Colorectal cancer postoperative risk assessment method, device and equipment and storage medium

Abstract

The invention discloses a colorectal cancer postoperative risk assessment method, device and equipment and a storage medium, and relates to the technical field of colorectal cancer postoperative assessment, and the colorectal cancer postoperative risk assessment method comprises the following steps: based on a multi-head attention mechanism, segmenting a centrum horizontal CT image of a preoperative and postoperative third lumbar vertebra of a target object, and obtaining a centrum horizontal CT image of a second lumbar vertebra of the target object; obtaining a skeletal muscle area SMA, a visceral fat area VFA and a subcutaneous fat area SFA corresponding to preoperative and postoperative operations; respectively calculating a skeletal muscle index SMI and a visceral fat to subcutaneous fat ratio VSR according to SMA, VFA and SFA before and after the operation; according to the gender of the target object and the preoperative and postoperative SMI and VSR, judging whether the target object is sarcopenia and / or visceral fat obesity; and dividing risk levels according to preoperative and postoperative judgment results. Based on a multi-head attention mechanism, automatic accurate segmentation and risk assessment of the colorectal cancer postoperative body composition are realized, and the analysis efficiency and accuracy are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of postoperative assessment technology for colorectal cancer, specifically to a method, device, equipment, and storage medium for postoperative risk assessment of colorectal cancer. Background Technology

[0002] Sarcopenia and visceral fat obesity are key risk factors for poor long-term prognosis after CRC (colorectal cancer) surgery. Currently, CT (Computed Tomography) has become one of the gold standards for quantifying body composition at the tissue and organ level due to its high resolution. Among these, the area of ​​adipose tissue and skeletal muscle measured on a single-slice axial CT image at the level of the third lumbar vertebra shows a strong correlation with the overall distribution of the body and is widely used in oncological assessment.

[0003] Traditional CT imaging assessment involves doctors manually segmenting skeletal muscle and adipose tissue regions based on the different HU (Hunsfield Unit) ranges of each tissue.

[0004] However, existing technologies generally suffer from the following problems: traditional manual sketching methods have significant shortcomings. On the one hand, they are labor-intensive and time-consuming, making it difficult to meet the needs of efficient clinical assessment; on the other hand, manual operation is easily affected by the subjective factors of doctors, resulting in human error. In addition, there is a lack of efficient and automated long-term prognostic risk assessment systems after CRC surgery in clinical practice, which cannot provide sufficient and reliable support for subsequent treatment and follow-up decisions. Summary of the Invention

[0005] This application provides a method, device, equipment, and storage medium for postoperative risk assessment of colorectal cancer. Based on a multi-head attention mechanism, it achieves automated and precise segmentation and risk assessment of body components after colorectal cancer surgery, significantly improving analysis efficiency and accuracy.

[0006] In a first aspect, embodiments of this application provide a method for postoperative risk assessment of colorectal cancer, the method comprising: Based on the multi-head attention mechanism, the vertebral body level CT images of the third lumbar vertebra of the target object before and after surgery were segmented to obtain the skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative conditions. Based on the preoperative and postoperative SMA, VFA and SFA, the skeletal muscle index SMI = SMA / target subject height² and the visceral fat to subcutaneous fat ratio VSR = VFA / SFA were calculated respectively. Based on the gender of the target subject and the preoperative and postoperative SMI and VSR, determine whether it is sarcopenia and / or visceral fat obesity; Risk levels are determined based on pre- and post-operative assessments.

[0007] In conjunction with the first aspect, in one implementation, determining whether the patient has sarcopenia and / or visceral fat obesity based on the target subject's gender and the preoperative and postoperative SMI and VSR includes: The target group is male. If the SMI is less than the first threshold, it is judged as sarcopenia. If the VSR is greater than or equal to the second threshold, it is judged as visceral fat obesity. The target group is women. If the SMI is less than the third threshold, it is judged as sarcopenia. If the VSR is greater than or equal to the fourth threshold, it is judged as visceral fat obesity.

[0008] In conjunction with the first aspect, in one embodiment, the first threshold is 39.4 cm² / m², the second threshold is 0.6, the third threshold is 32.5 cm² / m², and the fourth threshold is 1.1.

[0009] In conjunction with the first aspect, in one implementation, the step of classifying risk levels based on pre- and post-operative assessment results includes: If sarcopenia and / or visceral fat obesity are present both before and after surgery, the risk is high. If sarcopenia and / or visceral fat obesity are present before or after surgery, the risk level is medium. If there is no sarcopenia or visceral fat obesity before and after the operation, the risk is low.

[0010] In conjunction with the first aspect, in one implementation, the segmentation of the preoperative and postoperative vertebral body CT images of the target subject based on the multi-head attention mechanism to obtain the skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative data includes: The vertebral body level CT images of the third lumbar vertebra of the target object before and after surgery were input into the body composition analysis model based on the improved multi-head attention mechanism for segmentation, and the skeletal muscle area (SMA), visceral fat area (VFA) and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative conditions were obtained. The body composition analysis model is configured as follows: Based on a 3D segmentation network, an encoder is configured to downsample and extract deep features from the image, and a decoder is configured to upsample and restore the image resolution and output the segmentation results. In addition, a multi-head attention module is set in each feature extraction stage of the encoder.

[0011] In conjunction with the first aspect, in one implementation, prior to segmenting the preoperative and postoperative vertebral body level CT images of the target object's third lumbar vertebra based on the multi-head attention mechanism, the method further includes: Extract preoperative and postoperative abdominal CT non-contrast scan images of the target subject; Based on preoperative and postoperative abdominal CT non-contrast scan images, vertebral level CT images of the third lumbar vertebra were selected from the preoperative and postoperative images of the target subjects.

[0012] In conjunction with the first aspect, in one implementation, it further includes: Spatial registration was performed on the preoperative and postoperative vertebral CT images of the third lumbar vertebra of the target subject. The registered image is normalized to the set HU range.

[0013] Secondly, embodiments of this application provide a postoperative risk assessment device for colorectal cancer, the postoperative risk assessment device for colorectal cancer comprising: The segmentation module is used to segment the vertebral body level CT images of the third lumbar vertebra of the target object before and after surgery based on the multi-head attention mechanism, and obtain the skeletal muscle area (SMA), visceral fat area (VFA) and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative conditions. The analysis module is used to calculate the skeletal muscle index SMI=SMA / target subject height² and the visceral fat to subcutaneous fat ratio VSR=VFA / SFA based on the SMA, VFA and SFA before and after the operation; and to determine whether the target subject has sarcopenia and / or visceral fat obesity based on the gender of the target subject and the SMI and VSR before and after the operation. The assessment module is used to classify risk levels based on pre- and post-operative assessment results.

[0014] Thirdly, embodiments of this application provide a postoperative risk assessment device for colorectal cancer, the postoperative risk assessment device for colorectal cancer includes a processor, a memory, and a postoperative risk assessment program for colorectal cancer stored in the memory and executable by the processor, wherein when the postoperative risk assessment program for colorectal cancer is executed by the processor, the steps of the above-described postoperative risk assessment method for colorectal cancer are implemented.

[0015] Fourthly, a computer-readable storage medium storing a postoperative risk assessment program for colorectal cancer, wherein when the postoperative risk assessment program for colorectal cancer is executed by a processor, the steps of the above-described postoperative risk assessment method for colorectal cancer are implemented.

[0016] The beneficial effects of the technical solutions provided in this application include at least the following: The postoperative risk assessment method for colorectal cancer in this application is based on a multi-head attention mechanism. It segments the vertebral body CT images of the third lumbar vertebra of the target subject before and after surgery to obtain the skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative values. Based on the preoperative and postoperative SMA, VFA, and SFA, the skeletal muscle index (SMI) and the visceral fat to subcutaneous fat ratio (VSR) are calculated, respectively. Based on the gender of the target subject and the preoperative and postoperative SMI and VSR, it is determined whether the subject has sarcopenia and / or visceral fat obesity. Based on the preoperative and postoperative assessment results, the risk level is classified.

[0017] The body composition analysis method using a multi-head attention mechanism replaces the traditional manual delineation method. On the one hand, it significantly reduces the workload of doctors, eliminating the need for a lot of time to manually segment images and significantly improving the efficiency of body composition analysis. On the other hand, the segmentation process is not affected by human subjective factors, effectively reducing human error and improving the accuracy of skeletal muscle and adipose tissue segmentation and related index calculation. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an embodiment of the colorectal cancer postoperative risk assessment method of this application; Figure 2 This is an image showing the automatic body composition segmentation results of this application; Figure 3 This is a flowchart illustrating the workflow in the embodiments of this application; Figure 4 This is a structural block diagram of an embodiment of the colorectal cancer postoperative risk assessment device of this application; Figure 5 This is a schematic diagram of the hardware structure of the postoperative risk assessment device for colorectal cancer involved in the embodiments of this application. Detailed Implementation

[0019] 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 some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0021] In one aspect, embodiments of this application provide a method for postoperative risk assessment of colorectal cancer.

[0022] In one embodiment, reference is made to Figure 1 , Figure 1 This is a schematic flowchart of an embodiment of the risk assessment method for colorectal cancer surgery according to this application. Figure 1 As shown, postoperative risk assessment methods for colorectal cancer include: S1. Based on the multi-head attention mechanism, the vertebral body level CT images of the third lumbar vertebra of the target object before and after surgery are segmented to obtain the skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative conditions. It is worth noting that before step S1, it is necessary to collect the basic data required for subsequent analysis, including two parts: first, the patient's (target subject's) gender and height data from clinical records; and second, the patient's imaging data. By connecting to the hospital's Picture Archiving and Communication System (PACS), the patient's preoperative and postoperative abdominal CT non-contrast scan images are extracted and used as the raw data for deep learning model processing. Specifically, the image data extraction can be automated by interface connecting to the PACS system, avoiding the tedious manual export operation.

[0023] Then, data preprocessing is performed to optimize the raw image data and ensure the accuracy of subsequent segmentation and the effectiveness of preoperative and postoperative comparisons. First, single-slice axial CT images at the vertebral body level of the third lumbar vertebra are selected. Second, the Iterative Closest Point (ICP) algorithm based on skeletal contour features is used to register the preoperative and postoperative CT images of the same patient, ensuring precise spatial alignment and eliminating interference caused by differences in scanning position. Finally, image normalization is performed, standardizing the pixel values ​​of the CT images to [-1000, 400] HU to ensure that the feature information of various tissues can be preserved and effectively analyzed, providing a standardized data format for subsequent model input.

[0024] After data preprocessing, automatic image segmentation can be performed, achieving high-precision segmentation of the total abdominal muscles, visceral fat, and subcutaneous fat within the abdominal cavity. This embodiment is based on a body composition analysis model with an improved multi-head attention mechanism. This model is based on the 3D-Unet framework and retains the encoder-decoder symmetrical structure (the encoder is responsible for downsampling to extract deep features of the image, and the decoder is responsible for upsampling to restore the image resolution and output the segmentation results). In each feature extraction stage of the encoder (a total of 4 layers), a multi-head attention module is added. This module calculates multiple attention heads in parallel, focusing on features of different scales and anatomical structures within the abdominal cavity (muscle fiber texture features, gray-level difference features between visceral fat and organ boundaries, and continuous distribution features of subcutaneous fat). It assigns higher weights to key features, suppresses the interference of background noise (such as intestinal gas and scanning artifacts) on segmentation, and enhances the model's ability to learn complex abdominal structures.

[0025] The body composition analysis model was pre-trained using a large amount of labeled single-slice axial CT images of the third lumbar vertebra to accurately identify different tissue regions. The pre-processed CT images were then input into the trained model. (See [link to relevant documentation]). Figure 2 As shown, the model can automatically generate the boundaries of skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA). Specifically, SMA corresponds to the regions of the rectus abdominis, external oblique, internal oblique, transversus abdominis, quadratus lumborum, psoas major, and erector spinae muscles; VFA is the adipose tissue region surrounding the abdominal viscera; and SFA is the continuous adipose tissue region between the skin and the muscle layer.

[0026] It is worth noting that image registration and normalization during data preprocessing ensure the comparability of pre- and post-operative images. Combined with gender-specific index calculations from the body composition analysis module and risk level classification from the outcome assessment module, post-operative risk stratification analysis of patients can be achieved. Compared to traditional assessments that rely solely on weight changes, this stratification method can more accurately identify high-risk patients and solves the problem that traditional assessments cannot distinguish between muscle and adipose tissue quality and are difficult to accurately determine risk.

[0027] A body composition analysis model employing an improved multi-head attention mechanism replaces the traditional manual delineation method. On the one hand, this significantly reduces the workload of doctors, eliminating the need for extensive manual image segmentation and greatly improving the efficiency of body composition analysis, thus meeting the assessment needs of a large number of patients in clinical practice. On the other hand, the model is trained on a large amount of labeled data, and the segmentation process is not affected by subjective human factors, effectively reducing human error and improving the accuracy of skeletal muscle and adipose tissue segmentation and related index calculations, solving the problems of low efficiency and large errors in the traditional manual method.

[0028] S2. Based on the SMA, VFA and SFA mentioned before and after the operation, calculate the skeletal muscle index SMI = SMA / target height² and the visceral fat to subcutaneous fat ratio VSR = VFA / SFA respectively. S3. Based on the gender of the target subject and the SMI and VSR mentioned before and after the operation, determine whether it is sarcopenia and / or visceral fat obesity. In this embodiment, threshold values ​​for the indicators are set according to the gender of the target object. For males, if the SMI is less than the first threshold, they are diagnosed with sarcopenia; if the VSR is greater than or equal to the second threshold, they are diagnosed with visceral fat obesity. For females, if the SMI is less than the third threshold, they are diagnosed with sarcopenia; if the VSR is greater than or equal to the fourth threshold, they are diagnosed with visceral fat obesity. Preferably, the first threshold can be set to 39.4 cm² / m², the second threshold to 0.6, the third threshold to 32.5 cm² / m², and the fourth threshold to 1.1. Of course, the thresholds can be adjusted adaptively as needed, and this embodiment does not impose any limitations on this.

[0029] In practice, the following formulas are used for calculation: ① Skeletal Muscle Index (SMI) = SMA / height² (unit: cm² / m²), and sarcopenia is determined based on gender-specific reference standards. Specifically, sarcopenia is defined as an SMI < 39.4 cm² / m² for men and an SMI < 32.5 cm² / m² for women; ② Visceral Fat to Subcutaneous Fat Ratio (VSR) = VFA / SFA. Visceral fat obesity is also assessed based on gender reference standards. Visceral fat obesity is defined as a VSR ≥ 0.6 for men and a VSR ≥ 1.1 for women.

[0030] S4. Based on the preoperative and postoperative assessment results, classify the risk level.

[0031] Understandably, by comparing the SMI and VSR results before and after surgery, the patient's postoperative risk level can be determined. In this embodiment, the risk levels are defined as follows: ① High risk: sarcopenia and / or visceral fat obesity both before and after surgery; ② Medium risk: sarcopenia and / or visceral fat obesity both before or after surgery; ③ Low risk: no sarcopenia or visceral fat obesity both before and after surgery. A risk level report is then generated to assist physicians in assessing the patient's long-term survival risk, thereby guiding the development of follow-up plans and the selection of interventional treatment strategies.

[0032] It should be noted that the risk assessment results output above can help clinicians intuitively understand the long-term survival risk of patients after surgery, and provide an objective and reliable basis for adjusting subsequent treatment plans (such as developing nutritional support plans for patients with sarcopenia) and developing follow-up plans (such as shortening the follow-up interval for high-risk patients). This fills the gap in the clinical lack of an efficient, accurate and automated risk assessment system for colorectal cancer surgery, and helps to improve the quality of postoperative management of patients and enhance long-term treatment outcomes.

[0033] See Figure 3 As shown, the present invention will be further illustrated below through embodiments. Taking a male patient as an example, the implementation process is as follows: Data acquisition: ① Clinical data: Extracted patient gender (male) and height (1.74m); ② Imaging data: Extracted abdominal CT scan images from PACS before surgery (July 2015) and 12 months after surgery (August 2016).

[0034] Data preprocessing: ① Image screening: Single-layer axial CT images of the third lumbar vertebra at two time points were manually selected; ② Image registration: ICP registration was performed on the CT images 12 months after surgery, based on the preoperative vertebral body outline of the third lumbar vertebra; ③ Image normalization: The pixel values ​​of the two images were normalized to [-1000, 400] HU.

[0035] Automatic image segmentation: Input two preprocessed images into the model and output the body component segmentation results: ① Preoperative: SMA=118.3 cm², VFA=205.5 cm², SFA=141.6 cm²; ② 12 months postoperative: SMA=111.6 cm², VFA=135.7 cm², SFA=117.2 cm².

[0036] Body composition analysis: ① Calculation indicators: Preoperative SMI=118.3 / (1.74)²≈39.1 cm² / m² (<39.4, sarcopenia present), VSR=205.5 / 141.6≈1.45 (≥0.6, visceral fat obesity present); 12 months postoperatively SMI=111.6 / (1.74)²≈36.9 cm² / m² (sarcopenia present), VSR=135.7 / 117.2≈1.2 (visceral fat obesity present); ② Output results: Sarcopenia and visceral fat obesity were present both preoperatively and 12 months postoperatively.

[0037] Outcome assessment: Based on the body composition analysis, a postoperative risk level report was generated for the patient, and the patient was assessed as high risk.

[0038] In summary, the postoperative risk assessment method for colorectal cancer in this application is based on a multi-head attention mechanism. It segments the vertebral body-level CT images of the third lumbar vertebra of the target subject before and after surgery to obtain the skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative values. Based on the preoperative and postoperative SMA, VFA, and SFA, the skeletal muscle index (SMI) and the visceral fat to subcutaneous fat ratio (VSR) are calculated, respectively. Based on the target subject's gender and the preoperative and postoperative SMI and VSR, it is determined whether the subject has sarcopenia and / or visceral fat obesity. Based on the preoperative and postoperative assessment results, a risk level is assigned.

[0039] The body composition analysis method using a multi-head attention mechanism replaces the traditional manual delineation method. On the one hand, it significantly reduces the workload of doctors, eliminating the need for a lot of time to manually segment images and significantly improving the efficiency of body composition analysis. On the other hand, the segmentation process is not affected by human subjective factors, effectively reducing human error and improving the accuracy of skeletal muscle and adipose tissue segmentation and related index calculation.

[0040] Secondly, embodiments of this application also provide a postoperative risk assessment device for colorectal cancer.

[0041] In one embodiment, reference is made to Figure 4 , Figure 4 This is a schematic diagram of the functional modules of an embodiment of the colorectal cancer postoperative risk assessment device of this application. Figure 4As shown, the postoperative risk assessment device for colorectal cancer includes a segmentation module, an analysis module, and an assessment module.

[0042] The segmentation module is used to segment the vertebral body level CT images of the third lumbar vertebra of the target object before and after surgery based on the multi-head attention mechanism, and obtain the skeletal muscle area (SMA), visceral fat area (VFA) and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative conditions. The analysis module is used to calculate the skeletal muscle index SMI=SMA / target subject height² and the visceral fat to subcutaneous fat ratio VSR=VFA / SFA based on the SMA, VFA and SFA before and after the operation; and to determine whether the target subject has sarcopenia and / or visceral fat obesity based on the gender of the target subject and the SMI and VSR before and after the operation. The assessment module is used to classify risk levels based on pre- and post-operative assessment results.

[0043] Further, in one embodiment, the analysis module determines whether the target subject has sarcopenia and / or visceral fat obesity based on the target subject's gender and the preoperative and postoperative SMI and VSR, including: The target group is male. If the SMI is less than the first threshold, it is judged as sarcopenia. If the VSR is greater than or equal to the second threshold, it is judged as visceral fat obesity. The target group is women. If the SMI is less than the third threshold, it is judged as sarcopenia. If the VSR is greater than or equal to the fourth threshold, it is judged as visceral fat obesity.

[0044] Furthermore, in one embodiment, the first threshold is 39.4 cm² / m², the second threshold is 0.6, the third threshold is 32.5 cm² / m², and the fourth threshold is 1.1.

[0045] Furthermore, in one embodiment, the assessment module classifies risk levels based on pre- and post-operative assessment results, including: If sarcopenia and / or visceral fat obesity are present both before and after surgery, the risk is high. If sarcopenia and / or visceral fat obesity are present before or after surgery, the risk level is medium. If there is no sarcopenia or visceral fat obesity before and after the operation, the risk is low.

[0046] Further, in one embodiment, the segmentation module, based on a multi-head attention mechanism, segments the preoperative and postoperative vertebral CT images of the third lumbar vertebra of the target object to obtain the skeletal muscle area (SMA), visceral fat area (VFA), and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative data, including: The vertebral body level CT images of the third lumbar vertebra of the target object before and after surgery were input into the body composition analysis model based on the improved multi-head attention mechanism for segmentation, and the skeletal muscle area (SMA), visceral fat area (VFA) and subcutaneous fat area (SFA) corresponding to the preoperative and postoperative conditions were obtained. The body composition analysis model is configured as follows: Based on a 3D segmentation network, an encoder is configured to downsample and extract deep features from the image, and a decoder is configured to upsample and restore the image resolution and output the segmentation results. In addition, a multi-head attention module is set in each feature extraction stage of the encoder.

[0047] Furthermore, in one embodiment, an image processing module is also included. Before segmenting the preoperative and postoperative vertebral body level CT images of the target object's third lumbar vertebra based on the multi-head attention mechanism, the image processing module is used to: Extract preoperative and postoperative abdominal CT non-contrast scan images of the target subject; Based on preoperative and postoperative abdominal CT non-contrast scan images, vertebral level CT images of the third lumbar vertebra were selected from the preoperative and postoperative images of the target subjects.

[0048] Furthermore, in one embodiment, the image processing module is also used for: Spatial registration was performed on the preoperative and postoperative vertebral CT images of the third lumbar vertebra of the target subject. The registered image is normalized to the set HU range.

[0049] The functions of each module in the above-mentioned colorectal cancer postoperative risk assessment device correspond to the steps in the above-mentioned colorectal cancer postoperative risk assessment method embodiment, and their functions and implementation processes will not be described in detail here.

[0050] Thirdly, embodiments of this application provide a postoperative risk assessment device for colorectal cancer. The postoperative risk assessment device for colorectal cancer can be a personal computer (PC), laptop computer, server, or other device with data processing capabilities.

[0051] Reference Figure 5 , Figure 5 This is a schematic diagram of the hardware structure of the postoperative risk assessment device for colorectal cancer involved in the embodiments of this application. In the embodiments of this application, the postoperative risk assessment device for colorectal cancer may include a processor, a memory, a communication interface, and a communication bus.

[0052] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.

[0053] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting internal components of the colorectal cancer postoperative risk assessment device, as well as interfaces used for interconnecting the colorectal cancer postoperative risk assessment device with other devices (such as other computing devices or user equipment). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can be displays, keyboards, etc.

[0054] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0055] The processor can be a general-purpose processor, which can call the postoperative risk assessment program for colorectal cancer stored in memory and execute the postoperative risk assessment method for colorectal cancer provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the postoperative risk assessment program for colorectal cancer is called can refer to the various embodiments of the postoperative risk assessment method for colorectal cancer in this application, and will not be repeated here.

[0056] Those skilled in the art will understand that Figure 5 The hardware structure shown does not constitute a limitation of this application and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0057] Fourthly, embodiments of this application also provide a readable storage medium.

[0058] This application has a readable storage medium storing a postoperative risk assessment program for colorectal cancer, wherein when the postoperative risk assessment program for colorectal cancer is executed by a processor, it implements the steps of the postoperative risk assessment method for colorectal cancer as described above.

[0059] The method used when performing the postoperative risk assessment procedure for colorectal cancer can be referred to in the various embodiments of the postoperative risk assessment method for colorectal cancer of this application, and will not be repeated here.

[0060] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0061] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0062] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0063] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0064] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0065] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0066] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method of risk assessment after colorectal cancer surgery, characterized in that, The colorectal cancer postoperative risk assessment method comprises: Based on the multi-head attention mechanism, the CT images of the target object before and after the operation are segmented at the level of the vertebral body of the third lumbar vertebra, and the skeletal muscle area SMA, visceral fat area VFA and subcutaneous fat area SFA corresponding to the preoperative and postoperative are obtained; According to the SMA, VFA and SFA before and after the operation, the skeletal muscle index SMI=SMA / height of the target object² and the visceral fat to subcutaneous fat ratio VSR=VFA / SFA are calculated respectively; According to the gender of the target object and the SMI and VSR before and after the operation, it is determined whether it is muscle loss and / or visceral fat obesity; According to the determination results before and after the operation, the risk level is divided.

2. The method for risk assessment of colorectal cancer after surgery according to claim 1, wherein, The determination of whether it is muscle loss and / or visceral fat obesity according to the gender of the target object and the SMI and VSR before and after the operation comprises: If the SMI of the target object is less than the first threshold value, it is determined to be muscle loss, and if the VSR is greater than or equal to the second threshold value, it is determined to be visceral fat obesity; If the SMI of the target object is less than the third threshold value, it is determined to be muscle loss, and if the VSR is greater than or equal to the fourth threshold value, it is determined to be visceral fat obesity.

3. The colorectal cancer postoperative risk assessment method of claim 2, wherein: The first threshold value is 39.4 cm² / m², the second threshold value is 0.6, the third threshold value is 32.5 cm² / m², and the fourth threshold value is 1.

1.

4. The method of assessing risk of colorectal cancer post-surgery of claim 1, wherein, The division of the risk level according to the determination results before and after the operation comprises: If there is muscle loss and / or visceral fat obesity before and after the operation, it is high risk; If there is muscle loss and / or visceral fat obesity before or after the operation, it is medium risk; If there is no muscle loss and visceral fat obesity before and after the operation, it is low risk.

5. The method of assessing risk of colorectal cancer post-surgery of claim 1, wherein, The segmentation of the CT images of the target object before and after the operation at the level of the vertebral body of the third lumbar vertebra based on the multi-head attention mechanism comprises: Input the CT images of the target object before and after the operation at the level of the vertebral body of the third lumbar vertebra into the body composition analysis model based on the improved multi-head attention mechanism for segmentation to obtain the skeletal muscle area SMA, visceral fat area VFA and subcutaneous fat area SFA corresponding to the preoperative and postoperative; Wherein, the body composition analysis model is configured as: Based on the 3D segmentation network as the basic framework, the encoder is configured to downsample to extract deep image features, and the decoder is configured to upsample to restore image resolution and output segmentation results, and each layer of the encoder is provided with a multi-head attention module.

6. The method of assessing risk of colorectal cancer post-surgery of claim 1, wherein, Before the segmentation of the CT images of the target object before and after the operation at the level of the vertebral body of the third lumbar vertebra based on the multi-head attention mechanism, it further comprises: Extract the abdominal CT non-enhanced scan images of the target object before and after the operation; Based on the abdominal CT non-enhanced scan images before and after the operation, the CT images of the target object before and after the operation at the level of the vertebral body of the third lumbar vertebra are screened.

7. The method of assessing the risk of colorectal cancer post surgery as claimed in claim 6, wherein, Further comprising: perform spatial registration on the preoperative and postoperative CT images of the vertebral body level of the third lumbar vertebra of the target object; standardize the registered images to a set HU range.

8. A colorectal cancer post-operative risk assessment device, characterized in that, The colorectal cancer postoperative risk assessment device comprises: a segmentation module configured to segment the preoperative and postoperative CT images of the vertebral body level of the third lumbar vertebra of the target object based on a multi-head attention mechanism to obtain a skeletal muscle area SMA, a visceral fat area VFA, and a subcutaneous fat area SFA corresponding to the preoperative and postoperative bone; an analysis module configured to calculate a skeletal muscle index SMI=SMA / height of the target object and a visceral fat to subcutaneous fat ratio VSR=VFA / SFA based on the preoperative and postoperative SMA, VFA, and SFA, and determine whether the target object has sarcopenia and / or visceral obesity based on the gender of the target object and the preoperative and postoperative SMI and VSR; an assessment module configured to divide the target object into risk levels based on the determination results of the preoperative and postoperative target object.

9. A colorectal cancer postoperative risk assessment device, characterized by, The colorectal cancer postoperative risk assessment device comprises a processor, a memory, and a colorectal cancer postoperative risk assessment program stored on the memory and executable by the processor, wherein the colorectal cancer postoperative risk assessment program, when executed by the processor, implements the steps of the colorectal cancer postoperative risk assessment method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a colorectal cancer postoperative risk assessment program, wherein the colorectal cancer postoperative risk assessment program, when executed by a processor, implements the steps of the colorectal cancer postoperative risk assessment method according to any one of claims 1 to 7.

Images