Radioactive normal tissue damage probability prediction method and system based on deep learning

Abstract

The invention relates to a radioactive normal tissue damage probability prediction method and system based on deep learning, and belongs to the technical field of medical data processing. The method comprises the following steps: acquiring tumor dose data, basic information and tumor diagnosis information of a patient; the tumor dose data is formulated by a doctor according to tumor diagnosis information of a patient before radiotherapy of the patient; the basic information at least comprises the age of the patient; obtaining a dose volume histogram of the patient based on the tumor dose data; based on the dose volume histogram, acquiring volume parameters and bioequivalent dose of normal tissues of the patient; and inputting the volume parameter and the bioequivalent dose of the normal tissue of the patient, the basic information of the patient and the tumor diagnosis information into a normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the normal tissue of the patient. Compared with a traditional statistical model or empirical formula, the method is higher in prediction precision, and the damage probability of the normal tissue can be evaluated more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a method and system for predicting the probability of radiation normal tissue damage based on deep learning. Background Art

[0002] Radiotherapy is one of the important means in tumor treatment. Radiotherapy is a local treatment method that uses various rays of different energies to irradiate tumors to inhibit and kill cancer cells. These rays can destroy the internal components of human tissue cells, directly or indirectly act on DNA molecules, inhibit or kill tumor cells, and thus achieve the purpose of treatment. However, during radiotherapy, in addition to radiating and destroying tumors, the rays will also have a certain impact on normal human tissues.

[0003] However, studies have found that most of the current radiotherapy for patients' tumors is formulated by doctors based on experience, or predicted by statistical models to assist in predicting the probability of damage. In other words, the existing methods are all based on fixed statistical formulas or doctors' empirical rules to formulate radiotherapy doses, but the level of doctors varies, and they are subjective, lacking consistency and standardization in the formulation of radiotherapy doses, and also lacking accuracy. Traditional statistical models are too simple to capture complex relationships, and the accuracy of predictions is limited. In other words, the existing methods will affect the accuracy of doctors' formulation of radiotherapy doses. If the formulated radiotherapy dose is too high, the risk of complications in normal tissues will increase; and if the formulated radiotherapy dose is too low, the control effect of tumors will be affected. Summary of the invention

[0004] In order to solve the above-mentioned problems in the prior art, the present invention provides a method and system for predicting the probability of radiation normal tissue damage based on deep learning.

[0005] In a first aspect, an embodiment of the present application provides a method for predicting the probability of radiation normal tissue damage based on deep learning, comprising: obtaining tumor dose data, basic information and tumor diagnosis information of a patient; wherein the tumor dose data includes data formulated by a doctor based on the tumor diagnosis information of the patient before radiotherapy; the basic information at least includes the age of the patient; based on the tumor dose data, obtaining a dose volume histogram of the patient; based on the dose volume histogram, obtaining a volume parameter and a biological equivalent dose of the normal tissue of the patient; wherein the conversion formula of the biological equivalent dose is: ;in, represents the biological equivalent dose when the physical dose extracted from the dose-volume histogram after conversion is 2 Gy as the fractional dose; Indicates that the fractionated dose in the radiotherapy plan is , the number of divisions is The total physical dose at represents the cell proliferation correction factor, The value of is taken as 3Gy; the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information are input into the normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue.

[0006] Optionally, the tumor diagnostic information includes tumor type; the normal tissue damage probability prediction model constructed based on deep learning is a joint model group; different joint model groups are pre-constructed according to different tumor types to form a prediction model library; the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnostic information are input into the normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue, including: based on the patient's tumor type, determining a target joint model group from the prediction model library; inputting the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnostic information into the target joint model group, so as to output the damage probabilities of multiple normal tissues associated with the patient's tumor type through the target joint model group.

[0007] Optionally, there are multiple normal tissue damage probability prediction models constructed based on deep learning, and they are network models of different types; the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the basic information and tumor diagnosis information of the patient are input into the normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue, including: inputting the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the basic information and tumor diagnosis information of the patient into multiple network models, so that each network model outputs the damage probability of the patient's normal tissue; obtaining the weight of each network model; obtaining a combined damage probability based on the weight of each network model and the damage probability of the patient's normal tissue output by each network model; wherein the combined damage probability is the final obtained damage probability of the patient's normal tissue.

[0008] Optionally, obtaining the weight of each network model includes: determining the age range of the patient based on the basic information of the patient; determining the range of the patient's tumor size based on the patient's tumor diagnosis information; determining the weight of each network model according to the age range of the patient, the range of the patient's tumor size, and the model accuracy of each network model in the age range and for the range of the tumor size.

[0009] Optionally, obtaining the weight of each of the network models includes: determining the weight of each of the network models according to an input data type of the patient's tumor diagnosis information.

[0010] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a convolutional neural network model; the structure of the convolutional neural network model includes: multiple convolutional layers, batch normalization layers, dropout layers, maximum pooling layers and fully connected layers.

[0011] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a deep neural network model; the structure of the deep neural network model includes: being constructed with multiple fully connected layers as the core; and also including a batch normalization layer and a Dropout layer.

[0012] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a residual network model; the structure of the residual network model includes: being constructed with the residual block as the core; and including a convolutional layer, a batch normalization layer, an activation function layer and a fully connected layer.

[0013] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a long short-term memory network model; the structure of the long short-term memory network model includes: being constructed with multiple LSTM layers as the core; and also including a batch normalization layer and a Dropout layer.

[0014] In a second aspect, an embodiment of the present application provides a radiation normal tissue damage probability prediction system based on deep learning, comprising: a first acquisition module, used to obtain a patient's tumor dose data, basic information and tumor diagnosis information; wherein the tumor dose data includes the data prepared by the doctor based on the patient's tumor diagnosis information before radiotherapy; the basic information includes at least the patient's age; a second acquisition module, used to obtain the patient's dose volume histogram based on the tumor dose data; a third acquisition module, used to obtain the volume parameters and biological equivalent dose of the patient's normal tissue based on the dose volume histogram; wherein the conversion formula of the biological equivalent dose is: ;in, represents the biological equivalent dose when the physical dose extracted from the dose-volume histogram after conversion is 2 Gy as the fractional dose; Indicates that the fractionated dose in the radiotherapy plan is , the number of divisions is The total physical dose at represents the cell proliferation correction factor, The value of is taken as 3Gy; a prediction module is used to input the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information into a normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue.

[0015] The beneficial effects of the present invention include: First, by introducing a normal tissue damage probability prediction model based on deep learning in the embodiments of the present application, high-dimensional features can be extracted from complex data, and the potential relationship between the patient's tumor dose data, tumor diagnostic information and individual information can be fully explored. Compared with traditional statistical models or empirical formulas, this method has higher prediction accuracy and can more accurately evaluate the probability of normal tissue damage.

[0016] Second, the prediction model provided by the embodiment of the present application is based on deep learning and is combined with a prediction method of multi-dimensional information fusion analysis, which can improve data utilization and make up for the problem that the traditional model can only rely on a single data, resulting in low prediction accuracy.

[0017] Third, considering the consistency of data utilization of the prediction model using deep learning, in the embodiment of the present application, the physical dose directly obtained from the dose-volume histogram is converted in advance, and the physical dose is converted into a biological equivalent dose with 2Gy as the fractionated dose. This can more scientifically input the patient's dose distribution into the deep learning prediction model, enhance the model's predictive ability and clinical relevance, and further standardize the dose distribution of different radiotherapy schemes, making different treatment plans comparable.

[0018] Fourth, the above scheme has wide adaptability and can be applied to various tumor types and their corresponding normal tissue damage prediction needs, and has strong versatility and scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of the steps of a first method for predicting the probability of radiation normal tissue damage based on deep learning provided by an embodiment of the present invention; Figure 2 A schematic diagram of the prediction effect of the convolutional neural network model provided in an embodiment of the present invention; Figure 3 A schematic diagram of the prediction effect of the deep neural network model provided in an embodiment of the present invention; Figure 4 A schematic diagram of the prediction effect of the residual network model provided in an embodiment of the present invention; Figure 5 A schematic diagram of the prediction effect of the long short-term memory network model provided by an embodiment of the present invention; Figure 6A flowchart of the steps of a second method for predicting the probability of radiation normal tissue damage based on deep learning provided by an embodiment of the present invention; Figure 7 A flowchart of the steps of a third method for predicting the probability of radiation normal tissue damage based on deep learning provided by an embodiment of the present invention; Figure 8 A flowchart of the steps of a fourth method for predicting the probability of radiation normal tissue damage based on deep learning provided by an embodiment of the present invention; Fig. 9 A module block diagram of a radiation normal tissue damage probability prediction system based on deep learning provided for the implementation of the present invention; Fig.10 A module block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0021] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0022] Studies have found that most of the current radiotherapy for patients' tumors is determined by doctors based on experience, or by predicting the probability of damage based on statistical models as an aid. In other words, the existing methods are all based on fixed statistical formulas or doctors' empirical rules to determine radiotherapy doses, but the level of doctors varies, and they are subjective, lacking consistency and standardization in the formulation of radiotherapy doses, and also lacking accuracy. Traditional statistical models are too simple to capture complex relationships, and the accuracy of predictions is limited. In other words, the existing methods will affect the accuracy of doctors' formulation of radiotherapy doses. If the formulated radiotherapy dose is too high, the risk of complications in normal tissues will increase; and if the formulated radiotherapy dose is too low, the control effect of the tumor will be affected.

[0023] In view of the above problems, the present application proposes the following embodiments to solve the above technical problems.

[0024] See also Figure 1 , an embodiment of the present application provides a method for predicting the probability of radiation normal tissue damage based on deep learning, including: steps 101 to 104.

[0025] Step 101: Obtain the patient's tumor dose data, basic information, and tumor diagnosis information.

[0026] Among them, the tumor dose data includes the data prepared by the doctor based on the patient's tumor diagnosis information before the patient is treated with radiotherapy, and the tumor dose data includes the radiotherapy dose. Here, the patient's tumor diagnosis information may include but is not limited to tumor type, tumor size, tumor location, etc. After obtaining the patient's tumor diagnosis information, the doctor formulates the tumor dose data. The above-mentioned basic information of the patient includes at least the patient's age.

[0027] Step 102: Based on the tumor dose data, a dose volume histogram of the patient is obtained.

[0028] Among them, a dose volume histogram (DVH, Dose and Volume Histogram) corresponding to the tumor dose data can be automatically generated directly based on the radiotherapy planning system.

[0029] Step 103: Based on the dose-volume histogram, volume parameters and biological equivalent dose of normal tissues of the patient are obtained.

[0030] It should be noted that since the dose-volume histogram is a physical dose, it is not convenient for unified standards and subsequent predictions. Therefore, the physical dose is extracted from the dose-volume histogram and converted into a biological equivalent dose.

[0031] The conversion formula for bioequivalent dose is: ;in, It represents the biological equivalent dose when the physical dose extracted from the dose-volume histogram after conversion is 2 Gy as the fractional dose; Indicates that the fractionated dose in the radiotherapy plan is , the number of divisions is The total physical dose at Indicates cell proliferation correction factor. Considering that normal tissue is a normal late-response tissue, The value is taken as 3Gy.

[0032] Step 104: The volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information are input into the normal tissue damage probability prediction model constructed based on deep learning to obtain the patient's normal tissue damage probability.

[0033] This application pre-builds a normal tissue damage probability prediction model based on deep learning to predict the probability of damage to the patient's normal tissue. The deep learning model can automatically extract high-order features from large-scale data and adapt to complex nonlinear relationships.

[0034] It should be noted that this solution is applicable to most tumors and the damage probability of a normal tissue corresponding to the tumor.

[0035] For example, the normal tissue damage probability prediction model constructed based on deep learning can be a pre-constructed damage prediction model for lung tissue corresponding to lung cancer. Then one application scenario of the above scheme is that the damage probability prediction for lung tissue can be obtained based on the tumor dose data, basic information and tumor diagnosis information of lung cancer patients.

[0036] For example, the normal tissue damage probability prediction model constructed based on deep learning can be a pre-constructed damage prediction model for the corresponding throat soft tissue for nasopharyngeal carcinoma. Another application scenario of the above scheme is that the damage probability prediction for the throat soft tissue can be obtained based on the tumor dose data, basic information and tumor diagnosis information of nasopharyngeal carcinoma patients.

[0037] Of course, in other application scenarios, the tumor type may also be brain tumor, esophageal cancer, etc., and the corresponding normal tissue may also be brain stem tissue, optic nerve tissue, lung tissue, etc.

[0038] In summary, the method for predicting the probability of radiation normal tissue damage based on deep learning provided in the embodiments of the present application has the following beneficial effects: First, by introducing a normal tissue damage probability prediction model based on deep learning in the embodiments of the present application, high-dimensional features can be extracted from complex data, and the potential relationship between the patient's tumor dose data, tumor diagnostic information and individual information can be fully explored. Compared with traditional statistical models or empirical formulas, this method has higher prediction accuracy and can more accurately evaluate the probability of normal tissue damage.

[0039] Second, the prediction model provided by the embodiment of the present application is based on deep learning and is combined with a prediction method of multi-dimensional information fusion analysis, which can improve data utilization and make up for the problem that the traditional model can only rely on a single data, resulting in low prediction accuracy.

[0040] Third, considering the consistency of data utilization of the prediction model using deep learning, in the embodiment of the present application, the physical dose directly obtained from the dose-volume histogram is converted in advance, and the physical dose is converted into a biological equivalent dose with 2Gy as the fractionated dose. This can more scientifically input the patient's dose distribution into the deep learning prediction model, enhance the model's predictive ability and clinical relevance, and further standardize the dose distribution of different radiotherapy schemes, making different treatment plans comparable.

[0041] Fourth, the above scheme has wide adaptability and can be applied to various tumor types and their corresponding normal tissue damage prediction needs, and has strong versatility and scalability.

[0042] In an embodiment of the present application, a variety of different types of network models are pre-constructed to construct the normal tissue damage probability model based on deep learning in the above embodiment.

[0043] The following describes various types of network models provided in the embodiments of the present application.

[0044] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a convolutional neural network model (CNN model).

[0045] The structure of the convolutional neural network model includes: multiple convolutional layers, batch normalization layers, dropout layers, maximum pooling layers, and fully connected layers.

[0046] That is, the embodiment of the present application constructs a complex convolutional neural network model, which includes multiple convolutional layers, batch normalization layers, Dropout layers, maximum pooling layers, and fully connected layers. The convolutional blocks are constructed by defining the conv_block function. In each convolutional block, the input data is sequentially operated through the convolutional layer, batch normalization, and activation function. This structure helps to extract and enhance the local features of the data. In the construction of the entire model, an initial convolutional layer, batch normalization, and activation function are first set. Then, different groups of convolutional blocks are added in a loop for multiple times, and the number of convolutional blocks in each group is different. Add a maximum pooling layer at an appropriate position to reduce the data dimension and highlight important features. Finally, the features are converted into vectors through a global average pooling layer, and then classified through a fully connected layer, which can be used to process one-dimensional data.

[0047] Because the convolutional neural network model is good at capturing spatial correlation, it can be better applied to scenarios where tumors are close to complex normal tissues, such as head and neck tumors and brain tumors.

[0048] See also Figure 2 , Figure 2 An image showing the prediction effect of a tumor based on a convolutional neural network model provided in an embodiment of the present application, wherein the horizontal axis is the patient's tumor dose data and the vertical axis is the probability of damage to the patient's normal tissue.

[0049] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a deep neural network model (DNN model).

[0050] The structure of the deep neural network model includes: building with multiple fully connected layers as the core; also including batch normalization layers and Dropout layers.

[0051] The model construction of this deep neural network model adopts a sequential approach. The reshape layer is first used to adjust the input data to a suitable shape to prepare for the fully connected layer processing. Each fully connected layer uses the ReLU activation function to introduce nonlinearity, enabling the network to learn complex functional relationships. A batch normalization layer is added after each fully connected layer to accelerate the training process and improve the stability of the model, reducing the problem of internal covariate shift. The Dropout layer randomly discards neuron outputs in a certain proportion to prevent overfitting of the model and enhance generalization ability. Finally, the fully connected output layer maps the previously extracted features to the appropriate output dimension. This deep neural network model can process data of various dimensions.

[0052] See also Figure 3 , Figure 3 An image showing the prediction effect of a tumor based on a deep neural network model provided in an embodiment of the present application, wherein the horizontal axis is the patient's tumor dose data and the vertical axis is the probability of damage to the patient's normal tissue.

[0053] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a residual network model (ResNet model).

[0054] The structure of the residual network model includes: building with the residual block as the core; and including convolutional layers, batch normalization layers, activation function layers, and fully connected layers.

[0055] Specifically, the residual network model builds a neural network model with residual blocks as the core, and covers convolutional layers, batch normalization layers, activation function layers, and fully connected layers. First, the residual block is constructed by defining the residual_block function. In the function, the input is first saved as a shortcut. After two convolutional layers and corresponding batch normalization and activation function operations, the dimension of the shortcut can be adjusted according to the change in step size or number of channels. Then the convolution result and the shortcut are added and activated again to achieve residual connection, which effectively solves the gradient problem in deep network training. In model construction, an initial convolutional layer, batch normalization and activation function, as well as a maximum pooling layer are first set. Then, different groups of residual blocks are added in multiple cycles according to the design. The number and parameters of residual blocks in each group are different, and the complex features of the image are gradually extracted. Finally, the features are converted into vectors through the global average pooling layer, and then classified through the fully connected layer. It is suitable for processing data with complex hierarchical structures such as images.

[0056] See also Figure 4 , Figure 4 An image showing the prediction effect of a tumor based on the residual network model provided in an embodiment of the present application, wherein the horizontal axis is the patient's tumor dose data, and the vertical axis is the patient's normal tissue damage probability.

[0057] Optionally, the normal tissue damage probability prediction model constructed based on deep learning is a long short-term memory network model (LSTM model).

[0058] The structure of the long short-term memory network model includes: building with multiple LSTM layers as the core; and also including batch normalization layers and Dropout layers.

[0059] Specifically, firstly, a neural network model with long short-term memory unit as the core is constructed, which includes multiple LSTM layers, batch normalization layers and Dropout layers. The LSTM module is constructed by defining the lstm_module function. In each module, the input data first passes through an LSTM layer. The LSTM layer uses its special gate structure (input gate, forget gate and output gate) to process the long-term dependency in the sequence data. Then, the data is normalized by the batch normalization layer to stabilize the training process and reduce the impact of data distribution changes on model training. Then, the Dropout layer randomly discards the neuron output with a certain probability to prevent overfitting. In the process of model construction, the shape of the input data is first specified, including the sequence length and the feature dimension of each time step, and then lstm_module is added multiple times to construct LSTM structures of different levels. The return_sequences of the first few lstm_modules are set to True to continue processing sequence information, and the return_sequences of the last lstm_module is set to False to extract the final representation of the entire sequence. Finally, the features output by the LSTM module are mapped to the appropriate output dimension through the fully connected layer, which is specially used to process sequence data.

[0060] See also Figure 5 , Figure 5 An image showing the prediction effect of a tumor based on the long short-term memory network model provided in an embodiment of the present application, wherein the horizontal axis is the patient's tumor dose data and the vertical axis is the patient's normal tissue damage probability.

[0061] It should be noted that the above network models provided in the embodiments of the present application all need to be converted into biological equivalent doses, that is, converted into equivalent uniform doses of each network model.

[0062] The equivalent uniform dose of each network model can be calculated as follows: ; In this formula, It represents equivalent uniform dose; The normal tissue biological equivalent dose is divided into Equal parts, For the Equal doses, is The corresponding normal tissue volume; is the Lebecque norm.

[0063] The following describes the training process of the normal tissue damage probability prediction model based on deep construction provided in the embodiment of the present application. The training process can be applied to any of the above-mentioned types of network models, and of course can also be applied to other types of network models not listed.

[0064] Specifically, the network model is trained through the following steps, including: obtaining sample data; wherein the sample data is the detection data of tumor patients before and after radiotherapy; the detection data of tumor patients after radiotherapy is the normal tissue detection data of tumor patients sixty months after radiotherapy; based on the sample data, the initial network model is trained, and the model parameters of the initial network model are adjusted to obtain a network model in the mathematical model library.

[0065] The above sample data are the test data of tumor patients before and after radiotherapy. Among them, the test data before radiotherapy mainly include the patient's tumor dose data, basic information and tumor diagnosis information, while the test data after radiotherapy mainly includes the test of the patient's normal tissue damage. The test data after radiotherapy is used for labeling, and then supervised training of the network model.

[0066] The study found that in the prediction of the probability of normal tissue damage, the difficulty of the network model construction process lies in the lack of sample size. Since the damage of normal tissue is usually a gradual process, some damage may not be obvious in the short term after radiotherapy, but will gradually appear over a longer period of time. For example, radiotherapy (affecting the temporal lobe) leads to the gradual decline of cognitive functions such as memory and language, and this change can only be manifested within a few years after radiotherapy. Short-term test data alone may not accurately predict these side effects and the effects of normal tissue damage. Therefore, in the embodiment of the present application, the initial network model is trained by following up the test data of at least 500 patients for sixty months after radiotherapy as sample data. The above method can provide richer data features for the network model. The sixty-month data after radiotherapy can provide rich information including the patient's physiological state, side effects after radiotherapy, tissue repair process, etc. These data features can make the model training more comprehensive and avoid missing key factors that may affect damage prediction. Compared with relying only on short-term data, long-term follow-up data can make the prediction results more accurate and comprehensive. At the same time, the effects of radiotherapy on patients with different individual characteristics (such as age) vary greatly. Long-term data can help identify individual patients' unique response patterns to radiotherapy, thereby providing more basis for personalized treatment plans.

[0067] It should be noted that the test data of cancer patients after radiotherapy may include computed tomography (CT), magnetic resonance imaging (MRI), blood tests, etc.

[0068] In the embodiments of the present application, the sample data will also be preprocessed, including: removing outliers, noise data and duplicate data to ensure the accuracy and consistency of the data; performing efficient data integration; integrating various types of data from different devices and different time points, unifying the data format into a standardized medical digital imaging and communication format and the international system of units for subsequent processing; performing reasonable data transformation: selecting appropriate transformation methods according to the characteristics of the data, such as logarithmic transformation, standardization, normalization and other operations, for example, performing logarithmic transformation on some data to reduce the skewness of the data, and standardizing other data to make its mean 0 and standard deviation 1. Make the data meet the input requirements of the model and improve the stability and convergence speed of the model; perform scientific data reduction: reduce the dimension and complexity of the data through methods such as feature selection or dimensionality reduction, reduce the computational cost, improve the computational efficiency, and avoid overfitting problems.

[0069] In addition, in order to make the results more accurate, the preprocessed sample data were grouped based on the multiple cross-validation method; the data were divided into training set, validation set, and test set; 70% of the data were randomly divided as training set, 20% as validation set, and 10% as test set. The training set is used for model training and parameter adjustment, and its size should be large enough to ensure that the model can fully learn the characteristics and laws of the data; the validation set is used to evaluate the performance of the model during the training process and select the best model parameters. Its data should be representative and reflect the distribution characteristics of the overall data; the test set is used to finally evaluate the generalization ability of the model, and its data should be independent of the training set and validation set to truly test the performance of the model on new data; the test set can assist clinicians in predicting temporal lobe damage due to radiotherapy doses and observe the stability and accuracy of the model. By randomly dividing the data set multiple times, the result deviation caused by the randomness of data division is reduced, and the reliability and stability of the model are improved.

[0070] For model verification, the embodiment of the present application uses the above verification set to verify the model through multiple evaluation indicators; the model is verified using multiple evaluation indicators, including but not limited to Evaluation indicators and AUC area evaluation indicators are used to comprehensively evaluate the model performance.

[0071] It reflects the proportion of the dependent variable variation that can be explained by the model. is a relative measure that can be used to compare similar models trained on the same data. When it is close to 1, it means that the model can well explain the variability of the dependent variable and has a high degree of fit. When it is close to 0, it means that the model cannot explain the variability of the dependent variable and the degree of fit is low. Reflects the model's ability to explain the data. The higher the A value of indicates that the model better explains the variation in the data.

[0072] The AUC area is used to evaluate the classification performance of the model, especially in binary classification problems. The larger the AUC area, the stronger the model's ability to distinguish. For the AUC area, by drawing the receiver operating characteristic (ROC) curve, calculating the area under the curve, and selecting the model with a larger AUC area.

[0073] In addition, it should be noted that the weight and bias optimization modules are designed in the training process of the above four network models. The back propagation algorithm is used in the above four network models, that is, starting from the output layer, the partial derivatives of the weights and biases of each layer are calculated according to the loss function, and the parameters are updated according to a certain optimization algorithm until the optimal model configuration is optimized; The differences in optimization algorithms of the above four deep learning-based network models are as follows: The convolutional neural network model contains special layers such as convolutional layers and pooling layers. The weight update of the convolutional layer is achieved by sliding the convolution kernel on the input data to calculate the gradient.

[0074] The deep neural network model is mainly composed of fully connected layers, and weight updates are achieved based on the neuron connection method of the fully connected layer. For a fully connected layer with n input neurons and m output neurons, the weight matrix is ​​an m*n matrix and the bias is an m-dimensional vector. When updating the weights, it is necessary to consider the contribution of each input neuron to each output neuron, and update the weight matrix and bias vector through gradient calculation.

[0075] The core of the residual network model is the residual block, which affects the update of weights and biases through the jump connections in the residual block. When calculating the gradient, the input is directly added to the output of the subsequent layer so that the gradient can be propagated more directly between different layers, alleviating the gradient vanishing problem in deep networks. This makes it easier for the network to learn the identity mapping, that is, the output is the input itself.

[0076] The LSTM network model has a special gate structure, namely the input gate, forget gate and output gate. The forget gate determines which information is discarded from the cell state. The update of its weights and biases comprehensively considers multiple factors such as the current input, the hidden state at the previous moment, and the cell state. In the back propagation process, it is necessary to derive the operations of these gates according to the loss function, and then update the corresponding weights and biases to learn the long-term dependencies in the sequence data. Different optimization algorithms are used to continuously adjust the model parameters to achieve the optimal model configuration, so that the model can accurately fit the relationship between input data such as the patient's radiotherapy dose and volume data and the probability of normal tissue complications.

[0077] See also Figure 6 , Figure 6 The specific training process of the four network models provided in the embodiment of the present application is shown. First, data, i.e., sample data, is collected, and then the data is preprocessed. After that, 70% of the data is randomly divided as a training set, 20% as a validation set, and 10% as a test set. The four network models are trained with the training set, and then after the iterative training with the training set is completed, the model is evaluated with the validation set. According to the evaluation results, it is determined whether the optimal model configuration has been reached. If not, the model is continued to be iteratively trained; if so, the model is predicted with the test set. Finally, the prediction results are presented.

[0078] Optionally, the tumor diagnostic information includes the tumor type; the normal tissue damage probability prediction model constructed based on deep learning is a joint model group; different joint model groups are constructed in advance according to different tumor types to form a prediction model library.

[0079] Exemplarily, the tumor type is a brain tumor, and a joint model group A is constructed, which includes a prediction model for brainstem tissue, a prediction model for temporal lobe tissue, and a prediction model for salivary gland tissue.

[0080] Exemplarily, the tumor type is lung cancer, and a joint model group B is constructed, which includes a prediction model for lung tissue, a prediction model for esophageal tissue, and a prediction model for the heart.

[0081] The joint model group A and the joint model group B in the above example are both stored in the prediction model library. Different joint model groups can be understood as multi-task prediction model groups constructed separately for different tumor types, and the damage probability of normal tissue output by different joint model groups is associated with the tumor type.

[0082] For details, please refer to Figure 7The above steps input the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information into the normal tissue damage probability prediction model constructed based on deep learning to obtain the patient's normal tissue damage probability, including: step 701~step 702.

[0083] Step 701: Based on the patient's tumor type, determine a target joint model group from a prediction model library.

[0084] According to the above example, assuming that the patient's tumor type is a brain tumor, the joint model group A is selected from the prediction model library.

[0085] Step 702: Input the volume parameters and biological equivalent dose of the patient's normal tissues, as well as the patient's basic information and tumor diagnosis information into the target joint model group, so as to output the damage probability of multiple normal tissues associated with the patient's tumor type through the target joint model group.

[0086] Then, the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis are input into the determined target joint model, thereby realizing the prediction of the damage probability of multiple normal tissues associated with the tumor type.

[0087] Continuing with the above example, the model group A is combined to output the probability of brainstem tissue damage, the probability of temporal lobe tissue damage, and the probability of salivary gland tissue damage.

[0088] It can be seen that the embodiment of the present application provides a multi-task prediction method for the patient's tumor type, which can obtain damage probability prediction results for multiple related normal tissues based on the patient's tumor type. However, the traditional method usually only predicts the damage probability for a certain normal tissue that is easily affected, and lacks an assessment of other normal tissues of tumor patients. Here, through a single input, the damage probability prediction of multiple related normal tissues can be obtained, which can assist doctors in more comprehensively understanding the potential impact after radiotherapy. In summary, this solution realizes the joint prediction of the damage probability of multiple normal tissues, realizes personalization and efficiency by constructing a joint model group for different tumor types. It can not only comprehensively evaluate the impact of radiotherapy on multiple normal tissues, but also help doctors optimize treatment plans and provide patients with better treatment experience and effects.

[0089] Optionally, there are multiple normal tissue damage probability prediction models constructed based on deep learning, and they are network models of different types.

[0090] See also Figure 8The above steps input the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information into the normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue, including: step 801 to step 803.

[0091] Step 801: input the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information into multiple network models, so that each network model outputs the damage probability of the patient's normal tissue.

[0092] In one embodiment, four network models as described in the above embodiments may be constructed for the same normal tissue. For example, the four models are all suitable for predicting the probability of damage to lung tissue. The volume parameters and biological equivalent dose of the patient's lung tissue, as well as the patient's basic information and tumor diagnosis information may be input into the four network models, so that the probability of damage to the patient's lung tissue may be output through each network model.

[0093] Step 802: Obtain the weight of each network model.

[0094] Step 803: Based on the weight of each network model and the damage probability of the normal tissue of the patient output by each network model, a joint damage probability is obtained.

[0095] The combined injury probability is the final injury probability of the patient's normal tissue.

[0096] Here, it can be understood that a weight is assigned to the damage probability output by each network model, and then the final joint damage probability is obtained by summing them up.

[0097] That is, the embodiment of the present application provides a method for joint prediction by combining multiple different types of network models. The probability of normal tissue damage of the patient is independently predicted by multiple network models, and then the outputs of different models are fused using weights, integrating the advantages of different network models, thereby improving the accuracy and credibility of the overall prediction. This method can reduce the limitations of the output results caused by a single type of model.

[0098] It should be noted that the weight of each of the above network models may be static or pre-set.

[0099] Of course, in some embodiments, the weight of each of the above network models can also be adjusted dynamically.

[0100] Optionally, an embodiment of the present application provides a method for dynamically adjusting weights, that is, the above steps of obtaining the weight of each network model may specifically include: determining the age range to which the patient belongs based on the patient's basic information; determining the range of the patient's tumor size based on the patient's tumor diagnosis information; determining the weight of each network model according to the patient's age range, the range of the patient's tumor size, and the model accuracy of each network model in the age range and for the range of the tumor size.

[0101] For example, the age ranges may be pre-set as 20-30 years old, 30-40 years old, 40-50 years old, 50-60 years old, and so on. Then the age range to which the patient belongs is determined. Assuming that the patient is 35 years old, the patient belongs to the age range of 30-40 years old.

[0102] For example, the size range of the tumor can be preset, and the size range is set according to different tumor types. For example, the conventional setting can be 0-3 cm, 3-5 cm, 5-10 cm, etc. Then determine the range of the patient's tumor size. Assuming that the patient's tumor size is 7 cm, it belongs to the 5-10 cm range.

[0103] After determining the age range of 30 to 40 years old and the tumor size range of 5 to 10 cm, the weight of each network model is determined according to the model accuracy of the above four network models in this age range and the range of tumor size. The determination of model accuracy is the accuracy of model prediction in this range, and the determination of this accuracy can be obtained through the test data of the test set in the above embodiment.

[0104] Initially, the weight of each network model can be set to 0.25. Assuming that the convolutional neural network model has the highest accuracy among people aged 30 to 40, and the deep neural network model has the lowest accuracy, the corresponding weight of the convolutional neural network model can be increased, and the corresponding weight of the deep neural network model can be reduced.

[0105] Assuming that in the range of tumor size of 5-10 cm, the accuracy is arranged from high to low as follows: residual network model, long short-term memory network model, deep neural network model, convolutional neural network model, then the corresponding weights of the residual network model and the long short-term memory network model can be increased, and the corresponding weights of the deep neural network model and the convolutional neural network model can be reduced. And increase and decrease according to the gradient, such as the final weight of the residual network model can be 0.4, the weight of the long short-term memory network model can be 0.3, the weight of the deep neural network model can be 0.2, and the weight of the convolutional neural network model can be 0.1 (this is an example, and the influence of age is not considered).

[0106] It can be seen that the embodiment of the present application provides a more accurate prediction, introduces a mechanism for dynamically adjusting weights, combines the patient's age range, tumor size range, and the prediction accuracy of each network model under these conditions, and dynamically optimizes the weight distribution of each network model, thereby ensuring that under different patient characteristics, the weight distribution matches the model performance, thereby significantly improving the accuracy of the final prediction results.

[0107] Optionally, an embodiment of the present application provides another way to dynamically adjust weights, that is, the above steps to obtain the weight of each network model may specifically include: determining the weight of each network model according to the input data type of the patient's tumor diagnosis information.

[0108] The input data types here may include time series data, image data, etc.

[0109] For example, since the residual network model is more suitable for extracting features from complex image data, especially keeping features undistorted in deep networks, it is suitable for processing high-dimensional tumor diagnosis information (such as multimodal image data). Therefore, when the patient's tumor diagnosis information involves image data (such as CT), the weight of the residual network model can be increased.

[0110] For example, since the long short-term memory network model can perform superiorly on time series data, when the patient's tumor diagnosis information involves data such as tumor diagnosis progress records, the weight of the long short-term memory network model can be increased.

[0111] In summary, the above method achieves the best match between model characteristics and data features by dynamically determining the weight of each network model according to the data type of tumor diagnostic information, thereby comprehensively improving the accuracy, reliability and adaptability of predictions. This method not only fully utilizes the unique advantages of each network model, but also dynamically adapts to different tumor types and data scenarios, supports personalized and multimodal processing, and has significant clinical application value.

[0112] See also Fig. 9 Based on the same inventive concept, the embodiment of the present application provides a radiation normal tissue damage probability prediction system 900 based on deep learning, including: a first acquisition module 901, used to obtain the patient's tumor dose data, basic information and tumor diagnosis information; wherein the tumor dose data includes the doctor's formulation based on the patient's tumor diagnosis information before radiotherapy; the basic information at least includes the patient's age; a second acquisition module 902, used to obtain the patient's dose volume histogram based on the tumor dose data; a third acquisition module 903, used to obtain the patient's normal tissue volume parameters and biological equivalent dose based on the dose volume histogram; wherein the conversion formula of the biological equivalent dose is: ;in, represents the biological equivalent dose when the physical dose extracted from the dose-volume histogram after conversion is 2 Gy as the fractional dose; Indicates that the fractionated dose in the radiotherapy plan is , the number of divisions is The total physical dose at represents the cell proliferation correction factor, The value of is taken as 3Gy; the prediction module 904 is used to input the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the basic information and tumor diagnosis information of the patient into the normal tissue damage probability prediction model constructed based on deep learning, to obtain the damage probability of the patient's normal tissue.

[0113] See also Fig.10 Based on the same inventive concept, the present application embodiment provides a module frame of an electronic device 1000 that applies the above method. The electronic device 1000 includes: at least one processor 1001 ( Fig.10 Only one is shown in the figure), memory 1002, a computer program 1003 stored in the memory 1002 and executable on at least one processor 1001, and when the processor 1001 executes the computer program 1003, the steps of the method in any of the foregoing embodiments are implemented.

[0114] The electronic device 1000 may be a server, a personal computer, a notebook computer, etc.

[0115] Those skilled in the art will understand that Fig.10 The electronic device 1000 is merely an example and does not constitute a limitation on the electronic device 1000 , and may include more or less components than those shown in the figure, or may combine certain components, or may have different components.

[0116] The processor 1001 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0117] In some embodiments, the memory 1002 may be an internal storage unit of the electronic device 1000, such as a hard disk or memory of the electronic device 1000. In other embodiments, the memory 1002 may also be an external storage device of the electronic device 1000, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1000. Further, the memory 1002 may also include both an internal storage unit of the electronic device 1000 and an external storage device.

[0118] It should be noted that the above-mentioned systems, devices, etc. are based on the same concept as the method embodiments of the present application. The modules designed for the systems, the steps performed by the devices, and the technical effects brought about can all be found in the method embodiment section and will not be repeated here.

[0119] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0120] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0121] An embodiment of the present application provides a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0122] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / electronic device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electrical carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0123] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0124] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example 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 performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0125] In the embodiments provided in the present application, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0126] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0127] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for predicting the probability of radiation normal tissue damage based on deep learning, characterized in that: include: Obtaining the patient's tumor dose data, basic information and tumor diagnosis information; wherein the tumor dose data includes the dose data formulated by the doctor based on the patient's tumor diagnosis information before radiotherapy; the basic information includes at least the patient's age; Based on the tumor dose data, obtaining a dose volume histogram of the patient; Based on the dose-volume histogram, the volume parameters and biological equivalent dose of the patient's normal tissue are obtained; wherein the conversion formula of the biological equivalent dose is: ;in, represents the biological equivalent dose when the physical dose extracted from the dose-volume histogram after conversion is 2 Gy as the fractional dose; Indicates that the fractionated dose in the radiotherapy plan is , the number of divisions is The total physical dose at represents the cell proliferation correction factor, The value of is 3Gy; The volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information are input into a normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue.

2. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 1, characterized in that: The tumor diagnosis information includes tumor type; the normal tissue damage probability prediction model constructed based on deep learning is a joint model group; different joint model groups are constructed in advance according to different tumor types to form a prediction model library; the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the basic information and tumor diagnosis information of the patient are input into the normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue, including: Based on the tumor type of the patient, determining a target combined model group from the prediction model library; The volume parameters and biological equivalent dose of the patient's normal tissues, as well as the patient's basic information and tumor diagnosis information are input into the target joint model group, so as to output the damage probability of multiple normal tissues associated with the patient's tumor type through the target joint model group.

3. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 1, characterized in that: There are multiple normal tissue damage probability prediction models constructed based on deep learning, and they are network models of different types; the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the basic information and tumor diagnosis information of the patient are input into the normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue, including: Inputting the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information into a plurality of network models, so that each network model outputs the damage probability of the patient's normal tissue; Get the weights of each network model; Based on the weight of each network model and the damage probability of the normal tissue of the patient output by each network model, a combined damage probability is obtained; wherein the combined damage probability is the final obtained damage probability of the normal tissue of the patient.

4. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 3, characterized in that: The obtaining of the weight of each network model includes: Based on the basic information of the patient, determining the age range to which the patient belongs; Based on the patient's tumor diagnosis information, determining the range interval to which the patient's tumor size belongs; The weight of each network model is determined according to the age range of the patient, the range of the patient's tumor size, and the model accuracy of each network model in the age range and for the range of the tumor size.

5. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 3, characterized in that: The obtaining of the weight of each of the network models comprises: The weight of each network model is determined according to the input data type of the patient's tumor diagnosis information.

6. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 1, characterized in that: The normal tissue damage probability prediction model built based on deep learning is a convolutional neural network model; The structure of the convolutional neural network model includes: multiple convolutional layers, batch normalization layers, dropout layers, maximum pooling layers and fully connected layers.

7. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 1, characterized in that: The normal tissue damage probability prediction model built based on deep learning is a deep neural network model; The structure of the deep neural network model includes: being constructed with multiple fully connected layers as the core; and also including a batch normalization layer and a Dropout layer.

8. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 1, characterized in that: The normal tissue damage probability prediction model built based on deep learning is a residual network model; The structure of the residual network model includes: building with the residual block as the core; and including a convolutional layer, a batch normalization layer, an activation function layer and a fully connected layer.

9. The method for predicting the probability of radiation normal tissue damage based on deep learning according to claim 1, characterized in that: The normal tissue damage probability prediction model built based on deep learning is a long short-term memory network model; The structure of the long short-term memory network model includes: building with multiple LSTM layers as the core; and also including a batch normalization layer and a Dropout layer.

10. A radiation normal tissue damage probability prediction system based on deep learning, characterized in that: include: The first acquisition module is used to acquire the patient's tumor dose data, basic information and tumor diagnosis information; wherein the tumor dose data includes the data set by the doctor based on the patient's tumor diagnosis information before radiotherapy; the basic information at least includes the patient's age; A second acquisition module, configured to acquire a dose volume histogram of the patient based on the tumor dose data; The third acquisition module is used to acquire the volume parameters and biological equivalent dose of the normal tissue of the patient based on the dose volume histogram; wherein the conversion formula of the biological equivalent dose is: ;in, represents the biological equivalent dose when the physical dose extracted from the dose-volume histogram after conversion is 2 Gy as the fractional dose; Indicates that the fractionated dose in the radiotherapy plan is , the number of divisions is The total physical dose at represents cell proliferation correction factor, The value of is 3Gy; The prediction module is used to input the volume parameters and biological equivalent dose of the patient's normal tissue, as well as the patient's basic information and tumor diagnosis information into a normal tissue damage probability prediction model constructed based on deep learning to obtain the damage probability of the patient's normal tissue.

Images