around the tumor, and around the tumor ring, the maximum sub-volume volume in the region of interest, and the volume of the region of interest. Through this tumor heterogeneity index calculation method, combined with global and local information, the rich spatial information of three-dimensional stereoscopic radiographic data is fully utilized, and the role of tumor microenvironment information in tumor heterogeneity assessment is brought into play. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a flowchart of a tumor heterogeneity assessment method based on multiple regions of interest in radiographic images, as described in an embodiment of the present invention.

[0055] Figure 2 This is a schematic diagram of tumor subvolume pre-segmentation and feature extraction in an embodiment of the present invention;

[0056] Figure 3 This is a schematic diagram of subvolume unsupervised clustering in an embodiment of the present invention;

[0057] Figure 4 This is a schematic diagram of the tumor heterogeneity assessment system based on multiple regions of interest in radiographic imaging, as described in an embodiment of the present invention.

[0058] Figure 5 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0059] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0060] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0061] Radiological imaging: a clinical diagnostic technique, such as computed tomography (CT) or magnetic resonance imaging (MRI);

[0062] Region of multiple interest: mainly refers to the areas within the tumor, around the tumor, and around the tumor ring in radiographic images;

[0063] Tumor heterogeneity: Tumor cells exhibit a variety of different biological characteristics and phenotypes, and the assessment of tumor heterogeneity is crucial for improving the effectiveness of tumor treatment.

[0064] Please see Figure 1-3 This embodiment proposes a method for assessing tumor heterogeneity based on multiple regions of interest in radiographic images, comprising the following steps:

[0065] Step 1: Subvolume pre-segmentation of the region of interest, the specific steps are as follows:

[0066] (1-1) Preprocess the image containing the tumor region mask;

[0067] Preferably, different preprocessing methods are selected according to the different modalities of the original radiographic images. The MRI images are preprocessed using standard normalization calculations, as shown in the following formula:

[0068]

[0069] Where x is the pixel intensity value of the original image, x scaled σ represents the pixel intensity value of the MRI image after standard normalization, μ represents the mean pixel intensity value of the original image, and σ represents the standard deviation of the pixel intensity value of the original image.

[0070] The CT images are preprocessed using maximum and minimum value normalization calculations, as shown in the following formula:

[0071]

[0072] Where x s ′ caled x represents the pixel intensity value of the CT image after normalization of the maximum and minimum values. min x represents the minimum pixel intensity value in the original image. max This represents the maximum intensity value of each pixel in the original image.

[0073] (1-2) Within the subvolume pre-segmentation region, a linear iterative clustering algorithm is used to pre-segment the region of interest containing pixels into multiple supervoxels. After iterative calculation, the subvolume pre-segmentation is completed. The subvolume pre-segmentation region is formed by determining the bounding box based on the size and location of the tumor in the radiographic image. The region of interest includes the intratumoral, peritumoral, and tumor ring regions.

[0074] Optionally, the bounding box can be polygonal, including triangles, rectangles, rhombuses, pentagons, and hexagons;

[0075] Optionally, the bounding box can be a rounded polygon.

[0076] Optionally, the bounding box can be circular or elliptical in shape;

[0077] Furthermore, within the region encompassed by the tumor bounding box, a simple linear iterative clustering algorithm is used to segment the region of interest, which has a total of N pixels, into n hypervoxels. Specifically, the steps include:

[0078] (1-2-1) Initialize k cluster centers, with the distance between adjacent centers being...

[0079] (1-2-2) Calculate the gradient value within a cube neighborhood with a side length of 2S and the origin at the calculated cluster center. Select the point with the smallest gradient change as the new cluster center. Calculate the distance D between each pixel and the cluster center and re-divide into k cluster centers. Let any two pixels be j(a, b) and c(d, c). j ,b j ,c j ) and i(a i ,b i ,c i ), as shown in the following formula:

[0080]

[0081]

[0082]

[0083] Where d cColor distance is represented by the difference in intensity values ​​between two pixels, d s Spatial distance, represented by d, is the spatial distance between two pixels in an image. c and d s It consists of two parts, where m is the maximum color distance;

[0084] Optional, the intensity value I between two pixels j and I i Difference representation;

[0085] (1-2-3) After an iterative process, the subvolume pre-segmentation is finally completed.

[0086] Step 2: Local feature extraction, specifically:

[0087] Radiomics features are extracted from pre-segmented subvolume regions to obtain feature values; the radiomics features include first-order histogram features and texture features.

[0088] For example, such as Figure 2 As shown, guided by prior clinical knowledge, the radiomics open-source toolkit PyRadiomics was used to extract Radiomics features from pre-segmented subvolume regions. Each supervoxel contains one pixel. After feature extraction of n supervoxels in the subvolume region, 19 first-order histogram features and 73 texture features were obtained, forming a two-dimensional network.

[0089] Step 3: Unsupervised clustering generates sub-volumes, such as... Figure 3 As shown, it includes the following steps:

[0090] (3-1) For all the obtained feature values, use Gaussian mixture model to perform unsupervised clustering at the population level. When performing unsupervised clustering at the population level, the category K is clustered according to the set value.

[0091] Optionally, the category K takes values ​​in the range K∈[1,5];

[0092] (3-2) For all the obtained feature values, perform unsupervised clustering at the population level using a Gaussian mixture model, and evaluate the clustering effect using the Bayesian Information Criterion (BIC). Take the value of K with the smallest BIC value. best The optimal number of subvolume categories for tumor subvolume segmentation;

[0093] Preferably, the generation of sub-volume specifically involves:

[0094] (3-2-1) For all the obtained feature values, use Gaussian mixture model to perform unsupervised clustering at the population level; when performing unsupervised clustering at the population level, the category K is clustered according to the set value;

[0095] (3-2-2) After clustering is completed, the Bayesian Information Criterion (BIC) is used to evaluate the clustering effect. The calculation formula is as follows:

[0096] BIC = -2ln L + a ln n1

[0097] Where a is the total number of parameters in the Gaussian mixture model, n1 is the sample size, L is the likelihood function of the data under the Gaussian mixture model, and K is the value of the minimum BIC value. best The final number of sub-volume categories is used to determine the sub-volume categories.

[0098] Each patient's tumor mask is determined by It consists of one or more categories combined.

[0099] Step 4: Calculation of tumor heterogeneity index, which includes the following steps:

[0100] (4-1) Obtain the number of optimal sub-volume categories for the tumor interior, tumor peritumoral region, and tumor ring, respectively. The number of optimal sub-volume categories for the tumor interior, tumor peritumoral region, and tumor ring reflects the complexity of the region of interest.

[0101] (4-2) Calculate the tumor heterogeneity index based on the number of optimal subvolume categories within the tumor, around the tumor, and around the tumor ring, the maximum subvolume volume in the region of interest, and the volume of the region of interest, and use the tumor heterogeneity index to evaluate tumor heterogeneity;

[0102] Preferably, the tumor heterogeneity index is calculated as follows:

[0103]

[0104] Where ITSH_index is the tumor heterogeneity index, C tumor C peri and C ring SubV represents the complexity of the tumor interior, peritumoral region, and tumor ring on radiographic images, respectively; tumor_max SubV peri_max and SubV ring_max V represents the maximum subvolume volume in the region of interest; tumor_total V Peri_total and V ring_total These represent the volumes of the region of interest.

[0105] It should be noted that, for the sake of simplicity, the aforementioned method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously.

[0106] Based on the same idea as the tumor heterogeneity assessment method based on multiple regions of interest (MOIs) in radiographic images described in the above embodiments, this invention also provides a tumor heterogeneity assessment system based on MOIs in radiographic images. This system can be used to perform the aforementioned tumor heterogeneity assessment method based on MOIs in radiographic images. For ease of explanation, the schematic diagram of the embodiment of the tumor heterogeneity assessment system based on MOIs in radiographic images only shows the parts relevant to the embodiments of this invention. Those skilled in the art will understand that the illustrated structure does not constitute a limitation on the device, and it may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0107] Please see Figure 4 In another embodiment of this application, a tumor heterogeneity assessment method system 10 based on multiple regions of interest in radiographic images is provided. The system includes a subvolume pre-segmentation module 11, a feature extraction module 12, an unsupervised clustering module 13, and a computation execution module 14.

[0108] The subvolume pre-segmentation module 11 is used to process a given radiographic image containing a tumor region mask. After preprocessing the image, the simple linear iterative clustering algorithm SLIC is used to pre-segment the region of interest containing pixels into supervoxels within the subvolume pre-segmentation area. After iterative calculation, the subvolume pre-segmentation is completed.

[0109] The feature extraction module 12 extracts Radiomics features from the pre-segmented sub-volume regions to obtain feature values;

[0110] The unsupervised clustering module 13 performs unsupervised clustering at the population level using a Gaussian mixture model on all the obtained feature values, and evaluates the clustering effect using the Bayesian Information Criterion (BIC) to generate subvolumes.

[0111] The calculation execution module 14 obtains the number of optimal subvolume categories for each of the multiple regions of interest, calculates the tumor heterogeneity index, and obtains an assessment of tumor heterogeneity.

[0112] It should be noted that the tumor heterogeneity assessment system based on multiple regions of interest in radiographic images of the present invention corresponds one-to-one with the tumor heterogeneity assessment method based on multiple regions of interest in radiographic images of the present invention. The technical features and beneficial effects described in the embodiments of the tumor heterogeneity assessment method based on multiple regions of interest in radiographic images described above are applicable to the embodiments of the tumor heterogeneity assessment method based on multiple regions of interest in radiographic images. For details, please refer to the description in the embodiments of the method of the present invention, which will not be repeated here.

[0113] Furthermore, in the embodiments of the tumor heterogeneity assessment system based on multiple regions of interest in radiographic images described above, the logical division of each program module is merely illustrative. In practical applications, the above functions can be assigned to different program modules as needed, for example, for the sake of corresponding hardware configuration requirements or the convenience of software implementation. That is, the internal structure of the tumor heterogeneity assessment system based on multiple regions of interest in radiographic images can be divided into different program modules to complete all or part of the functions described above.

[0114] Please see Figure 5 In one embodiment, an electronic device is provided for implementing a tumor heterogeneity assessment method based on multiple regions of interest in radiographic images. The electronic device 20 may include a first processor 21, a first memory 22, and a bus, and may also include a computer program stored in the first memory 22 and executable on the first processor 21, such as a tumor heterogeneity assessment program 23 based on multiple regions of interest in radiographic images.

[0115] The first memory 22 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the first memory 22 can be an internal storage unit of the electronic device 20, such as the portable hard drive of the electronic device 20. In other embodiments, the first memory 22 can be an external storage device of the electronic device 20, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 20. Furthermore, the first memory 22 can include both internal storage units and external storage devices of the electronic device 20. The first memory 22 can be used not only to store application software and various types of data installed on the electronic device 20, such as the code of a tumor heterogeneity assessment program 23 based on multiple regions of interest in radiographic images, but also to temporarily store data that has been output or will be output.

[0116] In some embodiments, the first processor 21 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The first processor 21 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the first memory 22 and calls data stored in the first memory 22 to perform various functions of the electronic device 20 and process data.

[0117] Figure 5 Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 5 The structure shown does not constitute a limitation on the electronic device 20, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0118] The tumor heterogeneity assessment program 23 based on multiple regions of interest in radiographic images, stored in the first memory 22 of the electronic device 20, is a combination of multiple instructions that, when run in the first processor 21, can achieve the following:

[0119] Given a radiographic image containing a tumor region mask, after preprocessing the image, the Simple Linear Iterative Clustering (SLIC) algorithm is used to pre-segment the region of interest (ROI) containing pixels into supervoxels within the subvolume pre-segmentation region. Subvolume pre-segmentation is then completed through iterative calculations.

[0120] Guided by prior clinical knowledge, the radiomics open-source toolkit PyRadiomics was used to extract Radiomics features from pre-segmented subvolume regions, obtaining feature values.

[0121] For all the obtained feature values, unsupervised clustering at the population level is performed using a Gaussian mixture model, and the clustering effect is evaluated using the Bayesian Information Criterion (BIC), generating subvolume clusters.

[0122] The optimal number of subvolume categories for each region of interest is obtained, and the tumor heterogeneity index is calculated to obtain an assessment of tumor heterogeneity.

[0123] Furthermore, if the modules / units integrated in the electronic device 20 are implemented as software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0124] 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 program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0125] 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.

[0126] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for assessing tumor heterogeneity based on multiple regions of interest in radiographic imaging, characterized in that, Includes the following steps: After preprocessing the radiographic image containing the tumor region mask, a linear iterative clustering algorithm is used within the subvolume pre-segmentation region to pre-segment the region of interest (MOI) containing pixels into multiple supervoxels. Subvolume pre-segmentation is completed through iterative calculation. The subvolume pre-segmentation region is formed by determining the bounding box based on the size and location of the tumor in the radiographic image. The MOI includes intratumoral, peritumoral, and tumor ring regions. The sub-volume pre-segmentation specifically refers to: Subvolume pre-segmentation regions are determined using geometric bounding boxes based on the size and location of the tumor in the radiographic images; Within the region encompassed by the tumor bounding box, a simple linear iterative clustering algorithm was used to count the total number of pixels. The region of interest is segmented into Each supervoxel specifically includes the following steps: initialization There are cluster centers, and the distance between adjacent centers is . ; With the cluster center as the origin and a side length of... The gradient value is calculated within the cube neighborhood, and the point with the smallest gradient change is selected as the new cluster center. Calculate the distance between each pixel and the cluster center. And re-divided Cluster centers; After an iterative process, the subvolume pre-segmentation is finally completed; The calculation of the distance between each pixel and the cluster center point And re-divided The number of cluster centers is calculated using the following formula: in Representing color distance, expressed as the intensity value I between two pixels. j and I i Difference representation, This represents spatial distance, which is the spatial distance between two pixels in the image. The spatial coordinates of the two pixels are... and , These are the coordinate parameters of the pixel, and the distance. Depend on and It consists of two parts. Maximum color distance Radiomics features are extracted from pre-segmented subvolume regions to obtain feature values; the radiomics features include first-order histogram features and texture features. For all the obtained feature values, Gaussian mixture model is used to perform unsupervised clustering at the population level, and the Bayesian information criterion (BIC) is used to evaluate the clustering effect, taking the value with the smallest BIC value. The optimal number of subvolume categories for tumor subvolume segmentation; This represents the optimal number of cluster centers; The number of optimal sub-volume categories within the tumor, around the tumor, and around the tumor ring is obtained respectively, and the number of optimal sub-volume categories within the tumor, around the tumor, and around the tumor ring reflects the complexity of the region of interest. The tumor heterogeneity index is calculated based on the number of optimal subvolume categories within the tumor, around the tumor, and around the tumor ring, the maximum subvolume volume in the region of interest, and the volume of the region of interest. The tumor heterogeneity is then assessed using the tumor heterogeneity index.

2. The tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging according to claim 1, characterized in that, The preprocessing of radiographic images containing tumor region masks also includes the following steps: Different preprocessing methods are selected based on the modality of the original radiographic images; the original image modalities include CT and MRI; MRI images are preprocessed using standard normalization calculations, as shown in the following formula: in These are the pixel intensity values ​​of the original image. The pixel intensity values ​​of the MRI image after standard normalization. The average pixel intensity value of the original image. The standard deviation of pixel intensity values ​​in the original image; The CT images are preprocessed using maximum and minimum value normalization calculations, as shown in the following formula: in These are the pixel intensity values ​​of the CT image after normalization of the maximum and minimum values. The minimum pixel intensity value in the original image. This represents the maximum intensity value of each pixel in the original image.

3. The tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging according to claim 1, characterized in that, The Bayesian Information Criterion (BIC) is used to evaluate the clustering effect, and the clustering effect is evaluated when the BIC value is the smallest. The optimal number of subvolume categories for tumor subvolume segmentation is as follows: For all the obtained feature values, Gaussian mixture models are used to perform unsupervised clustering at the population level; when performing unsupervised clustering at the population level, the categories... Clustering is performed based on the set values; After clustering is completed, the Bayesian Information Criterion (BIC) is used to evaluate the clustering effect. The calculation formula is as follows: in, This represents the total number of parameters in the Gaussian mixture model. For sample size, Let BIC be the likelihood function of the data under the Gaussian mixture model, and take the value when the BIC is minimized. The final number of sub-volume categories is used to determine the sub-volume categories. , , , ; Each patient's tumor mask is determined by , , , It consists of one or more categories combined.

4. The tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging according to claim 1, characterized in that, The tumor heterogeneity index is calculated using the following formula: in It is a tumor heterogeneity index. , and These respectively represent the complexity of the tumor interior, surrounding tumor, and tumor ring on radiographic images; , and These represent the maximum subvolume volume in the region of interest; , and These represent the volumes of the region of interest.

5. The tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging according to claim 1, characterized in that, Guided by prior clinical knowledge, Radiomics radiomics features were extracted from pre-segmented subvolume regions using the open-source radiomics toolkit PyRadiomics.

6. A tumor heterogeneity assessment system based on multiple regions of interest in radiographic imaging, characterized in that, The tumor heterogeneity assessment method based on multiple regions of interest in radiographic images, as applied in any one of claims 1-5, includes a subvolume pre-segmentation module, a feature extraction module, an unsupervised clustering module, and a computation execution module; The subvolume pre-segmentation module is used to process a given radiographic image containing a tumor region mask. After preprocessing the image, a linear iterative clustering algorithm is used to pre-segment the region of interest containing pixels into supervoxels within the subvolume pre-segmentation area. After iterative calculation, the subvolume pre-segmentation is completed. The feature extraction module extracts Radiomics features from the pre-segmented sub-volume regions to obtain feature values; The unsupervised clustering module performs population-level unsupervised clustering using a Gaussian mixture model on all obtained feature values, and evaluates the clustering effect using the Bayesian Information Criterion (BIC) to generate subvolumes. The computation execution module obtains the optimal number of subvolume categories for each of the multiple regions of interest, calculates the tumor heterogeneity index, and obtains an assessment of tumor heterogeneity.

7. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores computer program instructions that can be executed by the at least one processor to enable the at least one processor to perform the tumor heterogeneity assessment method based on multiple regions of interest in radiographic imaging as described in any one of claims 1-6.

8. A computer-readable storage medium storing a program, characterized in that, When the program is executed by the processor, it implements the tumor heterogeneity assessment method based on multiple regions of interest in radiographic images as described in any one of claims 1-6.

Images