A rectal cancer radiotherapy reaction analysis method and system based on image processing

Abstract

The application discloses a rectal cancer radiotherapy reaction analysis method based on image processing, which comprises the following steps: collecting image information of rectal cancer radiotherapy; radiotherapy reaction analysis model; collecting picture information of rectal cancer of a patient; pre-processing the picture information; extracting features; radiotherapy simulation; determining the change of the diameter of rectal cancer after radiotherapy; a rectal cancer radiotherapy system based on image processing, comprising: a model building information acquisition module; a radiotherapy reaction analysis model building module; a rectal cancer data information acquisition module; a picture information preprocessing module; a picture feature extraction module; a radiotherapy prediction module; a radiotherapy scheme module; the application extracts features in picture information through a convolutional neural network algorithm, and simulates treatment of rectal cancer according to the feature information; through multiple simulation processes, the diameter of rectal cancer after multiple simulation treatments is obtained; whether the diameter of rectal cancer after simulation treatment is within a threshold value is determined, and then the treatment scheme is determined.

Description

Technical Field

[0001] This invention relates to the field of radiotherapy technology for rectal cancer, specifically to a method and system for analyzing radiotherapy response to rectal cancer based on image processing. Background Technology

[0002] Colorectal cancer is a common malignant tumor. With changes in residents' lifestyles and dietary structures, the incidence of colorectal cancer is on the rise, ranking third in incidence and fifth in mortality among malignant tumors.

[0003] Traditional imaging methods have limited value in the early diagnosis, prognosis, and prediction of radiotherapy and chemotherapy effects of colorectal cancer. In recent years, with the continuous development of image processing and computer technology, radiomics analysis methods have been applied to accurately assess colorectal cancer patients before surgery and predict treatment efficacy.

[0004] Radiomics is an emerging technology that extracts subtle quantitative features from medical images and analyzes the relationship between these features and clinical and genetic data to build predictive models. This technology can be used not only for automated tumor staging but also for predicting patient survival.

[0005] Deep learning-based radiomics differs from traditional radiomics in that the constructed models can automatically learn to extract and select image features and make predictions, thereby more comprehensively and deeply mining information from images. Currently, the most commonly used methods in radiomics are convolutional neural networks and sparse autoencoders. Compared to traditional radiomics, deep learning has the following advantages: First, traditional radiomics features are general but lack specificity; for example, some features are difficult to apply to deformable tumors like colorectal cancer. Second, traditional radiomics requires explicit computational formulas, thus limiting the definition of features, while deep learning uses convolutional neural networks to extract features and learns from data continuously, thus possessing strong specificity.

[0006] Chinese invention patent application CN117238488A discloses a method and system for analyzing radiotherapy responses to rectal cancer based on image analysis, belonging to the field of radiotherapy response technology. More specifically, it relates to a method and system for analyzing radiotherapy responses to rectal cancer based on image analysis. This invention acquires human body data and radiotherapy data, inputs the preprocessed human body data and radiotherapy data into a trained neural network prediction model to obtain the relevant response levels after radiotherapy; then, it acquires images of all disease areas during the radiotherapy process, and obtains the volume or area of ​​the disease areas based on the disease area images, establishes a time-series disease area change map based on the volume or area; and outputs a decision guidance table based on the disease area change map and the relevant response levels after radiotherapy.

[0007] The technical solutions described in the above applications obtain the human body's tolerance for radiotherapy based on the relevant reaction levels after radiotherapy, obtain the changes in the disease area based on the disease area change map, and then provide decision guidance for medical staff to make decisions on subsequent radiotherapy treatment plans through a decision guidance table. However, in combination with existing technologies and the above applications, the convolutional neural network algorithm extracts features from rectal cancer image information to obtain various data information of rectal cancer, but it still requires medical staff to determine the subsequent data plan, and cannot achieve multiple simulations of treatment effects, etc.

[0008] Therefore, the present invention provides a method and system for analyzing radiotherapy response to rectal cancer based on image processing. Summary of the Invention

[0009] (a) Technical problems to be solved

[0010] To address the shortcomings of existing technologies, this invention provides a method and system for analyzing radiotherapy responses to rectal cancer based on image processing. The method employs a convolutional neural network algorithm to process images of a patient's rectal cancer. First, the image information is preprocessed to improve accuracy and clarity. Then, the convolutional neural network algorithm extracts features from the image information and simulates treatment based on these features. Through multiple simulations, the diameter of the rectal cancer after each simulated treatment is obtained. The diameter of the rectal cancer after simulated treatment is then used to determine whether it falls within a threshold, thereby determining the appropriate treatment plan. This solves the technical problems described in the background section.

[0011] (II) Technical Solution

[0012] To achieve the above objectives, the present invention provides the following technical solution:

[0013] A method for analyzing radiotherapy response to rectal cancer based on image processing, comprising,

[0014] Image information from rectal cancer radiotherapy was collected, and a large amount of data from rectal cancer radiotherapy was collected from the hospital. The data from rectal cancer radiotherapy was divided into two groups, namely group A and group B. Group A was the data group for model building and group B was the data group for model testing.

[0015] The radiotherapy response analysis model uses a convolutional neural network algorithm.

[0016] The input is: V = conv2(W, X,"valid") + b.

[0017] The output is: In this system, the input and output formulas are the same for each convolutional layer, and each convolutional layer has a different weight matrix W, where W, X, and Y are in matrix form. Let the last fully connected layer be the L-th layer, and let its output be a vector y.L The expected output is d.

[0018] The total error formula is then:

[0019] Here, `conv2()` is a Matlab function for convolution operations. The third parameter, `valid`, specifies the type of convolution operation. `W` is the kernel matrix, `X` is the input matrix, and `b` is the bias. It is the activation function, where d and y in the total error are the expected output and the network output vectors, respectively, and ||x||2 represents the 2-norm of vector x, calculated as follows:

[0020] The patient's rectal cancer images were collected using medical equipment, and the image data was input into a radiotherapy response analysis model for calculation.

[0021] Furthermore, before inputting the patient's rectal cancer images into the radiotherapy response analysis model, the image information is preprocessed by converting the images to grayscale, removing interference noise, and normalizing the images.

[0022] The radiotherapy response analysis model extracts features from image information and makes radiotherapy predictions based on these features.

[0023] First, input the number of radiotherapy sessions and the duration. In each iteration, initialize the number of radiotherapy sessions and the duration to be equal to the predicted number of radiotherapy sessions and the duration. Then, update the number of radiotherapy sessions and the duration, setting the number of radiotherapy sessions as z and the duration as t. The update method formula is as follows:

[0024]

[0025] Among them, D z D t These represent the number of radiotherapy sessions and the duration of radiotherapy in this iteration, respectively. p D and D represent the predicted dose and adjusted dose, respectively; m, n, and k represent the diameter of the rectal cancer; and D(k) represents the calculated diameter of the rectal cancer. p (k) represents the predicted diameter after rectal cancer treatment;

[0026] Based on the number of radiotherapy sessions, radiotherapy duration, and a combination of both, the change in rectal cancer diameter after radiotherapy is predicted. Taking into account the patient's age, gender, and survival status, the required number of radiotherapy sessions to reduce the rectal cancer diameter to a threshold range is determined, and this threshold range is set as j. min <D(k)<j max Where j is the value of rectal cancer after treatment, j min j is the minimum value for rectal cancer after treatment. max This represents the maximum value for rectal cancer after radiotherapy.

[0027] Furthermore, the data information for rectal cancer radiotherapy includes historical data and images of rectal cancer radiotherapy patients from various hospitals. Group A data contains no fewer than 200 data points, and Group B data contains no fewer than 50 data points. The radiotherapy response analysis model is tested using Group B data, and the model is allowed to learn autonomously, improving the accuracy of feature extraction from the image information.

[0028] Furthermore, in the radiotherapy response analysis model, the convolution kernel is set to 2×2, and the sliding stride of the convolution kernel in the input matrix is ​​2. The number of zero-padding layers is set to q. The number of zero-padding layers is adjusted according to the size of the convolution kernel, the stride, and the size of the input matrix so that the convolution kernel slides exactly to the edge and the sliding stride is greater than the size of the convolution kernel.

[0029] Let the input matrix size be w, the convolution kernel size be v, the stride be s, and the number of zero-padding layers be q. Then the formula for calculating the size of the feature map generated after convolution is:

[0030] Where w' is the size of the feature map.

[0031] Furthermore, the collection of patients' rectal cancer image information includes MRI images or computed tomography (CT) scans, and semantic analysis is performed based on the MRI images or CT scans, i.e., describing the rectal cancer and annotating the rectal cancer area. The MRI images or CT scans are input into the computer and manually annotated to select the area where the tumor is located and annotate the areas of suspected rectal cancer.

[0032] Furthermore, grayscale processing is used to convert image information into a grayscale image. That is, when R=G=B, the color represents a grayscale color, where the value of R=G=B is called the grayscale value.

[0033] The weighted average method is used for calculation:

[0034] f(i,j)=0.30R(i,j)+0.59G(i,j)+011B(i,j), where i and j represent the position coordinates in the image information.

[0035] Furthermore, interference noise is removed using a two-dimensional Gaussian function for calculation.

[0036] Here, the value of G(x,y) is the weight on the matrix, and x and y are the coordinate positions in the image information, and σ is a set value. Combined with the convolution kernel, σ is set to 3.0, and the calculated weights are normalized.

[0037] Furthermore, normalization is used to map the weights to the interval [0, 1], which facilitates the radiotherapy response analysis model in calculating and processing the image information;

[0038]

[0039] Where g is the result of weight normalization of the two-dimensional Gaussian function calculation, and the output range is between 0 and 1. (x,y) The weights G are calculated for a two-dimensional Gaussian function. (x,y)min G is the minimum value among the weights calculated for a two-dimensional Gaussian function. (x,y)max This is the maximum value among the weights calculated for a two-dimensional Gaussian function.

[0040] Furthermore, the relationship between the initial number of radiotherapy sessions and the set predicted dose is determined by comparing the initial value of the number of radiotherapy sessions with the set value of the predicted dose. If the initial number of radiotherapy sessions is greater than the predicted dose, the dose is adjusted to the initial number of radiotherapy sessions; otherwise, the dose is adjusted to the predicted dose.

[0041] Similarly, the relationship between the initial radiotherapy duration and the set predicted duration is determined by judging the initial value of the radiotherapy duration and the set value of the predicted dose. If the initial radiotherapy duration is longer than the predicted duration, the dose is adjusted to the initial radiotherapy duration; otherwise, the dose is adjusted to the predicted duration.

[0042] By combining the number of radiotherapy sessions and the duration of radiotherapy, the diameter of rectal cancer is measured, and the change in the diameter of rectal cancer during treatment is predicted. That is, if the diameter of rectal cancer after treatment is larger than the predicted diameter, the number of radiotherapy sessions and the duration of radiotherapy are changed again; if the diameter of rectal cancer after treatment is smaller than the predicted diameter, the number of radiotherapy sessions and the duration of radiotherapy are changed again.

[0043] Furthermore, by combining the patient's age, gender, and survival status, the model determines how many radiotherapy sessions and for how long the patient can undergo, while maintaining the patient's vital signs. For older patients, those with poorer survival, or who are female, the dosage is appropriately reduced; for younger patients, those with better survival, or who are male, the dosage is appropriately increased. This allows the radiotherapy response analysis model to minimize the predicted diameter of rectal cancer under the predicted treatment conditions.

[0044] An image processing-based radiotherapy system for rectal cancer includes:

[0045] The model building information acquisition module acquires data on rectal cancer radiotherapy within hospitals and queries data on rectal cancer radiotherapy from various hospitals online through big data analysis to improve the data sample size. The collected rectal cancer radiotherapy data is divided into group A and group B, with group A being the model building data group and group B being the model testing data group.

[0046] The radiotherapy response analysis model building module uses data from group A as the basic data and data from group B for learning and testing. This is used to test the accuracy of the radiotherapy response analysis model and to provide the model with a large amount of simulation learning.

[0047] The rectal cancer data information acquisition module examines patients using medical equipment in the hospital, acquires MRI images or computed tomography scans, performs semantic analysis and description of rectal cancer, and marks rectal tumor areas on MRI or computed tomography images to facilitate accurate feature extraction.

[0048] The image information preprocessing module performs grayscale conversion, noise removal, and normalization on the patient's rectal cancer images to improve the clarity and accuracy of the image information.

[0049] The image feature extraction module extracts features from image information using a radiotherapy response analysis model.

[0050] The radiotherapy prediction module predicts the patient's radiotherapy process through a radiotherapy response analysis model. By iterating the number of radiotherapy sessions and the duration of radiotherapy, it determines the predicted diameter of rectal cancer and the calculated diameter of rectal cancer, and then determines the treatment plan.

[0051] The radiotherapy protocol module records multiple simulation data from the radiotherapy response analysis model and assigns values ​​based on a set threshold range. min <D(k)<j max The calculation determines whether the rectal cancer diameter falls within a threshold range, thus determining the feasibility of the treatment plan. If the rectal cancer diameter is within the threshold range multiple times, the treatment plan is feasible; if the rectal cancer diameter is greater than the threshold range multiple times, the treatment plan is not feasible. max If so, then the treatment plan is not feasible.

[0052] (III) Beneficial Effects

[0053] This invention provides a method and system for analyzing radiotherapy response to rectal cancer based on image processing, which has the following beneficial effects:

[0054] This invention constructs a radiotherapy response analysis model by querying a large amount of rectal cancer radiotherapy data. The large amount of rectal cancer radiotherapy data is classified, and the radiotherapy response analysis model is constructed using group A data as the basic data. Group B data is used for learning and detection, which is used to detect the accuracy of the radiotherapy response analysis model and to provide the radiotherapy response analysis model with a large amount of simulation learning.

[0055] The patient's rectal cancer information is processed by converting rectal cancer images to grayscale, removing interference noise, and normalizing the images to improve image clarity and accuracy. This facilitates feature extraction in the radiotherapy response analysis model, enabling the acquisition of various data information from rectal cancer patients. Then, a convolutional neural network algorithm is used to extract features from the images, and based on these features, simulated treatment is performed on the rectal cancer. Through multiple simulations, the diameter of the rectal cancer after each simulated treatment is obtained. The diameter is then used to determine whether it falls within a threshold, thus determining the appropriate treatment plan. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating the computer-based method for analyzing radiotherapy responses to rectal cancer according to the present invention.

[0057] Figure 2 This is a schematic diagram of the structure of the rectal cancer radiotherapy system of the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Please see Figure 1 This invention provides a method for analyzing the radiotherapy response to rectal cancer based on image processing, comprising:

[0060] Step 1: Collect image information of rectal cancer radiotherapy. Collect a large amount of data on rectal cancer radiotherapy from the hospital and divide the data into two groups, namely Group A and Group B. Group A is the data group for model building and Group B is the data group for model testing.

[0061] The data information for rectal cancer radiotherapy includes data and images of patients with a history of rectal cancer radiotherapy in the hospital. There are no fewer than 200 data points in Group A and no fewer than 50 data points in Group B. The radiotherapy response analysis model is tested using the data from Group B and is allowed to learn autonomously to improve the accuracy of feature extraction from the image information.

[0062] When using this technology, collecting a large amount of rectal cancer data can provide basic data support for the radiotherapy response analysis model, facilitate the testing of the model, enable the model to learn from a large amount of data, test its accuracy, and improve its learning ability.

[0063] Step 2: The radiotherapy response analysis model uses a convolutional neural network algorithm.

[0064] The input is: V = conv2(W, X,"valid") + b.

[0065] The output is: In this system, the input and output formulas are the same for each convolutional layer, and each convolutional layer has a different weight matrix W, where W, X, and Y are in matrix form. Let the last fully connected layer be the L-th layer, and let its output be a vector y. L The expected output is d.

[0066] The total error formula is then:

[0067] Here, `conv2()` is a Matlab function for convolution operations. The third parameter, `valid`, specifies the type of convolution operation. `W` is the kernel matrix, `X` is the input matrix, and `b` is the bias. It is the activation function, where d and y in the total error are the expected output and the network output vectors, respectively, and ||x||2 represents the 2-norm of vector x, calculated as follows:

[0068] In the radiotherapy response analysis model, the convolution kernel is set to 2×2, and the sliding stride of the convolution kernel in the input matrix is ​​2. The number of zero-padding layers is set to q. The number of zero-padding layers is adjusted according to the size of the convolution kernel, the stride, and the size of the input matrix so that the convolution kernel slides exactly to the edge and the sliding stride is greater than the size of the convolution kernel.

[0069] Let the input matrix size be w, the convolution kernel size be v, the stride be s, and the number of zero-padding layers be q. Then the formula for calculating the size of the feature map generated after convolution is:

[0070] Where w' is the size of the feature map;

[0071] When in use, the radiotherapy response analysis model is supported by a convolutional neural network algorithm, enabling the model to extract features from the patient's rectal cancer images and obtain various data information about the patient's rectal cancer. In order to ensure that the rectal cancer images can be completely scanned by the convolutional kernel, the sliding stride is set for processing.

[0072] Step 3: Collect images of the patient's rectal cancer. Use medical equipment to collect images of the patient's rectal cancer and input the image data into the radiotherapy response analysis model for calculation.

[0073] The process involves collecting images of patients with rectal cancer, including MRI images or computed tomography (CT) scans. Semantic analysis is then performed based on the MRI images or CT scans to describe the rectal cancer and to annotate the rectal cancer regions. The MRI images or CT scans are input into a computer and manually annotated to select the areas where the tumor is located and to annotate any suspected rectal cancer areas.

[0074] When using this technology, images of rectal cancer from various patients are collected based on existing medical techniques. This facilitates accurate detection and comparison of the radiotherapy response analysis model, allows for clear acquisition of rectal cancer data, and improves the extraction of rectal cancer features through semantic analysis and annotation of rectal cancer regions. This reduces the scanning processing time of the radiotherapy response analysis model, thus facilitating subsequent simulation of large amounts of data.

[0075] Step 4: Before inputting the patient's rectal cancer images into the radiotherapy response analysis model, the image information is preprocessed by converting it to grayscale, removing interference noise, and normalizing the image information.

[0076] S401. Grayscale processing is used to convert image information into a grayscale image. That is, when R=G=B, the color represents a grayscale color, and the value of R=G=B is called the grayscale value.

[0077] The weighted average method is used for calculation:

[0078] f(i,j)=0.30R(i,j)+0.59G(i,j)+011B(i,j), where i and j represent the position coordinates in the image information.

[0079] S402. Noise removal is performed using a two-dimensional Gaussian function.

[0080] Here, the value of G(x,y) is the weight on the matrix, and x and y are the coordinate positions in the image information, and σ is a set value. Combined with the convolution kernel, σ is set to 3.0, and the calculated weights are normalized.

[0081] S403. Normalization is used to map the weights to the interval [0, 1], which facilitates the radiotherapy response analysis model to calculate and process the image information.

[0082]

[0083] Where g is the result of weight normalization of the two-dimensional Gaussian function calculation, and the output range is between 0 and 1. (x,y) The weights G are calculated for a two-dimensional Gaussian function. (x,y)min G is the minimum value among the weights calculated for a two-dimensional Gaussian function. (x,y)max The maximum value among the weights calculated for a two-dimensional Gaussian function;

[0084] Preprocessing rectal cancer images can improve their accuracy and clarity, facilitating subsequent feature detection and extraction. Firstly, the images are converted to grayscale to aid in subsequent processing. Secondly, noise removal and normalization effectively reduce the noise in the images, further enhancing accuracy and clarity. Thirdly, normalization ensures the calculated output values ​​are within the range of 0-1, reducing image size and allowing for faster processing by the radiotherapy response analysis model.

[0085] Step 5: The radiotherapy response analysis model extracts features from the image information and makes radiotherapy predictions based on the feature information;

[0086] When using it, the convolutional neural network algorithm is used to extract features from the processed rectal cancer images, which can obtain various data information about rectal cancer, making it easier to simulate treatment later.

[0087] Step Six: First, input the number of radiotherapy sessions and duration. In each iteration, initialize the number of radiotherapy sessions and duration to be equal to the predicted number of radiotherapy sessions and duration. Then, update the number of radiotherapy sessions and duration, setting the number of radiotherapy sessions as z and the duration as t. The update method formula is as follows:

[0088]

[0089] Among them, D z D t These represent the number of radiotherapy sessions and the duration of radiotherapy in this iteration, respectively. pD and D represent the predicted dose and adjusted dose, respectively; m, n, and k represent the diameter of the rectal cancer; and D(k) represents the calculated diameter of the rectal cancer. p (k) represents the predicted diameter after rectal cancer treatment;

[0090] The relationship between the initial number of radiotherapy sessions and the set predicted dose is determined by comparing the initial value of the number of radiotherapy sessions with the set value of the predicted dose. If the initial number of radiotherapy sessions is greater than the predicted dose, the dose is adjusted to the initial number of radiotherapy sessions; otherwise, the dose is adjusted to the predicted dose.

[0091] Similarly, the relationship between the initial radiotherapy duration and the set predicted duration is determined by judging the initial value of the radiotherapy duration and the set value of the predicted dose. If the initial radiotherapy duration is longer than the predicted duration, the dose is adjusted to the initial radiotherapy duration; otherwise, the dose is adjusted to the predicted duration.

[0092] By combining the number of radiotherapy sessions and the duration of radiotherapy, the diameter of rectal cancer is detected, and the change in the diameter of rectal cancer after treatment is predicted. That is, if the diameter of rectal cancer after treatment is greater than the predicted diameter, the number of radiotherapy sessions and the duration of radiotherapy are changed again; if the diameter of rectal cancer after treatment is smaller than the predicted diameter, the number of radiotherapy sessions and the duration of radiotherapy are changed again.

[0093] When in use, by selecting the number of radiotherapy sessions and the duration of radiotherapy, and iterating on the number of radiotherapy sessions and the duration of radiotherapy, the diameter change of rectal cancer is calculated based on the combination of the number of radiotherapy sessions and the duration of radiotherapy, so as to determine the radiotherapy plan. Furthermore, based on the diameter of rectal cancer calculated from the simulated treatment of rectal cancer and the expected diameter of the treated rectal cancer, the termination condition of the iteration is determined, so as to accurately obtain the treatment plan.

[0094] Step 7: Based on the number of radiotherapy sessions, radiotherapy duration, and a combination of both, predict the change in rectal cancer diameter after radiotherapy. Considering the patient's age, gender, and survival status, determine the number of radiotherapy sessions required to reduce the rectal cancer diameter to a threshold range, and set the threshold range as j. min <D(k)<j max Where j is the value of rectal cancer after treatment, j min j is the minimum value for rectal cancer after treatment. max This represents the maximum value for rectal cancer after radiotherapy.

[0095] Combining the patient's age, gender, and survival status, the model determines the number of radiotherapy sessions and the duration of treatment based on the patient's age and survival status, while maintaining the patient's vital signs. For older patients, those with poorer survival status, or female patients, the dosage is appropriately reduced; for younger patients, those with better survival status, or male patients, the dosage is appropriately increased. This allows the radiotherapy response analysis model to minimize the diameter of rectal cancer under predicted treatment conditions.

[0096] When using this treatment, the appropriate treatment plan should be selected based on the patient's individual condition and within the range that the patient's body can tolerate, thereby improving the patient's safety.

[0097] See Figure 2 An image processing-based radiotherapy system for rectal cancer, comprising:

[0098] The model building information acquisition module acquires data on rectal cancer radiotherapy within hospitals and queries data on rectal cancer radiotherapy from various hospitals online through big data analysis to improve the data sample size. The collected rectal cancer radiotherapy data is divided into group A and group B, with group A being the model building data group and group B being the model testing data group.

[0099] The radiotherapy response analysis model building module uses data from group A as the basic data and data from group B for learning and testing. This is used to test the accuracy of the radiotherapy response analysis model and to provide the model with a large amount of simulation learning.

[0100] The rectal cancer data information acquisition module examines patients using medical equipment in the hospital, acquires MRI images or computed tomography scans, performs semantic analysis and description of rectal cancer, and marks rectal tumor areas on MRI or computed tomography images to facilitate accurate feature extraction.

[0101] The image information preprocessing module performs grayscale conversion, noise removal, and normalization on the patient's rectal cancer images to improve the clarity and accuracy of the image information.

[0102] The image feature extraction module extracts features from image information using a radiotherapy response analysis model.

[0103] The radiotherapy prediction module predicts the patient's radiotherapy process through a radiotherapy response analysis model. By iterating the number of radiotherapy sessions and the duration of radiotherapy, it determines the predicted diameter of rectal cancer and the calculated diameter of rectal cancer, and then determines the treatment plan.

[0104] The radiotherapy protocol module records multiple simulation data from the radiotherapy response analysis model and assigns values ​​based on a set threshold range. min <D(k)<j max The calculation determines whether the rectal cancer diameter falls within a threshold range, thus determining the feasibility of the treatment plan. If the rectal cancer diameter is within the threshold range multiple times, the treatment plan is feasible; if the rectal cancer diameter is greater than the threshold range multiple times, the treatment plan is not feasible. max If so, then the treatment plan is not feasible.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0106] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0108] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0109] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0110] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0111] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0112] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0113] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for analyzing radiotherapy response to rectal cancer based on image processing, characterized in that: include, Image information from rectal cancer radiotherapy was collected, and a large amount of data from rectal cancer radiotherapy was collected from the hospital. The data from rectal cancer radiotherapy was divided into two groups, namely group A and group B. Group A was the data group for model building and group B was the data group for model testing. The radiotherapy response analysis model uses a convolutional neural network algorithm. The input is: V = conv2(W, X,"valid") + b. The output is: In this system, the input and output formulas are the same for each convolutional layer, and each convolutional layer has a different weight matrix W, where W, X, and Y are in matrix form. Let the last fully connected layer be the L-th layer, and let its output be a vector y. L The expected output is d. The total error formula is then: Here, `conv2()` is a Matlab function for convolution operations. The third parameter, `valid`, specifies the type of convolution operation. `W` is the kernel matrix, `X` is the input matrix, and `b` is the bias. It is the activation function, where d and y in the total error are the expected output and the network output vectors, respectively, and ||x||2 represents the 2-norm of vector x, calculated as follows: The patient's rectal cancer images were collected using medical equipment, and the image data was input into a radiotherapy response analysis model for calculation. Furthermore, before inputting the patient's rectal cancer images into the radiotherapy response analysis model, the image information is preprocessed by converting the images to grayscale, removing interference noise, and normalizing the images. The radiotherapy response analysis model extracts features from image information and makes radiotherapy predictions based on these features. First, input the number of radiotherapy sessions and the duration. In each iteration, initialize the number of radiotherapy sessions and the duration to be equal to the predicted number of radiotherapy sessions and the duration. Then, update the number of radiotherapy sessions and the duration, setting the number of radiotherapy sessions as z and the duration as t. The update method formula is as follows: Among them, D z D t These represent the number of radiotherapy sessions and the duration of radiotherapy in this iteration, respectively. p D and D represent the predicted dose and adjusted dose, respectively; m, n, and k represent the diameter of the rectal cancer; and D(k) represents the calculated diameter of the rectal cancer. p (k) represents the predicted diameter after rectal cancer treatment; Based on the number of radiotherapy sessions, radiotherapy duration, and a combination of both, the change in rectal cancer diameter after radiotherapy is predicted. Taking into account the patient's age, gender, and survival status, the required number of radiotherapy sessions to reduce the rectal cancer diameter to a threshold range is determined, and this threshold range is set as j. min <D(k)<j max Where j is the value of rectal cancer after treatment, j min j is the minimum value for rectal cancer after treatment. max This represents the maximum value for rectal cancer after radiotherapy.

2. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 1, characterized in that: The data information for rectal cancer radiotherapy includes historical data and images of rectal cancer radiotherapy patients from various hospitals. Group A data contains no fewer than 200 data points, and Group B data contains no fewer than 50 data points. The radiotherapy response analysis model is tested using Group B data, and the model is allowed to learn autonomously to improve the accuracy of feature extraction from the image information.

3. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 1, characterized in that: In the radiotherapy response analysis model, the convolution kernel is set to 2×2, and the sliding stride of the convolution kernel in the input matrix is ​​2. The number of zero-padding layers is set to q. The number of zero-padding layers is adjusted according to the size of the convolution kernel, the stride, and the size of the input matrix so that the convolution kernel slides exactly to the edge and the sliding stride is greater than the size of the convolution kernel. Let the input matrix size be w, the convolution kernel size be v, the stride be s, and the number of zero-padding layers be q. Then the formula for calculating the size of the feature map generated after convolution is: Where w' is the size of the feature map.

4. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 3, characterized in that: The process involves collecting images of the patient's rectal cancer, including MRI or CT scans, and performing semantic analysis based on the MRI or CT scans. This includes describing the rectal cancer and annotating the rectal cancer regions. The MRI or CT scans are input into the computer and manually annotated to select the areas where the tumor is located and any suspected rectal cancer areas.

5. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 1, characterized in that: Grayscale processing is used to convert image information into a grayscale image. That is, when R=G=B, the color represents a grayscale color, and the value of R=G=B is called the grayscale value. The weighted average method is used for calculation: f(i,j)=0.30R(i,j)+0.59G(i,j)+011B(i,j), where i and j represent the position coordinates in the image information.

6. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 3, characterized in that: The removal of interference noise is calculated using a two-dimensional Gaussian function; Here, the value of G(x,y) is the weight on the matrix, and x and y are the coordinate positions in the image information, and σ is a set value. Combined with the convolution kernel, σ is set to 3.0, and the calculated weights are normalized.

7. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 6, characterized in that: Normalization is used to map the weights to the interval [0, 1], which facilitates the radiotherapy response analysis model in calculating and processing image information; Where g is the result of weight normalization of the two-dimensional Gaussian function calculation, and the output range is between 0 and 1. (x,y) The weights G are calculated for a two-dimensional Gaussian function. (x,y)min G is the minimum value among the weights calculated for a two-dimensional Gaussian function. (x,y)max This is the maximum value among the weights calculated for a two-dimensional Gaussian function.

8. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 1, characterized in that: The relationship between the initial number of radiotherapy sessions and the set predicted dose is determined by comparing the initial value of the number of radiotherapy sessions with the set value of the predicted dose. If the initial number of radiotherapy sessions is greater than the predicted dose, the dose is adjusted to the initial number of radiotherapy sessions; otherwise, the dose is adjusted to the predicted dose. Similarly, the relationship between the initial radiotherapy duration and the set predicted duration is determined by judging the initial value of the radiotherapy duration and the set value of the predicted dose. If the initial radiotherapy duration is longer than the predicted duration, the dose is adjusted to the initial radiotherapy duration; otherwise, the dose is adjusted to the predicted duration. By combining the number of radiotherapy sessions and the duration of radiotherapy, the diameter of rectal cancer is measured, and the change in the diameter of rectal cancer during treatment is predicted. That is, if the diameter of rectal cancer after treatment is larger than the predicted diameter, the number of radiotherapy sessions and the duration of radiotherapy are changed again; if the diameter of rectal cancer after treatment is smaller than the predicted diameter, the number of radiotherapy sessions and the duration of radiotherapy are changed again.

9. The method for analyzing radiotherapy response to rectal cancer based on image processing according to claim 8, characterized in that: The radiotherapy response analysis model combines the patient's age, gender, and survival status to determine the number of radiotherapy sessions and the duration of treatment, while maintaining the patient's vital signs. For older patients, those with poorer survival, or who are female, the dosage is appropriately reduced; for younger patients, those with better survival, or who are male, the dosage is appropriately increased. This allows the radiotherapy response analysis model to minimize the predicted diameter of rectal cancer under the predicted treatment conditions.

10. A rectal cancer radiotherapy system based on image processing, characterized in that: include: The model building information acquisition module acquires data on rectal cancer radiotherapy within hospitals and queries data on rectal cancer radiotherapy from various hospitals online through big data analysis to improve the data sample size. The collected rectal cancer radiotherapy data is divided into group A and group B, with group A being the model building data group and group B being the model testing data group. The radiotherapy response analysis model building module uses data from group A as the basic data and data from group B for learning and testing. This is used to test the accuracy of the radiotherapy response analysis model and to provide the model with a large amount of simulation learning. The rectal cancer data information acquisition module examines patients using medical equipment in the hospital, acquires MRI images or computed tomography scans, performs semantic analysis and description of rectal cancer, and marks rectal tumor areas on MRI or computed tomography images to facilitate accurate feature extraction. The image information preprocessing module performs grayscale conversion, noise removal, and normalization on the patient's rectal cancer images to improve the clarity and accuracy of the image information. The image feature extraction module extracts features from image information using a radiotherapy response analysis model. The radiotherapy prediction module predicts the patient's radiotherapy process through a radiotherapy response analysis model. By iterating the number of radiotherapy sessions and the duration of radiotherapy, it determines the predicted diameter of rectal cancer and the calculated diameter of rectal cancer, and then determines the treatment plan. The radiotherapy protocol module records multiple simulation data from the radiotherapy response analysis model and assigns values ​​based on a set threshold range. min <D(k)<j max The calculation determines whether the rectal cancer diameter falls within a threshold range, thus determining the feasibility of the treatment plan. If the rectal cancer diameter is within the threshold range multiple times, the treatment plan is feasible; if the rectal cancer diameter is greater than the threshold range multiple times, the treatment plan is not feasible. max If so, then the treatment plan is not feasible.

Images