Glioma radiotherapy response prediction model construction method, medium and product

Abstract

The invention discloses a glioma radiotherapy response prediction model construction method, a medium and a product, and relates to the technical field of medical instruments. According to the method, feature extraction and screening are carried out on the magnetic resonance images based on the radiomics technology, the sample data set is constructed, then the linear regression model is trained by using the sample data set, the trained linear regression model is obtained to serve as the glioma radiotherapy response prediction model, and glioma radiotherapy response prediction is realized.

Description

Technical Field

[0001] The present application relates to the field of medical device technology, and in particular to a method, medium and product for constructing a glioma radiotherapy response prediction model. Background Art

[0002] Low-grade glioma (LGG) is a type of primary brain tumor derived from glial cells. Compared with high-grade glioma, LGG grows more slowly, and after treatment, the overall prognosis of patients is relatively good. However, despite its low malignancy, the biological heterogeneity of LGG makes its treatment effect difficult to predict, which poses a significant challenge to clinical treatment. Radiotherapy is an important adjuvant treatment for LGG after surgery, and its efficacy varies greatly among different patients. Although increasing the radiotherapy dose may enhance the therapeutic effect, it also increases the risk of radiation-related toxicity such as radiation necrosis. Conversely, although reducing the radiotherapy dose can reduce side effects, it may lead to insufficient tumor control and affect long-term efficacy. This dilemma highlights the need to develop personalized treatment plans based on the individual characteristics of patients.

[0003] Existing clinical evaluations mainly rely on doctors' empirical judgments based on factors such as the age, gender, molecular pathology information of the tumor, imaging manifestations, and clinical status of glioma patients. In particular, imaging manifestations and clinical status of patients are often subjective, and it is difficult to unify the evaluation standards among different doctors, resulting in a lack of consistency and rigorous evidence support for the judgment of radiotherapy response. The current empirical evaluation method relies on the molecular pathology information of the tumor, which can only be obtained by sequencing and analyzing tumor tissue obtained through craniotomy or biopsy surgery, so it is impossible to achieve non-invasive prediction before surgery.

[0004] Currently, there is a lack of effective tools to guide personalized radiotherapy strategies for LGG patients.

[0005] Therefore, establishing a glioma radiotherapy response prediction model and realizing the prediction of glioma radiotherapy response is crucial for optimizing treatment strategies and improving patients' postoperative radiotherapy effects. Summary of the invention

[0006] The purpose of this application is to provide a method, medium and product for constructing a glioma radiotherapy response prediction model, so as to establish a glioma radiotherapy response prediction model and realize the prediction of glioma radiotherapy response.

[0007] To achieve the above objectives, this application provides the following solutions.

[0008] In a first aspect, the present application provides a method for constructing a glioma radiotherapy response prediction model, comprising the following steps.

[0009] Acquire a longitudinal multi-time point magnetic resonance image set; the longitudinal multi-time point magnetic resonance image set includes a plurality of magnetic resonance image sequences, the magnetic resonance image sequence includes magnetic resonance images of low-grade glioma of the same patient at least three time points; the three time points are respectively before surgery, before radiotherapy after surgery, and after the end of radiotherapy; The product change of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance images after surgery and before radiotherapy and the magnetic resonance images after the end of radiotherapy in each magnetic resonance imaging sequence was calculated as the product change corresponding to each magnetic resonance imaging sequence; Determine the radiotherapy sensitivity corresponding to each magnetic resonance imaging sequence according to the product change corresponding to each magnetic resonance imaging sequence; Extracting and screening features of the pre-operative magnetic resonance images in each magnetic resonance image sequence to obtain a final feature set corresponding to each magnetic resonance image sequence; The final feature set corresponding to each magnetic resonance imaging sequence is used as input, and the radiotherapy sensitivity corresponding to each magnetic resonance imaging sequence is used as a label to construct a sample data set; The sample data set is used to train a linear regression model, and the trained linear regression model is obtained as a glioma radiotherapy response prediction model.

[0010] In a second aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for constructing a glioma radiotherapy response prediction model.

[0011] In a third aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned method for constructing a glioma radiotherapy response prediction model.

[0012] According to the specific embodiments provided in this application, this application has the following technical effects.

[0013] The present application provides a method, medium and product for constructing a glioma radiotherapy response prediction model. The present application extracts and screens features of magnetic resonance images based on imaging genomics technology, constructs a sample data set, and then uses the sample data set to train a linear regression model to obtain the trained linear regression model as a glioma radiotherapy response prediction model, thereby realizing the prediction of glioma radiotherapy response. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 A flowchart of a method for constructing a glioma radiotherapy response prediction model provided in one embodiment of the present application.

[0016] Figure 2 This is a prediction performance curve of the glioma radiotherapy sensitivity prediction model provided in one embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0019] Radiomics technology is the product of the development and progress of image pattern recognition technology. It can extract a large number of objective quantitative features that are difficult to identify with the naked eye from medical images, including signal intensity distribution, three-dimensional shape, texture characteristics, and wavelet transformed features of tumor images. This technology realizes the quantitative description of the characteristics of solid tumor images and provides a new direction for the predictive analysis of clinical diagnosis and treatment-related events of such diseases. In glioma research, radiomics has shown a wide range of application potential in many fields, especially in the prediction of treatment response. For example, the radiomics model developed by Grossmann P and Kickingereder P et al. can effectively screen out patients with recurrent glioblastoma who may benefit from bevacizumab (anti-angiogenic therapy). Similarly, George E et al. constructed another radiomics model to identify patients with recurrent glioblastoma who may benefit from durvalumab (immunotherapy). These models provide important support for personalized treatment and are expected to improve patient treatment outcomes.

[0020] Magnetic resonance imaging (MRI) plays a vital role in the diagnosis, treatment planning, and follow-up monitoring of gliomas. Different MRI sequences, including T1-weighted phase, T2-weighted phase, contrast-enhanced T1-weighted phase, and fluid-attenuated inversion recovery (FLAIR), provide complementary information for evaluating imaging characteristics such as tumor structure, edema, and necrosis. In clinical practice, the treatment response of gliomas is usually evaluated according to the Response Assessment in Neuro-Oncology (RANO) criteria, which are based on longitudinal MRI scans at multiple time points and evaluate the treatment effect by monitoring the dynamic changes of tumor volume. Given the core role of MRI in glioma management, imaging omics models based on MRI data have become an ideal tool for predicting the radiotherapy sensitivity of low-grade gliomas, which helps to achieve personalized treatment plans.

[0021] The embodiment of the present application uses MRI data to construct a linear regression model to predict the sensitivity of LGG to radiotherapy, hoping to assist doctors in selecting the most appropriate individualized treatment plan, improve the treatment effect, and avoid unnecessary physical and mental pain and economic burden.

[0022] In an exemplary embodiment, Figure 1 As shown, a method for constructing a glioma radiotherapy response prediction model is provided, including the following steps 101 to 106.

[0023] Step 101, obtaining a longitudinal multi-time point magnetic resonance image set; the longitudinal multi-time point magnetic resonance image set includes multiple magnetic resonance image sequences, and the magnetic resonance image sequence includes magnetic resonance images of low-grade glioma of the same patient at least three time points; the three time points are respectively before surgery, before radiotherapy after surgery, and after the end of radiotherapy.

[0024] Step 102, calculating the product change of the transverse diameter and the longitudinal diameter of the maximum cross-section of the tumor in the magnetic resonance image after surgery and before radiotherapy and the magnetic resonance image after the completion of radiotherapy in each magnetic resonance image sequence as the product change corresponding to each magnetic resonance image sequence.

[0025] Step 103: determining the radiotherapy sensitivity corresponding to each magnetic resonance image sequence according to the product change corresponding to each magnetic resonance image sequence.

[0026] Step 104 , extracting and screening features of the pre-operative magnetic resonance images in each magnetic resonance image sequence to obtain a final feature set corresponding to each magnetic resonance image sequence.

[0027] Step 105 , taking the final feature set corresponding to each magnetic resonance image sequence as input, taking the radiotherapy sensitivity corresponding to each magnetic resonance image sequence as a label, and constructing a sample data set.

[0028] Step 106: Use the sample data set to train a linear regression model to obtain the trained linear regression model as a glioma radiotherapy response prediction model.

[0029] By implementing the above steps 101 to 106, a glioma radiotherapy response prediction model can be established to achieve the prediction of glioma radiotherapy response.

[0030] In another exemplary embodiment of the present application, the above step 101 is implemented in the following manner.

[0031] Longitudinal multi-time point magnetic resonance images of lower-grade gliomas (Word Health Organization, WHO grade 1-2) were collected in the brain glioma database. The collected magnetic resonance image sequences included T1-weighted phase, T2-weighted phase, and enhanced T1-weighted phase, and the specific time points for collection were before surgery, within 1 week before postoperative radiotherapy, and 1 month after the end of postoperative radiotherapy.

[0032] Magnetic resonance imaging (MRI) technology can highlight different types of tissue or lesion characteristics through different scanning sequences and parameter settings. T1-weighted phase, T2-weighted phase and enhanced T1-weighted phase are several common imaging sequences, which are of great significance in different clinical applications.

[0033] T1-weighted phase is an image acquired by setting a shorter repetition time (TR) and echo time (TE). T1 stands for "spin-lattice relaxation time", which is the time required for water molecules to return to equilibrium due to the interaction between water molecules and surrounding tissues. T1-weighted phase is usually used to display anatomical structures.

[0034] T2-weighted phases are images acquired with longer repetition times and echo times. T2 stands for "spin-spin relaxation time," which is the time it takes for energy to disappear as a result of water molecules interacting with their surroundings. T2-weighted phases are suitable for showing brain edema, tumors, inflammation, and other pathological changes.

[0035] Enhanced T1-weighted phase uses contrast agents (usually gadolinium-containing contrast agents) to enhance the signal of the image on the basis of T1-weighted phase to help improve the visibility of certain lesion areas. The contrast agent enters the blood circulation through intravenous injection and accumulates at the specific lesion site, thereby enhancing the signal at that site. Enhanced T1-weighted phase is widely used for the detection and location of tumors and the evaluation of the blood supply of tumors. It can also be used to identify some other types of lesions, such as infection, spinal cord lesions, etc.

[0036] In another exemplary embodiment of the present application, step 102 of the present application calculates the change in the product of the transverse diameter and the longitudinal diameter of the maximum cross-section of the tumor before and after radiotherapy on the T2 weighted phases collected in step 101 before and after radiotherapy, and specifically includes the following steps 201-203.

[0037] Step 201, using the following formula to calculate the product of the transverse diameter and the longitudinal diameter of the maximum cross-section of the tumor on the post-operative magnetic resonance image as the first product.

[0038] .

[0039] in, It is the product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance image before surgery and radiotherapy, that is, the first product. is the transverse diameter of the largest cross section of the tumor in the magnetic resonance imaging after surgery and before radiotherapy. It is the longitudinal diameter of the largest cross section of the tumor in the magnetic resonance imaging after surgery and before radiotherapy. and vertical.

[0040] Step 202, using the following formula to calculate the product of the transverse diameter and the longitudinal diameter of the maximum cross-section of the tumor on the magnetic resonance image after the radiotherapy is completed, as the second product.

[0041] .

[0042] in, It is the product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance image after the end of radiotherapy, that is, the second product. It is the transverse diameter of the largest cross section of the tumor in the magnetic resonance imaging after the completion of radiotherapy. It is the longitudinal diameter of the largest cross section of the tumor in the magnetic resonance imaging after the end of radiotherapy. and vertical.

[0043] Step 203: Calculate a product change according to the first product and the second product using the following formula.

[0044] .

[0045] in, The product changes.

[0046] In another exemplary embodiment of the present application, in the above step 103, based on the neuro-oncology response evaluation criteria, the changes in tumor volume before and after radiotherapy are compared to determine the radiotherapy sensitivity, and the specific evaluation criteria are as follows.

[0047] If the tumor disappears completely after treatment, it is considered a complete response; if the tumor product decreases by ≥50% after treatment, it is considered a partial response; if the tumor volume decreases by <50% but >25%, it is considered a minor response; if the tumor volume is between a decrease of 25% and an increase of 25%, it is considered stable disease; if the tumor volume increases by ≥25%, it is considered progressive disease. Complete response, partial response, and minor response are defined as radiotherapy sensitivity, and stable disease and progressive disease are defined as radiotherapy resistance.

[0048] In another exemplary embodiment of the present application, the above step 104 is to segment the tumor habitat sub-regions of the pre-operative magnetic resonance images in the magnetic resonance image sequence, extract the imaging genomics features for different sub-regions, fuse the features of each sub-region through post-fusion technology to form a unified feature set, and apply feature dimension reduction and screening methods to obtain the final feature set corresponding to each magnetic resonance image sequence. Specifically, it includes the following steps 301-306.

[0049] Step 301, segment the tumor habitat sub-regions of the magnetic resonance image before surgery in the magnetic resonance image sequence to obtain different tumor habitat sub-regions; the different tumor habitat sub-regions include solid tumor sub-regions, cystic necrosis sub-regions and edema sub-regions. The edema sub-region is the edema region around the tumor, that is, the edema region within a preset range centered on the tumor.

[0050] Step 302 , extracting radiomics features of different tumor habitat sub-regions to obtain radiomics features of each tumor habitat sub-region.

[0051] Step 303 , the radiomics features of each tumor habitat sub-region are fused through post-fusion technology to form a unified feature set.

[0052] Step 304 , using the Maximum Relevancy Minimum Redundancy (mRMR) algorithm to calculate the discriminability of each radiomics feature in the feature set, and sorting the radiomics features in descending order of discriminability to obtain a radiomics feature sequence.

[0053] Step 305, performing correlation analysis on the first preset number of radiomics features in the radiomics feature sequence, selecting radiomics features that meet preset conditions from the first preset number of radiomics features in the radiomics feature sequence to form a radiomics feature candidate set; the preset conditions are: r>0.9 and p<0.05, r is the correlation coefficient of the Spearman correlation test, and p is the statistical test value.

[0054] Step 306: Use the LASSO (Least Absolute Shrinkage and Selection Operator) algorithm to select a second preset number of radiomics features from the radiomics feature candidate set to form a final feature set. Exemplarily, the radiomics features selected in this step include the following 21 items.

[0055] 1. original_ngtdm_Strength_CE-T1: original neighborhood grayscale difference matrix strength-CE-T1.

[0056] 2. wavelet-LLH_firstorder_RobustMeanAbsoluteDeviation_T2: wavelet-LLH first-order feature robust mean absolute deviation_T2.

[0057] 3. wavelet-LLH_glcm_ClusterProminence_T2: wavelet-LLH gray-level co-occurrence matrix cluster kurtosis_T2.

[0058] 4. original_ngtdm_Busyness_CE-T1: original neighborhood grayscale difference matrix busyness_CE-T1.

[0059] 5. wavelet-LLH_firstorder_Variance_CE: wavelet-LLH first-order characteristic variance_CE.

[0060] 6. wavelet-LLH_firstorder_RootMeanSquared_CE: wavelet-LLH first-order characteristic root mean square value_CE.

[0061] 7. wavelet-LLH_glcm_Autocorrelation_T2: wavelet-LLH gray-level co-occurrence matrix autocorrelation_T2.

[0062] 8. original_gldm_SmallDependenceLowGrayLevelEmphasis_CE-T1: original grayscale local dependency matrix small dependency low graylevel emphasis_CE-T1.

[0063] 9. log-sigma-5-0-mm-3D_gldm_DependenceEntropy_T1: log-sigma-5-0-mm-3D grayscale local dependency matrix dependency entropy_T1.

[0064] 10. original_gldm_SmallDependenceHighGrayLevelEmphasis_CE-T1: original grayscale local dependency matrix small dependency high grayscale emphasis_CE-T1.

[0065] 11. log-sigma-5-0-mm-3D_glszm_ZoneVariance_T1: log-sigma-5-0-mm-3D grayscale zone size matrix zone variance_T1.

[0066] 12. wavelet-LLH_firstorder_RootMeanSquared_T2: wavelet-LLH first-order characteristic root mean square_T2.

[0067] 13. log-sigma-5-0-mm-3D_glszm_ZoneVariance_CE: log-sigma-5-0-mm-3D grayscale zone size matrix zone variance_CE.

[0068] 14. log-sigma-5-0-mm-3D_glszm_ZonePercentage_T1: log-sigma-5-0-mm-3D grayscale zone size matrix zone percentage_T1.

[0069] 15. wavelet-LLH_glszm_GrayLevelNonUniformityNormalized_CE: wavelet-LLH grayscale region size matrix normalized grayscale non-uniformity_CE.

[0070] 16. log-sigma-5-0-mm-3D_glszm_ZonePercentage_CE: log-sigma-5-0-mm-3D grayscale zone size matrix zone percentage_CE.

[0071] 17. wavelet-LLH_glrlm_ShortRunLowGrayLevelEmphasis_CE: Wavelet-LLH gray level co-occurrence matrix short run low gray level emphasis_CE.

[0072] 18. wavelet-LLH_glrlm_RunVariance_CE: wavelet-LLH gray-level co-occurrence matrix run variance_CE.

[0073] 19. log-sigma-5-0-mm-3D_ngtdm_Contrast_T2: log-sigma-5-0-mm-3D grayscale gradient matrix contrast_T2.

[0074] 20. wavelet-LLH_gldm_DependenceNonUniformity_T2: wavelet-LLH grayscale local dependency matrix dependence non-uniformity_T2.

[0075] 21. wavelet-LLH_glszm_LargeAreaEmphasis_T2: Wavelet-LLH grayscale area size matrix Large Area Emphasis_T2.

[0076] The suffixes T2, T1, and CE indicate that the feature is extracted from the T2-weighted phase, T1-weighted phase, and enhanced T1-weighted phase, respectively; CE-T1 indicates that the feature is the difference between the feature values ​​extracted from the enhanced T1-weighted phase and the T1-weighted phase.

[0077] In an exemplary embodiment, the glioma radiotherapy response prediction model obtained in step 106 is: ; in, is the sensitivity score for radiotherapy response. When the prediction result is determined to be sensitive to radiotherapy, otherwise the prediction result is determined to be insensitive to radiotherapy. to They are the original neighborhood grayscale difference matrix intensity_CE-T1, wavelet-LLH first-order feature robust mean absolute deviation_T2, wavelet-LLH grayscale symbiosis matrix cluster kurtosis_T2, original neighborhood grayscale difference matrix busyness_CE-T1, wavelet-LLH first-order feature variance_CE, wavelet-LLH first-order feature root mean square value_CE, wavelet-LLH grayscale symbiosis matrix autocorrelation_T2, original grayscale local dependence matrix small dependence low grayscale emphasis_CE-T1, log-sigma-5-0-mm-3D grayscale local dependence matrix dependency entropy_T1, original grayscale local dependence matrix small dependence high grayscale emphasis_CE-T1, log-sigma-5-0-mm-3D grayscale region size matrix region variance_T1, wavelet-LLH first-order feature RMS_T2, log-sigma-5-0-mm-3D grayscale region size matrix region variance_CE, log-sigma-5-0-mm-3D grayscale region size matrix region percentage_T1, wavelet-LLH grayscale region size matrix normalized grayscale inhomogeneity_CE, log-sigma-5-0-mm-3D grayscale region size matrix region percentage_CE, wavelet-LLH grayscale co-occurrence matrix short run low grayscale emphasis_CE, wavelet-LLH grayscale co-occurrence matrix long run variance_CE, log-sigma-5-0-mm-3D grayscale gradient matrix contrast_T2, wavelet-LLH grayscale local dependence matrix dependence inhomogeneity_T2, and wavelet-LLH grayscale region size matrix large region emphasis_T2.

[0078] After the glioma radiotherapy response prediction model is established, the key imaging genomics features on the preoperative magnetic resonance images of glioma patients are extracted and input into the established glioma radiotherapy response prediction model to obtain the predicted response of the patient to postoperative radiotherapy.

[0079] Based on the above-mentioned glioma radiotherapy sensitivity prediction model, the effectiveness of predicting glioma radiotherapy response is as follows: Figure 2 As shown, Figure 2 The receiver operating characteristic curve of the glioma radiotherapy sensitivity prediction model is shown in the figure. The X-axis is "1-specificity", which indicates the proportion of negative samples that are misclassified as positive samples by the glioma radiotherapy sensitivity prediction model; the Y-axis is "sensitivity", which indicates the proportion of positive samples correctly identified by the glioma radiotherapy sensitivity prediction model. The area under the receiver operating characteristic curve is 0.975, indicating that the overall performance of the glioma radiotherapy sensitivity prediction model is excellent and suitable for high-precision classification tasks.

[0080] Based on the above embodiments, it can be seen that the present application can accurately calculate the patient's radiotherapy response by analyzing the quantitative imaging features of the patient's magnetic resonance images and combining them with the established model. This is an evaluation method based on big data and objective evidence, which significantly improves the accuracy and consistency of judgment and overcomes the limitations of traditional empirical judgment. Moreover, the present application can achieve preoperative non-invasive radiotherapy response prediction without surgery by analyzing the patient's magnetic resonance imaging features.

[0081] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0082] In another exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0083] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0084] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application 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), magnetoresistive 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 may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0085] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0086] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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.

[0087] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for constructing a glioma radiotherapy response prediction model, characterized in that: include: Acquire a longitudinal multi-time point magnetic resonance image set; the longitudinal multi-time point magnetic resonance image set includes a plurality of magnetic resonance image sequences, the magnetic resonance image sequence includes magnetic resonance images of low-grade glioma of the same patient at least three time points; the three time points are respectively before surgery, before radiotherapy after surgery, and after the end of radiotherapy; The product change of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance images after surgery and before radiotherapy and the magnetic resonance images after the end of radiotherapy in each magnetic resonance imaging sequence was calculated as the product change corresponding to each magnetic resonance imaging sequence; Determine the radiotherapy sensitivity corresponding to each magnetic resonance imaging sequence according to the product change corresponding to each magnetic resonance imaging sequence; Extracting and screening features of the pre-operative magnetic resonance images in each magnetic resonance image sequence to obtain a final feature set corresponding to each magnetic resonance image sequence; The final feature set corresponding to each magnetic resonance imaging sequence is used as input, and the radiotherapy sensitivity corresponding to each magnetic resonance imaging sequence is used as a label to construct a sample data set; The sample data set is used to train a linear regression model, and the trained linear regression model is obtained as a glioma radiotherapy response prediction model.

2. The method for constructing a glioma radiotherapy response prediction model according to claim 1, characterized in that: The change in the product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in each magnetic resonance imaging sequence between the magnetic resonance imaging after surgery and before radiotherapy and the magnetic resonance imaging after the completion of radiotherapy was calculated, including: The product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance image after surgery and before radiotherapy was calculated as the first product using the following formula; ； in, It is the product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance image before surgery and radiotherapy, that is, the first product. is the transverse diameter of the largest cross section of the tumor in the magnetic resonance imaging after surgery and before radiotherapy. It is the longitudinal diameter of the largest cross section of the tumor on magnetic resonance imaging after surgery and before radiotherapy; The product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor after the end of radiotherapy was calculated as the second product using the following formula; ； in, It is the product of the transverse diameter and the longitudinal diameter of the largest cross-section of the tumor in the magnetic resonance image after the end of radiotherapy, that is, the second product. It is the transverse diameter of the largest cross section of the tumor in the magnetic resonance imaging after the completion of radiotherapy. It is the longitudinal diameter of the largest cross section of the tumor in the magnetic resonance imaging after the end of radiotherapy; Calculate the product change according to the first product and the second product using the following formula: ； in, The product changes.

3. The method for constructing a glioma radiotherapy response prediction model according to claim 2, characterized in that: The radiotherapy sensitivity corresponding to each magnetic resonance imaging sequence is determined according to the product change corresponding to each magnetic resonance imaging sequence, specifically including: like , the radiotherapy sensitivity is determined to be radiotherapy sensitive, otherwise, the radiotherapy sensitivity is determined to be radiotherapy resistant.

4. The method for constructing a glioma radiotherapy response prediction model according to claim 2, characterized in that: Feature extraction and screening are performed on the preoperative magnetic resonance images in each magnetic resonance image sequence to obtain the final feature set corresponding to each magnetic resonance image sequence, including: Segmenting the tumor habitat sub-regions of the magnetic resonance images before surgery in the magnetic resonance image sequence to obtain different tumor habitat sub-regions; the different tumor habitat sub-regions include solid tumor sub-regions, cystic necrosis sub-regions and edema sub-regions; Extract radiomics features of different tumor habitat sub-regions to obtain radiomics features of each tumor habitat sub-region; Post-fusion technology was used to fuse the radiomic features of each tumor habitat sub-region to form a unified feature set; The mRMR algorithm was used to calculate the discriminability of each radiomics feature in the feature set, and the radiomics features were sorted in descending order of discriminability to obtain the radiomics feature sequence; Performing correlation analysis on the first preset number of radiomics features in the radiomics feature sequence, selecting radiomics features that meet preset conditions from the first preset number of radiomics features in the radiomics feature sequence to form a radiomics feature candidate set; the preset conditions are: r>0.9 and p<0.05, r is the correlation coefficient of the Spearman correlation test, and p is the statistical test value; The LASSO algorithm is used to select a second preset number of radiomics features from the radiomics feature candidate set to form the final feature set.

5. The method for constructing a glioma radiotherapy response prediction model according to claim 4, characterized in that: The final feature set includes: original neighborhood grayscale difference matrix intensity_CE-T1, wavelet-LLH first-order feature robust mean absolute deviation_T2, wavelet-LLH grayscale co-occurrence matrix cluster kurtosis_T2, original neighborhood grayscale difference matrix busyness_CE-T1, wavelet-LLH first-order feature variance_CE, wavelet-LLH first-order feature root mean square value_CE, wavelet-LLH grayscale co-occurrence matrix autocorrelation_T2, original grayscale local dependence matrix small dependence low grayscale emphasis_CE-T1, log-sigma-5-0-mm-3D grayscale local dependence matrix dependency entropy_T1, original grayscale local dependence matrix small dependence high grayscale emphasis_CE-T1, log-sigma-5-0-mm-3D grayscale region size matrix region variance_T1, wavelet-LLH The first-order features are root mean square_T2, log-sigma-5-0-mm-3D grayscale region size matrix region variance_CE, log-sigma-5-0-mm-3D grayscale region size matrix region percentage_T1, wavelet-LLH grayscale region size matrix normalized grayscale inhomogeneity_CE, log-sigma-5-0-mm-3D grayscale region size matrix region percentage_CE, wavelet-LLH grayscale co-occurrence matrix short run low grayscale emphasis_CE, wavelet-LLH grayscale co-occurrence matrix run long variance_CE, log-sigma-5-0-mm-3D grayscale gradient matrix contrast_T2, wavelet-LLH grayscale local dependence matrix dependence inhomogeneity_T2, and wavelet-LLH grayscale region size matrix large region emphasis_T2.

6. The method for constructing a glioma radiotherapy response prediction model according to claim 1, characterized in that: The glioma radiotherapy response prediction model is: ； in, is the sensitivity score for radiotherapy response. When the prediction result is determined to be sensitive to radiotherapy, otherwise the prediction result is determined to be insensitive to radiotherapy. to They are the original neighborhood grayscale difference matrix intensity_CE-T1, wavelet-LLH first-order feature robust mean absolute deviation_T2, wavelet-LLH grayscale symbiosis matrix cluster kurtosis_T2, original neighborhood grayscale difference matrix busyness_CE-T1, wavelet-LLH first-order feature variance_CE, wavelet-LLH first-order feature root mean square value_CE, wavelet-LLH grayscale symbiosis matrix autocorrelation_T2, original grayscale local dependence matrix small dependence low grayscale emphasis_CE-T1, log-sigma-5-0-mm-3D grayscale local dependence matrix dependency entropy_T1, original grayscale local dependence matrix small dependence high grayscale emphasis_CE-T1, log-sigma-5-0-mm-3D grayscale region size matrix region variance_T1, wavelet-LLH first-order feature RMS_T2, log-sigma-5-0-mm-3D grayscale region size matrix region variance_CE, log-sigma-5-0-mm-3D grayscale region size matrix region percentage_T1, wavelet-LLH grayscale region size matrix normalized grayscale inhomogeneity_CE, log-sigma-5-0-mm-3D grayscale region size matrix region percentage_CE, wavelet-LLH grayscale co-occurrence matrix short run low grayscale emphasis_CE, wavelet-LLH grayscale co-occurrence matrix long run variance_CE, log-sigma-5-0-mm-3D grayscale gradient matrix contrast_T2, wavelet-LLH grayscale local dependence matrix dependence inhomogeneity_T2, and wavelet-LLH grayscale region size matrix large region emphasis_T2.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for constructing a glioma radiotherapy response prediction model described in any one of claims 1 to 6 is implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for constructing a glioma radiotherapy response prediction model described in any one of claims 1 to 6 is implemented.

Images