CRM scoring method and device for rectal cancer T2WI image mesangial intrastructure and storage medium

Abstract

The invention discloses a CRM scoring method and device for a rectal cancer T2WI image mesangial intrastructure and a storage medium. The method comprises the steps that an original T2WI image is read; the T2WI image is marked, and the classification and the position of the neoplastic structure in the mesangial are determined; inputting the T2WI image and the marked ROI into a trained 3D semantic segmentation model for identification, and summarizing identification results to obtain the position condition of the neoplastic structure in each mesangial in the original T2WI image; calculating the shortest distance between the neoplastic structure in each mesangial and the mesenteric fascia; and comparing the plurality of calculated shortest distances with a CRM positive standard, judging whether the shortest distances are positive or not, obtaining a CRM score value, and taking the CRM score value as a prognosis index. By means of the method, the rectal cancer T2WI image mesangial intrastructure can be automatically detected, and the time cost and the labor cost of quantitative CRM work of the rectal cancer can be saved to a great extent.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a CRM scoring method and device as well as a storage medium for rectal cancer T2WI image mesentery structures. Background Art

[0002] The incidence and mortality of rectal cancer are increasing year by year, and MRI is of great significance in assessing the risk of recurrence of rectal cancer. Baseline MRI contains many features with prognostic value, including extramural vascular invasion, tumor deposition, and circumferential resection margin (CRM). Among them, patients with positive CRM have a higher local recurrence rate after surgery, which is closely related to survival. Therefore, accurate preoperative assessment of rectal cancer CRM is helpful for clinical formulation of reasonable treatment plans and prognosis assessment.

[0003] At present, in the routine clinical work of rectal cancer CRM determination process, it is necessary to first manually identify the tumor structure in the mesentery, and then measure the distance between each structure and MRF. Manual identification and measurement is a complicated task that requires rich clinical experience and consumes a lot of manpower and time. Summary of the invention

[0004] The main purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art and to provide a CRM scoring method and device for the intra-mesenteric structure of rectal cancer T2WI images, as well as a storage medium, which can automatically detect the intra-mesenteric structure of rectal cancer T2WI images, and can greatly save the time and labor costs of quantitative CRM work for rectal cancer.

[0005] The technical solution provided by the present invention is as follows:

[0006] A first aspect of the present invention provides a CRM scoring method for rectal cancer T2WI images of mesentery structures, comprising the following steps:

[0007] Step S1, reading T2WI images to determine and match the distribution of tumor structures in different mesentery in the T2WI images of rectal cancer;

[0008] Step S2, by annotating the T2WI image, classifying each voxel in the mesentery of the T2WI image, and determining the location of the tumor structure in the mesentery, wherein the tumor structure in the mesentery includes tumor foci (Tumor, T), tumor deposits (Tumor deposits, TD), extramural vascular invasion (Extramural vascular invasion, EMVI) and metastatic lymph nodes, and the normal rectum and mesorectal fascia (Mesorectal fascia, MRF) are key structures for determining the location of the tumor structure in the mesentery;

[0009] Step S3, establishing a 3D semantic segmentation model, and training the 3D semantic segmentation model to obtain a trained 3D semantic segmentation model, inputting the T2WI image and the annotated ROI into the trained 3D semantic segmentation model for recognition, and summarizing the recognition results to obtain the position of each tumor structure in the mesangium in the original T2WI image;

[0010] Step S4, calculating the shortest distance between each tumor structure in the mesentery and the mesorectal fascia;

[0011] Step S5, comparing the calculated multiple shortest distances with the CRM positive standard and determining whether it is positive, obtaining a CRM score value, and using the CRM score value as a prognostic indicator.

[0012] Furthermore, in step S3, the 3D semantic segmentation model is preferably a nn U-net semantic segmentation model, the 3D semantic segmentation model includes a training set and a validation set, and the objective evaluation index of the 3D semantic segmentation model is the Dice similarity coefficient.

[0013] Furthermore, in step S2, each voxel in the mesangium of the T2WI image is classified by annotating the T2WI image, specifically:

[0014] According to the definition of tumor structure in the mesentery, different colors are used for "voxel-level" annotation in the T2WI image, and each voxel in the mesentery of the T2WI image is divided into one of normal rectum, tumor foci (Tumor, T), tumor deposits (Tumor deposits, TD), extramural vascular invasion (Extramural vascular invasion, EMVI), metastatic lymph nodes and mesorectal fascia (Mesorectal fascia, MRF).

[0015] Furthermore, in step S4, the calculation of the shortest distance between each tumor structure in the mesangium and the MRF comprises the following steps:

[0016] Locate the border of the mesorectal fascia;

[0017] Calculate the distance from the tumor structure in the mesorectum to the mesorectal fascia: For the tumor area, the Euclidean distance from each tumor voxel to the mesorectal boundary is calculated to obtain the shortest distance from each tumor voxel to the mesorectal fascia boundary.

[0018] Furthermore, in step S5, the CRM positive standard is determined by measuring the distance between the intra-mesorectal tumor structure and the mesorectal fascia on the T2WI image. If the distance between the intra-mesorectal tumor structure and the mesorectal fascia is >2 mm, the CRM is negative; if the distance is <1 mm, the CRM is positive; if the distance is between 1 and 2 mm, the CRM is threatened.

[0019] Furthermore, in step S5, after obtaining the CRM score value, the following is further included:

[0020] According to the CRM score value, the patient cohort is divided into two groups: high CRM score group and low CRM score group;

[0021] The Kaplan-Meier method was used to draw the survival curves of different patient groups, and the log-rank test was used to evaluate whether there were statistical differences in the survival of different groups. The Cox proportional hazard regression model was used to evaluate the impact of the CRM score as an influencing factor on the survival of rectal cancer patients and its independence compared with other clinical indicators.

[0022] A second aspect of the present invention provides a CRM scoring device for rectal cancer T2WI image mesangial structure, using the above-mentioned CRM scoring method for rectal cancer T2WI image mesangial structure, comprising:

[0023] MRI image reading module, used to read the original T2WI image, and to determine and match the distribution of tumor structures in different mesentery in the T2WI image of rectal cancer;

[0024] The mesangial structure annotation module is used to annotate the T2WI image, classify each voxel in the mesangial area of ​​the T2WI image, and determine the location of the tumor structure in the mesangial area;

[0025] A 3D semantic segmentation model training module is used to establish a 3D semantic segmentation model and train the 3D semantic segmentation model to obtain a trained 3D semantic segmentation model, input the T2WI image and the annotated ROI into the trained 3D semantic segmentation model for recognition, and summarize the recognition results to obtain the position status of each tumor structure in the mesangium in the original T2WI image;

[0026] The shortest distance calculation module is used to calculate the shortest distance between each tumor structure in the mesentery and the mesorectal fascia;

[0027] The prognostic indicator calculation and patient stratification module is used to compare the calculated multiple shortest distances with the CRM positive standard and determine whether it is positive, obtain the CRM score value, and use the CRM score value as the prognostic indicator.

[0028] A third aspect of the present invention provides a storage medium, comprising:

[0029] A readable storage medium and a computer program, wherein the computer program is used to control the above-mentioned CRM scoring method for the intra-mesenteric structure of rectal cancer T2WI images.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] At present, in the CRM determination process of rectal cancer in routine clinical work, it is necessary to first manually identify the tumor structure in the mesentery, and then measure the distance between each structure and the MRF. Manual identification and measurement is a complicated task that requires rich clinical experience and consumes a lot of manpower and time. Therefore, through the method proposed in this application, it is possible to achieve fully automatic, repeatable, and highly robust detection of tumor structures in the mesentery on T2WI images of rectal cancer, which can greatly save the time and manpower costs of CRM scoring of rectal cancer, and further obtain CRM scores as prognostic factors independent of TNM staging, which promotes the clinical transformation of prognostic indicators and achieves more accurate tumor treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of a CRM scoring method for the intra-mesenteric structure of rectal cancer T2WI images in an embodiment of the present invention;

[0033] Figure 2 It is a schematic diagram of realizing automatic identification of the CRM status of the structure within the mesangium as negative in an embodiment of the present invention;

[0034] Figure 3 Schematic diagram of a CRM scoring device for the intra-mesenteric structure of rectal cancer T2WI images in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the embodiments described below are part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0036] Therefore, the detailed description of the embodiments of the present application provided below in conjunction with the accompanying drawings is intended to represent only selected embodiments of the present application, and is not intended to limit the scope of protection claimed in the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0037] Example 1

[0038] See also Figure 1 This embodiment provides a CRM scoring method for rectal cancer T2WI image mesentery structure, including the following steps:

[0039] Step S1, reading T2WI images to determine and match the distribution of tumor structures in different mesentery in the rectal cancer T2WI images. The original T2WI images are provided by the Affiliated Cancer Hospital of Fudan University.

[0040] Step S2: By annotating the T2WI images, each voxel in the mesentery of the T2WI images is classified to determine the location of the tumor structure in the mesentery. The tumor structure in the mesentery includes tumor foci (Tumor, T), tumor deposits (Tumordeposits, TD), extramural vascular invasion (Extramural vascular invasion, EMVI) and metastatic lymph nodes. The normal rectum and mesorectal fascia (Mesorectal fascia, MRF) are the key structures for determining the location of the tumor structure in the mesentery.

[0041] Step S3, establish a 3D semantic segmentation model, train the 3D semantic segmentation model, obtain a trained 3D semantic segmentation model, input the T2WI image and the annotated ROI into the trained 3D semantic segmentation model for recognition, and summarize the recognition results to obtain the position of each tumor structure in the mesentery in the original T2WI image. ROI is the abbreviation of Region of Interest, that is, the region of interest, including tumor foci, TD, EMVI, metastatic lymph nodes, normal rectum and MRF.

[0042] Step S4, calculating the shortest distance between each tumor structure in the mesentery and the mesorectal fascia.

[0043] Step S5, comparing the calculated multiple shortest distances with the CRM positive standard and determining whether it is positive, obtaining a CRM score value, and using the CRM score value as a prognostic indicator.

[0044] The mesorectal fascia contains normal tissues such as adipose fibrous connective tissue, superior and inferior rectal arteries and veins, rectal lymphatic vessels and lymph nodes, nerves that control the rectum, and intra-mesenteric tumor structures such as EMVI, metastatic lymph nodes, and TD. The mesorectum shows a uniform high signal on the non-fat-suppressed T2WI image, which can form a good contrast with its internal tissues and is conducive to the evaluation of lesions.

[0045] The above structures are defined as follows: Tumor foci refer to malignant tumors that occur above the rectal dentate line to the junction of the rectum and the sigmoid colon; when the T2WI image shows uniform or uneven equal or slightly high signals in the rectum, it is marked as a tumor foci identified by MR (Label 1). EMVI refers to tumor invasion of veins outside the rectal muscularis propria; when the following features are met: 1. Tumor signals appear in blood vessels around the tumor, 2. The lumen of the blood vessels is enlarged and the shape is irregular, it is marked as EMVI positive identified by MR (Label 2). TD is a satellite nodule in the perirectal fat tissue, which is formed entirely by tumor cells without residual lymph node components; when the irregular nodule appears in the mesentery that is discontinuous with the primary tumor and continuous with the venous branches in the mesentery, it is marked as MR identified TD (Label 3). Metastatic lymph nodes (Label 4) are located in the mesorectum. Metastatic lymph nodes are determined by combining the size and morphological characteristics of the lymph nodes. When the following characteristics are met: 1. The long diameter of the nodule is ≥9mm, the lymph node is considered a metastatic lymph node; 2. If the long diameter of the nodule is 5-9mm and there are two or more malignant signs, the lymph node is considered a metastatic lymph node; 3. If the long diameter of the nodule is <5mm and there are three malignant signs, the lymph node is considered a metastatic lymph node. The mesorectum is a term used in surgery to describe the tissue wrapped by the rectal fascia. It contains adipose fibrous connective tissue, superior and inferior rectal arteries and veins, rectal lymphatic vessels and lymph nodes, nerves that control the rectum, and other tissues. The fascia that surrounds the mesorectum is MRF. When it meets the characteristics of uniform high signal on unsuppressed T2WI and a circle of lower signal MRF on the periphery on unsuppressed T2WI, it is marked as mesorectum (Label 5). The normal rectum is located at the end of the large intestine and is about 15-16 cm long. If it meets the characteristics of a continuous intestinal tube that runs from the starting point, the sigmoid take-off, along the front of the sacrum and coccyx and ends at the anus, it is marked as a normal rectum (Label 6).

[0046] Optionally, in step S3, the 3D semantic segmentation model is preferably a nn U-net semantic segmentation model, the 3D semantic segmentation model includes a training set and a validation set, and the objective evaluation index of the 3D semantic segmentation model is the Dice similarity coefficient.

[0047] In this embodiment, the 3D semantic segmentation model is trained in the following manner:

[0048] Professional radiologists have delineated six different types of tissue categories on the T2WI images of rectal cancer used for training, namely normal rectum, tumor foci, tumor deposits, extrarectal vascular invasion, metastatic lymph nodes and mesorectal fascia, and then input them into the 3D semantic segmentation model for training; the training set consists of 500 patients diagnosed with rectal cancer by pathology, and the validation set consists of 300 cases. The objective evaluation index of the model segmentation is the Dice similarity coefficient (DSC) value. DSC is a set similarity measurement index, which is usually used to calculate the similarity of two samples. It can be understood as the overlap between the model prediction area and the annotated area of ​​the radiologist's gold standard.

[0049] The calculation formula of the Dice coefficient is:

[0050]

[0051] The Dice coefficient is defined as twice the area of ​​the intersection of A and B divided by the sum of the areas of A and B; TP = true positive, FP = false positive, FN = false negative. The higher the Dice coefficient, the better the performance of the model.

[0052] Optionally, in step S2, each voxel in the T2WI image is classified by annotating the T2WI image, specifically:

[0053] According to the definition of tumor structure in the mesentery, different colors are used for "voxel-level" annotation in the T2WI image, and each voxel in the mesentery of the T2WI image is divided into one of normal rectum, tumor foci (Tumor, T), tumor deposits (Tumor deposits, TD), extramural vascular invasion (Extramural vascular invasion, EMVI), metastatic lymph nodes and mesorectal fascia (Mesorectal fascia, MRF).

[0054] Optionally, in step S4, calculating the shortest distance between each tumor structure in the mesangium and the MRF comprises the following steps:

[0055] Locate the border of the mesorectal fascia;

[0056] Calculate the distance from the tumor structure in the mesorectum to the mesorectal fascia: For the tumor area, the Euclidean distance from each tumor voxel to the mesorectal boundary is calculated to obtain the shortest distance from each tumor voxel to the mesorectal fascia boundary.

[0057] In this embodiment, the shortest distance between each tumor structure in the mesangium and the MRF is calculated, including the following steps:

[0058] Find the border of the mesorectum: The periphery of the mesorectum region of interest is the mesorectum border, or MRF.

[0059] Calculate the distance from the tumor structure in the mesorectum to the mesorectum: For example, for the tumor region, calculate the shortest distance from each tumor voxel to the mesorectum boundary. This is achieved by calculating the Euclidean distance from each tumor voxel to the mesorectum boundary.

[0060] The same calculation is performed for other tumor structures in the mesentery. Calculation of the distance from the metastatic lymph node to the mesorectum: Since there may be multiple metastatic lymph node regions, the shortest distance from each voxel in each lymph node region to the mesorectum boundary needs to be calculated, and the smallest distance is selected from these shortest distances.

[0061] The same calculation is performed for other tumor structures in the mesangium. The process can be described as:

[0062] make,

[0063] P represents the set of points on the mesorectum boundary

[0064] T represents the set of voxels in the tumor region.

[0065] E represents the voxel set of the EMVI region.

[0066] Gi represents the set of voxels in the i-th TD region.

[0067] Ni represents the voxel set of the i-th lymph node region

[0068] d(x, y) represents the Euclidean distance from point x to point y

[0069] So,

[0070] The shortest distance from the tumor to the mesorectum can be expressed as:

[0071] min p ∈P,t∈T d(p,t)

[0072] The shortest distance from EMVI to the mesorectum can be expressed as:

[0073] min p ∈P,e∈T d(p,e)

[0074] The minimum distance from TD to the mesorectum can be expressed as;

[0075] min i (min p∈P, g∈Gi d(p, g))

[0076] The minimum distance from lymph node to mesorectum can be expressed as;

[0077] min i (min p∈P, n∈Ni d(p,n))

[0078] Finally, the minimum distance of the tumor structure in the mesentery was compared. If min>1mm, CRM was negative; otherwise, CRM was positive.

[0079] Optionally, in step S5, the CRM positive standard is determined by measuring the distance between the intra-mesorectal tumor structure and the mesorectal fascia on the T2WI image. If the distance between the intra-mesorectal tumor structure and the mesorectal fascia is >2 mm, the CRM is negative; if the distance is <1 mm, the CRM is positive; if the distance is between 1 and 2 mm, the CRM is threatened. Figure 2 As shown, in this embodiment, the shortest distances between tumor foci, tumor deposits, metastatic lymph nodes and MRF are measured. The tumor deposit distance is the shortest but the distance is >1 mm, and the CRM status is negative.

[0080] Optionally, in step S5, after obtaining the CRM score value, the following steps are further included:

[0081] According to the CRM score value, the patient cohort is divided into two groups: high CRM score group and low CRM score group;

[0082] The Kaplan-Meier method was used to draw the survival curves of different patient groups, and the log-rank test was used to evaluate whether there were statistical differences in the survival of different groups. The Cox proportional hazard regression model was used to evaluate the impact of the CRM score as an influencing factor on the survival of rectal cancer patients and its independence compared with other clinical indicators.

[0083] The present application provides a CRM scoring method for intra-mesenteric structures in T2WI images of rectal cancer. Through the method provided in the present application, it is possible to realize fully automatic, repeatable, and highly robust detection of intra-mesenteric tumor structures on T2WI images of rectal cancer, which can greatly save the time and labor costs of CRM scoring of rectal cancer, and further obtain the CRM score as a prognostic factor independent of TNM staging, thereby promoting the clinical transformation of prognostic indicators and realizing more accurate tumor treatment.

[0084] Example 2

[0085] like Figure 3 As shown, this embodiment provides a CRM scoring device for the intra-mesomal structure of rectal cancer T2WI images, which adopts the above-mentioned CRM scoring method for the intra-mesomal structure of rectal cancer T2WI images, including:

[0086] The MRI image reading module 201 is used to read the original T2WI image. The MRI image reading module is used to determine and match the distribution of tumor structures in different mesentery in the rectal cancer T2WI image.

[0087] The intra-mesenteric structure labeling module 202 is used to label the T2WI image, classify each voxel in the mesenteric region of the T2WI image, and determine the location of the tumor structure in the mesenteric region.

[0088] The 3D semantic segmentation model training module 203 is used to establish a 3D semantic segmentation model and train the 3D semantic segmentation model to obtain a trained 3D semantic segmentation model, input the T2WI image and the annotated ROI into the trained 3D semantic segmentation model for recognition, and summarize the recognition results to obtain the position status of each tumor structure in the mesangium in the original T2WI image.

[0089] The shortest distance calculation module 204 is used to calculate the shortest distance between each tumor structure in the mesentery and the mesorectal fascia.

[0090] The prognostic indicator calculation and patient stratification module 205 is used to compare the calculated multiple shortest distances with the CRM positive standard and determine whether it is positive, obtain the CRM score value, and use the CRM score value as the prognostic indicator.

[0091] The present embodiment provides a CRM scoring device for intra-mesenteric structures in rectal cancer T2WI images, which includes an MRI image reading module 201, an intra-mesenteric structure annotation module 202, a 3D semantic segmentation model training module 203, a shortest distance calculation module 204, and a prognosis index calculation and patient stratification module 205. The MRI image reading module 201 can determine and match the distribution of different intra-mesenteric tumor structures in the rectal cancer T2WI image, and then the intra-mesenteric structure annotation module 202 annotates the region of interest in the T2WI image with different colors to determine the position of the intra-mesenteric tumor structure. The 3D semantic segmentation model training module 203 determines the different intra-mesenteric tumor structures in each T2WI image, automatically identifies the intra-mesenteric tumor structure, and the shortest distance calculation module 204 calculates the shortest distance between each intra-mesenteric tumor structure output by the 3D semantic segmentation model and the MRF. Finally, the prognosis index calculation and patient stratification module 205 compares the calculated minimum shortest distance with the CRM positive standard (1 mm) to determine whether it is positive.

[0092] Example 3

[0093] This embodiment provides a storage medium, including:

[0094] A readable storage medium and a computer program, wherein the computer program is used to control the above-mentioned CRM scoring method for the intra-mesenteric structure of rectal cancer T2WI images.

[0095] The computer-readable storage medium in this embodiment can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that includes one or more available media integrated into it. The available media can be magnetic media (for example, floppy disks, hard disks, tapes), optical media (for example, DVDs), or semiconductor media (for example, SSDs), etc.

[0096] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0097] The above is only the best specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application shall be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A CRM scoring method for mesentery structures in T2WI images of rectal cancer, characterized in that: The following steps are involved: Step S1, reading images in the T2WI sequence respectively; Step S2, by annotating the T2WI image, classifying each voxel in the mesentery of the T2WI image, and determining the location of the tumor structure in the mesentery, wherein the tumor structure in the mesentery includes tumor foci (Tumor, T), tumor deposits (Tumor deposits, TD), extramural vascular invasion (Extramural vascular invasion, EMVI) and metastatic lymph nodes, and the normal rectum and mesorectal fascia (Mesorectal fascia, MRF) are key structures for determining the location of the tumor structure in the mesentery; Step S3, establishing a 3D semantic segmentation model, and training the 3D semantic segmentation model to obtain a trained 3D semantic segmentation model, inputting the T2WI image and the annotated ROI into the trained 3D semantic segmentation model for recognition, and summarizing the recognition results to obtain the position of each tumor structure in the mesangium in the original T2WI image; Step S4, calculating the shortest distance between each tumor structure in the mesentery and the mesorectal fascia; Step S5, comparing the calculated multiple shortest distances with the CRM positive standard and determining whether it is positive, obtaining a CRM score value, and using the CRM score value as a prognostic indicator.

2. The CRM scoring method for rectal cancer T2WI images of mesentery structures according to claim 1, characterized in that: In step S3, the 3D semantic segmentation model is a nn U-net semantic segmentation model, the 3D semantic segmentation model includes a training set and a validation set, and the objective evaluation index of the 3D semantic segmentation model is a Dice similarity coefficient.

3. The CRM scoring method for rectal cancer T2WI images of mesentery structures according to claim 1, characterized in that: In step S2, each voxel in the stenosis of the T2WI image is classified by annotating the T2WI image, specifically: According to the definition of tumor structures in the mesentery, different colors are used for "voxel-level" annotation in the T2WI image, and each voxel in the mesentery of the T2WI image is divided into one of the following categories: normal rectum, tumor foci (Tumor, T), tumor deposits (Tumor deposits, TD), extramural vascular invasion (Extramural vascular invasion, EMVI), metastatic lymph nodes and mesorectal fascia (Mesorectal fascia, MRF).

4. The CRM scoring method for rectal cancer T2WI images of mesentery structure according to claim 1, characterized in that: In step S4, the calculation of the shortest distance between each tumor structure in the mesangium and the MRF includes the following steps: Locate the border of the mesorectal fascia; Calculate the distance from the tumor structure in the mesorectum to the mesorectal fascia: For the tumor area, the Euclidean distance from each tumor voxel to the mesorectal boundary is calculated to obtain the shortest distance from each tumor voxel to the mesorectal fascia boundary.

5. The CRM scoring method for rectal cancer mesentery structure based on T2WI images according to any one of claims 1 to 4, characterized in that: In step S5, the CRM positive standard is determined by measuring the distance between the intra-mesorectal tumor structure and the mesorectal fascia on the T2WI image. If the distance between the intra-mesorectal tumor structure and the mesorectal fascia is >2 mm, the CRM is negative; if the distance is <1 mm, the CRM is positive; if the distance is between 1 and 2 mm, the CRM is threatened.

6. The CRM scoring method for rectal cancer T2WI images of mesentery structures according to claim 5, characterized in that: In step S5, after obtaining the CRM score value, the following steps are further included: According to the CRM score value, the patient cohort is divided into two groups: high CRM score group and low CRM score group; The Kaplan-Meier method was used to draw the survival curves of different patient groups, and the log-rank test was used to evaluate whether there were statistical differences in the survival of different groups. The Cox proportional hazard regression model was used to evaluate the impact of the CRM score as an influencing factor on the survival of rectal cancer patients and its independence compared with other clinical indicators.

7. A CRM scoring device for rectal cancer T2WI image mesangial structure, using the CRM scoring method for rectal cancer T2WI image mesangial structure according to any one of claims 1 to 6, characterized in that: include: MRI image reading module, used for reading original T2WI images, including axial or oblique axial T2WI; The mesangial structure annotation module is used to annotate the T2WI image, classify each voxel in the mesangium of the T2WI image, and determine the location of the tumor structure in the mesangium; A 3D semantic segmentation model training module is used to establish a 3D semantic segmentation model and train the 3D semantic segmentation model to obtain a trained 3D semantic segmentation model, input the T2WI image and the annotated ROI into the trained 3D semantic segmentation model for recognition, and summarize the recognition results to obtain the position status of each tumor structure in the mesangium in the original T2WI image; The shortest distance calculation module is used to calculate the shortest distance between each tumor structure in the mesentery and the mesorectal fascia; The prognostic indicator calculation and patient stratification module is used to compare the calculated multiple shortest distances with the CRM positive standard and determine whether it is positive, obtain the CRM score value, and use the CRM score value as the prognostic indicator.

8. A storage medium, characterized in that: include: A readable storage medium and a computer program, wherein the computer program is used to control the CRM scoring method for the intra-mesenteric structure of rectal cancer T2WI images as described in any one of claims 1-6.

Images