A hereditary cancer auxiliary diagnosis method, device and program product based on adaptive dynamic range loss function

Abstract

This application relates to the field of intelligent healthcare, specifically to a method, device, and program product for the auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function. It includes acquiring the image data and genetic information data of the subject; inputting the image data and genetic information data into a hereditary cancer discrimination model to obtain a discrimination result for whether it is a hereditary cancer; the hereditary cancer discrimination model is pre-trained using the image data and genetic information data, wherein the discrimination model is optimized by calculating the average absolute error of the predicted value and the true value at different scales and adjusting the dynamic error using an adaptive dynamic range loss function. This application can assist doctors in the diagnosis and discrimination of hereditary cancers and has significant clinical value.

Description

Technical Field

[0001] This application relates to the field of intelligent healthcare, specifically to a method, device, program product, and computer-readable storage medium for the auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function. Background Technology

[0002] Hereditary renal cell carcinoma is a type of kidney tumor with a familial tendency, accounting for approximately 2%-8% of all renal cell carcinomas. Although its incidence is low, its pathological types are complex, exhibiting various disease syndromes with different clinical manifestations and characteristic genetic mutations, posing significant challenges to clinical diagnosis and treatment. Traditional medical imaging diagnosis relies heavily on the experience and expertise of physicians, while genetic screening is based on patient family history and limited gene testing results; these methods have limitations in diagnostic efficiency and accuracy. Summary of the Invention

[0003] To address the aforementioned problems, this invention provides a method for the auxiliary diagnosis of hereditary cancers based on an adaptive dynamic range loss function. It aims to accurately and rapidly identify the characteristics of hereditary cancers from massive amounts of medical imaging data and genetic information using deep learning technology, thereby improving diagnostic efficiency and accuracy. Specifically, it includes:

[0004] Obtain imaging data and genetic information data of the test subjects;

[0005] The image data and genetic information data are input into the hereditary cancer discrimination model to obtain the discrimination result of whether it is hereditary cancer. The hereditary cancer discrimination model is obtained by pre-training with image data and genetic information data. The discrimination model is optimized by calculating the average value of the absolute error of the predicted value and the true value at different scales and adjusting the dynamic error through an adaptive dynamic range loss function.

[0006] The formula for calculating the adaptive dynamic range loss function is ADRL(y, Represented as: ) is represented as:

[0007] ADRL(y, ) = *exp(- )

[0008] Where y is the real label. This is the output value predicted by the model, where N is the total number of samples. It is the absolute error between the predicted value and the actual value. It is the standard deviation of the true label y. It is a very small positive number. This represents the mean of the true label y. It is a dynamically adjustable item.

[0009] The structure of the hereditary cancer discrimination model includes an input layer, an encoder layer, an attention layer, and a fully connected layer. Image data and genetic information data are input to the input layer, converted into vectors, and then fused. The vectors are then passed through the encoder layer, attention layer, and fully connected layer in sequence to obtain the discrimination result.

[0010] Optionally, the encoder layer performs vector encoding through a self-attention mechanism to obtain a self-attention feature vector, the attention layer performs feature processing through multi-head attention, the self-attention feature vector is input to the attention layer and then divided into N heads, where N is a natural number greater than 1, each head independently calculates N features, the N features are fused to obtain attention features, and the attention features are processed through a fully connected layer to obtain a discrimination result;

[0011] Optionally, the training process of the genetic discrimination model is as follows:

[0012] Acquire patient image datasets and genetic information datasets;

[0013] The image dataset and genetic information dataset are input into the hereditary cancer discrimination model to be trained for feature learning and loss value is calculated using an adaptive dynamic range loss function. The gradient is then used to optimize the model based on the loss value.

[0014] Repeat the above steps until the adaptive dynamic loss function remains unchanged, and you will obtain the genetic discriminant model.

[0015] The hereditary discrimination model also includes a hereditary cancer gene-assisted diagnosis module, which inputs the feature vector output by the hereditary discrimination model to the hereditary cancer gene-assisted diagnosis module for feature analysis to obtain the diagnostic result of the hereditary cancer type; the hereditary cancer gene-assisted diagnosis module includes a feature extraction layer and a classification layer, and the feature vector is processed through the feature extraction layer and the classification layer to obtain the diagnostic result.

[0016] Optionally, the diagnostic results may also include the type of genetic mutation.

[0017] The feature vector is the feature vector output by the encoder layer of the genetic discriminant model or the feature vector output by the fully connected layer.

[0018] The training process for the genetic cancer gene-assisted diagnosis module is as follows:

[0019] Obtain the feature vector set;

[0020] The feature vector set is input to the hereditary cancer gene-assisted diagnosis module for feature extraction, classification calculation, and loss calculation;

[0021] Repeat the above steps until the loss remains constant to obtain the genetic cancer gene-assisted diagnosis module.

[0022] The loss calculation of the hereditary cancer gene-assisted diagnosis module is performed using one or more of the following loss functions: cross-entropy loss function, logarithmic loss function, hinge loss function, and adaptive dynamic range loss function.

[0023] The purpose of this invention is to provide a computer program product comprising a computer program or instructions, wherein the computer program or instructions are executed by a processor to implement the above-described method for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function.

[0024] The purpose of this invention is to provide a computer device comprising a memory, a processor, and a computer program or instructions stored in the memory, wherein the computer program or instructions are executed by the processor to implement the above-described method for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function.

[0025] The purpose of this invention is to provide a computer-readable storage medium having a computer program or instructions stored thereon, which is executed by a processor to implement the above-described method for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function.

[0026] Advantages of this invention:

[0027] 1. This invention proposes an adaptive dynamic range loss function (ADRL), which first introduces the standard deviation... With its small constant ϵ, ADRL can automatically adapt to data distributions at different scales, reducing the need for data preprocessing. Secondly, it exhibits dynamic range sensitivity: by introducing an exponential function and dynamic variance θ, ADRL can dynamically adjust its sensitivity to error based on the location of the predicted value, helping the model learn better in situations of class imbalance or uneven data distribution. Finally, it combines absolute error and uncertainty: ADRL integrates absolute error and the uncertainty of model predictions, enabling it to acquire more information during model training in scenarios with uncertain or noisy data, thus ensuring a certain level of stability.

[0028] 2. This invention assists in the diagnosis of hereditary cancers by training a hereditary cancer discrimination model, which includes a two-stage module. The first stage uses an input layer, encoder layer, attention layer, and fully connected layer to determine whether it is a hereditary cancer. The second stage uses an input layer, encoder layer, attention layer, fully connected layer, feature extraction layer, and classification layer to determine the type of hereditary cancer and the type of genetic gene. This helps doctors use the model to diagnose hereditary cancers and reduces the workload of manual diagnosis. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A schematic diagram of a method for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function provided in an embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a hereditary cancer auxiliary diagnostic system based on an adaptive dynamic range loss function provided in an embodiment of the present invention;

[0032] Figure 3 A schematic diagram of a hereditary cancer auxiliary diagnostic device based on an adaptive dynamic range loss function provided in an embodiment of the present invention;

[0033] Figure 4 This is a schematic diagram of the structure and flow of the hereditary cancer discrimination model provided in an embodiment of the present invention. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0035] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as S101, S102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0036] Figure 1 A schematic diagram of the hereditary cancer auxiliary diagnosis method based on an adaptive dynamic range loss function provided in this embodiment of the invention specifically includes:

[0037] S101: Obtain the imaging data and genetic information data of the subject;

[0038] In one embodiment, the image data includes one or more of the following: CT, MRI, ultrasound, and X-ray images.

[0039] In one embodiment, genetic information includes, but is not limited to, the following types: family inheritance patterns, disease status of family members, gene mutation information (mutation site, mutation type), gene polymorphism, chromosomal abnormalities (numerical abnormalities, structural abnormalities), and pharmacogenetic information.

[0040] S102: Input the image data and genetic information data into the hereditary cancer discrimination model to obtain the discrimination result of whether it is hereditary cancer; the hereditary cancer discrimination model is obtained by pre-training the image data and genetic information data, wherein the discrimination model is optimized by calculating the average value of the absolute error of the predicted value and the true value at different scales and adjusting the dynamic error through the adaptive dynamic range loss function.

[0041] In one embodiment, the adaptive dynamic range loss function is calculated using the formula ADRL(y, ) is represented as:

[0042] ADRL(y, ) = *exp(- )

[0043] Where y is the real label. This is the output value predicted by the model, where N is the total number of samples. It is the absolute error between the predicted value and the actual value. It is the standard deviation of the true label y. It is a very small positive number. This represents the mean of the true label y. It is a dynamically adjustable item.

[0044] In one embodiment, the hereditary cancer discrimination model structure includes an input layer, an encoder layer, an attention layer, and a fully connected layer. Image data and genetic information data are input to the input layer, converted into vectors, and then fused. The vectors are then passed through the encoder layer, attention layer, and fully connected layer in sequence to obtain the discrimination result.

[0045] In one embodiment, the encoder layer performs vector encoding through a self-attention mechanism to obtain a self-attention feature vector. The attention layer performs feature processing through multi-head attention. After the self-attention feature vector is input to the attention layer, it is divided into N heads, where N is a natural number greater than 1. Each head independently calculates N features. The N features are fused to obtain attention features. The attention features are processed through a fully connected layer to obtain a discrimination result.

[0046] In one embodiment, the training process of the genetic discrimination model is as follows:

[0047] Acquire patient image datasets and genetic information datasets;

[0048] The image dataset and genetic information dataset are input into the hereditary cancer discrimination model to be trained for feature learning and loss value is calculated using an adaptive dynamic range loss function. The gradient is then used to optimize the model based on the loss value.

[0049] Repeat the above steps until the adaptive dynamic loss function remains unchanged, and you will obtain the genetic discriminant model.

[0050] In one embodiment, the hereditary discrimination model further includes a hereditary cancer gene-assisted diagnosis module, which inputs the feature vector output by the hereditary discrimination model to the hereditary cancer gene-assisted diagnosis module for feature analysis to obtain the diagnostic result of the hereditary cancer type; the hereditary cancer gene-assisted diagnosis module includes a feature extraction layer and a classification layer, and the feature vector is processed through the feature extraction layer and the classification layer to obtain the diagnostic result.

[0051] In one embodiment, the diagnostic result may also include the type of genetic mutation.

[0052] In one embodiment, the feature vector is the feature vector output by the encoder layer of the genetic discriminant model or the feature vector output by the fully connected layer.

[0053] In one embodiment, the training process of the hereditary cancer gene-assisted diagnostic module is as follows:

[0054] Obtain the feature vector set;

[0055] The feature vector set is input to the hereditary cancer gene-assisted diagnosis module for feature extraction, classification calculation, and loss calculation;

[0056] Repeat the above steps until the loss remains constant to obtain the genetic cancer gene-assisted diagnosis module.

[0057] In one embodiment, the loss calculation of the hereditary cancer gene-assisted diagnosis module is performed using one or more of the following loss functions: cross-entropy loss function, logarithmic loss function, hinge loss function, and adaptive dynamic range loss function.

[0058] In one specific embodiment, the structure of the hereditary cancer discrimination model is as follows:

[0059] 1. Data Acquisition Module: Used to collect patients' medical imaging data (such as CT, MRI, etc.) and genetic information (such as germline mutation detection results).

[0060] Two main model building modules: Based on deep learning technology, a large model is built for image analysis and genetic information interpretation. This model can accurately identify the characteristics of kidney tumors from medical images and, combined with genetic information, determine whether the tumor is hereditary kidney cancer. The specific structure of the large model module is as follows:

[0061] 2.1 Encoder layer;

[0062] The sequence X, which is a concatenation of medical image feature vectors and genetic information feature vectors, is processed by a self-attention mechanism to calculate the attention weight of each element in the sequence to other elements, thus obtaining the attention matrix.

[0063] Att = softmax(QK^T*X / d_k)V

[0064] Where Q (Query), K (Key), and V (Value) are the linear transformation results of the input sequence, and d_k is the dimension of the Key. Next, the output of the self-attention mechanism is further processed. The calculation formula is:

[0065] F = max(0, Att *W_1 + b_1)W_2 + b_2

[0066] Where W_1 and W_2 are weight matrices, and b_1 and b_2 are bias terms.

[0067] Output: The output of the encoder layer serves as the input to the next layer, or as the final feature representation, or as the input to the diagnostic module.

[0068] 2.3 Attention Mechanism Layer;

[0069] The encoder layer output employs a multi-head attention mechanism, dividing the input sequence into multiple heads, denoted as head_1,..., head_h. Each head undergoes independent self-attention computation, and the results are then concatenated. The computation result is as follows:

[0070] M= Concat(head_1, ..., head_h)W_3

[0071] Where head_i is the attention result of the i-th head, and W_3 is the weight matrix.

[0072] The output of the attention mechanism is used as the input to the next layer.

[0073] 2.4 Fully Connected Layer;

[0074] The output of the attention mechanism layer is processed by a linear transformation and activation function. The calculation formula is as follows:

[0075] D = σ(W_4x + b_4)

[0076] Where W_4 is the weight matrix, b_4 is the bias term, and σ is the ReLU activation function.

[0077] The output of the fully connected layer can be used as the final output or as the input to the diagnostic analysis module.

[0078] In one specific embodiment, the diagnostic analysis module: The output of the large model module is input into the diagnostic analysis module for further calculation, and the diagnostic result is output. The diagnostic result includes whether the tumor is hereditary kidney cancer and the possible types of genetic mutations. The model structure of the diagnostic analysis module is as follows:

[0079] Feature extraction layer: This layer extracts image features from medical image data and genetic information data through multiple convolution and pooling operations. The calculation formula is as follows:

[0080] T = σ(W_5 * D + b_5)

[0081] Where D is the output of the large model module, W_5 is the weight matrix, b_5 is the bias term, and σ is the ReLU activation function.

[0082] Next, genetic information feature extraction is performed, and the calculation formula is as follows:

[0083] U = σ(W_6T + b_6)

[0084] Where W_6 is the weight matrix, b_6 is the bias term, and σ is the ReLU activation function.

[0085] Classification layer: The extracted feature vectors are classified using a fully connected layer and a softmax function.

[0086] The formula for calculating a fully connected layer is:

[0087] P(y|x) = W_7U + b_7

[0088] Where W_7 is the weight matrix, b_7 is the bias term, and P(y|x) is the score of the corresponding category y in the output of the fully connected layer. The probability distribution of P(y|x) for each category is calculated. The category with the highest probability is selected as the diagnostic result, including the type of hereditary kidney cancer and the type of genetic mutation.

[0089] In one specific embodiment, the adaptive dynamic range loss function is:

[0090] ADRL(y, ) =

[0091] Where: y is the true label, ŷ is the model's predicted output value, N is the total number of samples, |y_i - ŷ_i| is the absolute error between the predicted and true values, STD(y) is the standard deviation of the true label y, used to measure the overall fluctuation range of the data, ϵ is a very small positive number with a denominator of zero, μ(y) represents the mean of the true label y, and θ is a dynamic adjustment term.

[0092] In one embodiment, image data and genetic information of the subject are acquired; the image data and genetic information are input into a hereditary cancer discrimination model to obtain a discrimination result; the structure of the hereditary cancer discrimination model includes an input layer, an encoder layer, an attention layer, a fully connected layer, a feature extraction layer, and a classification layer; the image data and genetic information are converted into image feature vectors and genetic feature vectors through the input layer; the image feature vectors and genetic feature vectors are concatenated and then sequentially processed through the encoder layer, attention layer, and fully connected layer to obtain processed features; the processed features are then extracted through the feature extraction layer and input to the classification layer for classification to obtain a second discrimination result (type of hereditary cancer, type of genetic mutation). Furthermore, the processed features obtained through the fully connected layer are used to calculate a first discrimination result (discrimination result for whether it is hereditary cancer). The model's structural flow is as follows: Figure 4 As shown.

[0093] In one embodiment, the hereditary cancer detection model is a pre-trained model. During training, an adaptive dynamic range loss function is used to calculate the model loss. Depending on the requirements, the structure of the hereditary cancer detection model can include any of the following:

[0094] 1. Input layer + encoder layer + attention layer + fully connected layer;

[0095] 2. Input layer + encoder layer + attention layer + fully connected layer + feature extraction layer + classification layer;

[0096] 3. Input layer + feature extraction layer + classification layer.

[0097] In one embodiment, when the structure of the hereditary cancer discrimination model is an input layer, a feature extraction layer, and a classification layer, the patient's medical image dataset and genetic information dataset are obtained.

[0098] The medical image dataset and genetic information data are input to the input layer for vector transformation to obtain image vector sets and genetic vector sets.

[0099] The image vector set and the genetic vector set are input into the feature extraction layer to extract feature vectors.

[0100] The feature vector is then input to the classification layer for identification to determine the type of hereditary cancer or the type of genetic mutation.

[0101] Alternatively, the feature vector output by a hereditary cancer discrimination model with a network structure of input layer + encoder layer + attention layer + fully connected layer can be obtained. The feature vector is then input to the feature extraction layer to extract features and obtain a secondary feature vector. The secondary feature vector is then input to the classification layer to determine the type of hereditary cancer or the type of genetic mutation.

[0102] Alternatively, it can simultaneously acquire medical image datasets, genetic information datasets, and feature vectors output by a hereditary cancer discrimination model with a network structure of input layer + encoder layer + attention layer + fully connected layer;

[0103] The medical image dataset and genetic information data are input to the input layer for vector transformation to obtain image vector sets and genetic vector sets.

[0104] The image vector set, genetic vector set, and feature vector are fused to obtain the fused vector;

[0105] The fused vector is fed into the feature extraction layer for feature extraction to obtain a cubic feature vector;

[0106] The three feature vectors are then input into the classification layer for discrimination to obtain the type of hereditary cancer or the type of genetic mutation.

[0107] The present invention also discloses a computer program product or system, including a computer program that, when executed by a processor, implements the steps of the above-described method for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function.

[0108] Figure 2 A schematic diagram of a hereditary cancer auxiliary diagnostic system based on an adaptive dynamic range loss function provided in this embodiment of the invention specifically includes:

[0109] Acquisition module: Acquires the imaging data and genetic information data of the test subject;

[0110] The discrimination module inputs the image data and genetic information data into the hereditary cancer discrimination model to obtain the discrimination result of whether it is hereditary cancer. The hereditary cancer discrimination model is pre-trained using image data and genetic information data. The discrimination model is optimized by calculating the average absolute error of the predicted value and the true value at different scales and adjusting the dynamic error through an adaptive dynamic range loss function.

[0111] In one embodiment, the system further includes a diagnostic module: the feature vector output by the hereditary discriminant model is input to the hereditary cancer gene-assisted diagnostic module for feature analysis to obtain a diagnostic result of the hereditary cancer type; the hereditary cancer gene-assisted diagnostic module includes a feature extraction layer and a classification layer, and the feature vector is processed through the feature extraction layer and the classification layer to obtain the diagnostic result.

[0112] In one embodiment, the diagnostic result may also include the type of genetic mutation.

[0113] In one embodiment, the system further includes a user interaction module for displaying diagnostic results, providing treatment suggestions, and receiving additional information input by the user for further analysis.

[0114] Figure 3 A schematic diagram of a hereditary cancer auxiliary diagnostic device based on an adaptive dynamic range loss function provided in this embodiment of the invention specifically includes:

[0115] A memory and a processor; the memory is used to store program instructions; the processor is used to invoke the program instructions, when any of the above-described methods for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function are executed.

[0116] The present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, is any of the above-described methods for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function.

[0117] The verification results of this verification embodiment show that assigning inherent weights to indications can improve the performance of this method compared to the default settings. 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. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for example, the division of units is merely 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 displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separated; 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 according to actual needs to achieve the purpose of this embodiment. Furthermore, the functional units in the various embodiments of the present invention 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. The integrated units described above can be implemented in hardware or as software functional units. Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0118] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0119] The computer device provided by the present invention has been described in detail above. For those skilled in the art, there will be changes in the specific implementation and application scope based on the ideas of the embodiments of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for auxiliary diagnosis of hereditary cancer based on an adaptive dynamic range loss function, characterized in that, include: Obtain imaging data and genetic information data of the test subjects; The image data and genetic information data are input into the hereditary cancer discrimination model to obtain the discrimination result of whether it is hereditary cancer; the hereditary cancer discrimination model is obtained by pre-training with image data and genetic information data, wherein the discrimination model is optimized by calculating the average value of the absolute error of the predicted value and the true value at different scales and adjusting the dynamic error through an adaptive dynamic range loss function; The formula for calculating the adaptive dynamic range loss function is ADRL(y, ) is represented as: ADRL(y, ) = *exp(- ) Where y is the real label. This is the output value predicted by the model, where N is the total number of samples. It is the absolute error between the predicted value and the actual value. It is the standard deviation of the true label y. It is a very small positive number. This represents the mean of the true label y. It is a dynamically adjustable item.

2. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 1, characterized in that, The structure of the hereditary cancer discrimination model includes an input layer, an encoder layer, an attention layer, and a fully connected layer. Image data and genetic information data are input to the input layer, converted into vectors, and then fused. The vectors are then passed through the encoder layer, attention layer, and fully connected layer in sequence to obtain the discrimination result.

3. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 2, characterized in that, The encoder layer uses a self-attention mechanism to encode vectors to obtain self-attention feature vectors. The attention layer performs feature processing through multi-head attention. After the self-attention feature vectors are input to the attention layer, they are divided into N heads, where N is a natural number greater than 1. Each head independently calculates N features. The N features are fused to obtain attention features. The attention features are processed through a fully connected layer to obtain the discrimination result.

4. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 1, characterized in that, The training process of the hereditary cancer discrimination model is as follows: Acquire patient image datasets and genetic information datasets; The image dataset and genetic information dataset are input into the hereditary cancer discrimination model to be trained for feature learning and loss value is calculated using an adaptive dynamic range loss function. The gradient is then used to optimize the model based on the loss value. Repeat the above steps until the adaptive dynamic loss function remains unchanged, and you will obtain the genetic discriminant model.

5. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 1, characterized in that, The hereditary cancer discrimination model also includes a hereditary cancer gene-assisted diagnosis module, which inputs the feature vector output by the hereditary discrimination model to the hereditary cancer gene-assisted diagnosis module for feature analysis to obtain the diagnostic result of the hereditary cancer type; the hereditary cancer gene-assisted diagnosis module includes a feature extraction layer and a classification layer, and the feature vector is processed through the feature extraction layer and the classification layer to obtain the diagnostic result.

6. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 5, characterized in that, The diagnostic results also include the type of genetic mutation.

7. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 5, characterized in that, The feature vector is the feature vector output by the encoder layer of the genetic discriminant model or the feature vector output by the fully connected layer.

8. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 5, characterized in that, The training process for the genetic cancer gene-assisted diagnosis module is as follows: Obtain the feature vector set; The feature vector set is input to the hereditary cancer gene-assisted diagnosis module for feature extraction, classification calculation, and loss calculation; Repeat the above steps until the loss remains constant to obtain the genetic cancer gene-assisted diagnosis module.

9. The method for auxiliary diagnosis of hereditary cancer based on adaptive dynamic range loss function according to claim 8, characterized in that, The loss calculation of the hereditary cancer gene-assisted diagnosis module is performed using one or more of the following loss functions: cross-entropy loss function, logarithmic loss function, hinge loss function, and adaptive dynamic range loss function.

10. A computer program product comprising a computer program or instructions, characterized in that, The computer program or instructions are executed by a processor to implement the hereditary cancer auxiliary diagnosis method based on an adaptive dynamic range loss function as described in any one of claims 1-9.

11. A computer device comprising a memory, a processor, and a computer program or instructions stored in the memory, characterized in that, The computer program or instructions are executed by a processor to implement the hereditary cancer auxiliary diagnosis method based on an adaptive dynamic range loss function as described in any one of claims 1-9.

12. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, The computer program or instructions are executed by a processor to implement the hereditary cancer auxiliary diagnosis method based on an adaptive dynamic range loss function as described in any one of claims 1-9.

Images