Radiographic dose adjusting system based on deep learning

Abstract

The invention discloses a radiation dose adjustment system based on deep learning, and relates to the technical field of neural networks, the system comprises an image analysis module, a clinical parameter analysis module, a dose decision module and a dose implementation monitoring module which work cooperatively in sequence, and each module forms a complete radiation dose adjustment closed loop through data interaction. According to the radiation dose adjusting system based on deep learning, tumor boundaries are accurately recognized through multi-step image processing, individualized dose distribution is achieved in combination with clinical parameter analysis, it is ensured that the dose is concentrated in a lesion area, radiation damage to healthy tissue is reduced, and the occurrence rate of side effects is reduced.

Description

Technical Field

[0001] This invention relates to the field of neural network technology, specifically to a radiation dose adjustment system based on deep learning. Background Technology

[0002] Neural networks are computational models inspired by the biological brain, used to simulate how the human brain processes information to perform various machine learning tasks. These networks consist of a large number of nodes (or neurons) connected in a hierarchical structure, capable of learning to recognize patterns and features through training. Neural network technology has wide applications in many fields, including speech recognition, image processing, and medical diagnosis. With improvements in hardware performance and the development of algorithms, neural network models, especially deep learning models, have become one of the key technologies driving the forefront of artificial intelligence. Among these, radiation dose adjustment systems refer to the use of artificial intelligence technologies such as neural networks to optimize dose distribution during radiotherapy. The purpose of such systems is to ensure that patients receive an appropriate and precise radiation dose in the disease-affected area while minimizing radiation exposure to surrounding healthy tissues. This technology can improve treatment effectiveness and reduce side effects, representing an important development direction in the field of modern medical radiology. A search revealed a deep learning-based radiation dose adjustment system disclosed in Chinese patent CN119027420B. By distinguishing the boundary between tumors and normal tissues and refining the characteristics of the tumor in the image, the system enables more precise and personalized radiation dose allocation. This detailed image analysis optimizes the treatment plan, ensuring the dose is concentrated in the lesion area and reducing the impact on healthy tissues. Furthermore, the integrated analysis of clinical parameters enhances the understanding of the patient's radiosensitivity, providing treatment plans that better meet individual needs. This dose adjustment method, based on in-depth analysis, effectively adjusts dose deviations during treatment by monitoring the execution effect of radiotherapy in real time.

[0003] With the development of neural network technology, its advantages in image processing and pattern recognition have made it possible to optimize radiotherapy. However, existing radiation dose adjustment technologies still have many problems that need to be solved. First, individual differences are not adequately considered. Current dosage regimens are often based on standardized templates and fail to deeply integrate individual patient information; Secondly, there is a lack of dynamic feedback and deviation correction mechanisms. During radiotherapy, factors such as fluctuations in equipment accuracy and subtle changes in patient position due to respiratory movements can cause deviations between the actual and planned doses. Third, the ability to conduct multi-dimensional data collaborative analysis is weak. Radiation dose adjustment requires the integration of multi-source data such as medical imaging, clinical parameters, and historical treatment cases, but existing technologies have failed to establish an effective data fusion and correlation analysis mechanism. Summary of the Invention

[0004] The purpose of this invention is to provide a deep learning-based radiation dose adjustment system to solve the problems mentioned in the background art.

[0005] By adopting the above technical solution, seamless information transmission between modules is achieved through preset data interfaces, forming a complete radiation dose adjustment process from medical image data input to implementation effect report generation. This ensures that each operation is based on accurate data from the preceding modules, and that subsequent modules can optimize and verify the results of the preceding steps.

[0006] In view of the above problems, the technical solution proposed by the present invention is as follows: A deep learning-based radiation dose adjustment system comprises an image analysis module, a clinical parameter analysis module, a dose decision module, and a dose implementation monitoring module that work sequentially and collaboratively. These modules interact to form a complete closed loop for radiation dose adjustment. The specific functions and data flow relationships are as follows: Image analysis module: After receiving medical image data input, it first reduces image noise through Gaussian smoothing to optimize the data basis, then uses Canny edge detection technology to accurately distinguish the boundary between tumor and normal tissue and refine the size and shape of the tumor, then extracts key features in the image through contrast enhancement algorithm, and finally summarizes the above processing results to generate tumor feature data, and transmits the tumor feature data to the clinical parameter analysis module. Clinical parameter analysis module: After receiving tumor feature data transmitted from the image analysis module, it integrates three types of clinical information: patient position information, tumor type, and previous radiation response records. Through the collaborative analysis of tumor feature data and clinical information, it calculates the tumor's radiosensitivity index, forms a collaborative clinical parameter result, and transmits the collaborative clinical parameter result to the dose decision module. Dosage Decision Module: After receiving the collaborative results of clinical parameters transmitted by the clinical parameter analysis module, it compares them with the existing radiation dose database, analyzes the correlation between pathological features and clinical data to clarify the basis for dose adjustment, adjusts individualized dose allocation based on this basis and refines the matching degree and adaptability of the dose, generates a dose adjustment plan, and transmits the dose adjustment plan to the dose implementation monitoring module. Dosage Implementation Monitoring Module: After receiving the dose adjustment plan transmitted by the dose decision module, it uses this as a benchmark to monitor the dose implementation effect in real time during radiotherapy, synchronously records the dynamic data of treatment, makes timely adjustments for treatment dose deviations found in the monitoring, and finally generates an implementation effect report that includes dose implementation effect, dynamic data and deviation adjustment status.

[0007] The implementation method of this system includes the following steps: S1: The patient's original images are acquired through CT and MRI. After noise reduction by Gaussian kernel smoothing, Canny edge detection is used to clarify the boundary between the tumor and normal tissue. Then, the contrast is enhanced to extract key features, and deep learning weights are combined to generate quantitative tumor feature data. S2: Receives tumor characteristic data, simultaneously collects patient treatment position angles, enters tumor type and previous radiation response records, and calculates radiosensitivity indicators through multiple rounds of parameter integration and coefficient adjustment to form clinical parameter synergy results; S3: Match the clinical parameters with the radiation dose database, screen similar cases, calculate and standardize the correlation score, clarify the degree of correlation through the Pearson correlation coefficient, and determine the basis for dose adjustment; S4: Based on the standard baseline dose, and taking into account factors such as correlation coefficients and individual differences, optimize the final dose adjustment plan to balance efficacy and tolerability. S5: The actual dose is captured in real time by the sensor according to the dosage plan, the dynamic dose data is recorded and generated, the planned dose is compared and the deviation exceeding the threshold is automatically corrected, and finally the information is integrated to generate an effect report, while backtracking and optimizing the parameters of the preceding modules.

[0008] As a preferred technical solution of the present invention, the Gaussian kernel in the image analysis module is designed as a two-dimensional normal distribution function, and the kernel size can be dynamically adjusted according to the resolution and noise level of the medical image. The kernel size ranges from 3×3 to 11×11.

[0009] As a preferred technical solution of the present invention, the positional information in the clinical parameter analysis module is the angle information of the patient during treatment, in degrees, with a value range of 0° to 360°, which is collected in real time by medical positioning equipment and transmitted to the system.

[0010] As a preferred embodiment of the present invention, the existing radiation dose database in the dose decision module contains at least 1,000 cases of clinical radiotherapy data, covering dose plans and treatment effect records corresponding to different tumor types, stages and individual patient characteristics, and the database supports regular updates and data expansion.

[0011] As a preferred embodiment of the present invention, the dose monitoring module adopts a real-time dose monitoring sensor with a sampling frequency of ≥1Hz, which can capture the instantaneous changes in radiation dose in real time and transmit the data to the system for dynamic analysis and deviation correction.

[0012] On the other hand, the present invention provides a method for a deep learning-based radiation dose adjustment system, comprising the following steps: S1: Medical Imaging Data Acquisition and Tumor Feature Extraction Raw patient image data is acquired using medical equipment such as CT and MRI and transmitted to the image analysis module. A two-dimensional normally distributed Gaussian kernel, with a kernel size ranging from 3×3 to 11×11, is dynamically adjusted according to image resolution and noise to smooth the raw images, reducing noise interference and preserving core tumor features. Based on the smoothed images, Canny edge detection technology is used to generate an edge enhancement map by calculating multi-directional gradients and synthesizing gradient amplitudes, accurately distinguishing the boundary between tumor and normal tissue, and refining the tumor size and shape. The edge enhancement map is then contrast-enhanced to extract key features such as tumor edge roughness and internal structural differences. Weighted summation of feature weights trained by deep learning is used to generate quantified tumor feature data.

[0013] S2: Clinical parameter integration and radiosensitivity analysis The clinical parameter analysis module receives tumor feature data output from the image analysis module and simultaneously collects the patient's treatment position angle (0°-360°) through medical positioning equipment, recording the tumor type and previous radiation response records. First, it integrates the tumor feature data and position information, introduces a moderating coefficient to balance the contributions of both, and generates an initial comprehensive dataset. Then, it combines the tumor type quantification value and adjusts the influence of the initial dataset and tumor type through weighting coefficients to generate an intermediate dataset. Referring to the previous radiation response records, it introduces a sensitivity moderating coefficient and a threshold to calculate the radiosensitivity index, forming a synergistic result of clinical parameters.

[0014] S3: Historical data comparison and determination of dosage adjustment basis The dose decision module matches the results of clinical parameters with the existing radiation dose database to screen similar clinical cases; calculates an initial correlation score; standardizes the score and then calculates the Pearson correlation coefficient to clarify the degree of linear correlation between pathological features and clinical data, and to determine the basis for dose adjustment.

[0015] S4: Generation of Individualized Dosing Adjustment Protocols Based on the standard clinical dose, a preliminary dose adjustment value is calculated by combining the correlation coefficient and the adjustment coefficient. The individual difference adjustment coefficient and the quantitative value of the patient's side effect intensity are introduced to further optimize the preliminary dose adjustment value and generate a final dose adjustment plan that takes into account both the therapeutic effect and the patient's tolerance.

[0016] S5: Treatment process monitoring and outcome report generation The dosage implementation monitoring module uses the dosage adjustment plan as a benchmark and employs sensors with a sampling frequency of ≥1Hz to capture the actual dose administered in real time. It records the actual dose data by combining weighted parameters such as the equipment deviation ratio and subtle changes in the patient's body position. It continuously records the actual dose and duration of each stage of treatment, introduces a time-weighted factor, and accumulates dynamic dose data to reflect the dose accumulation throughout the treatment. It compares the dynamic dose data with the planned dose, calculates the deviation ratio, and automatically corrects it if the deviation exceeds the threshold. It integrates the dosage implementation effect, dynamic data, and deviation adjustment status to generate an implementation effect report, while also retrospectively optimizing the parameters of the preceding modules.

[0017] Compared with existing technologies, the beneficial effects of this invention are as follows: This deep learning-based radiation dose adjustment system accurately identifies tumor boundaries through multi-step image processing and achieves individualized dose allocation by combining clinical parameter analysis, ensuring that the dose is concentrated in the lesion area, reducing radiation damage to healthy tissue, and lowering the incidence of side effects; it integrates medical images, clinical information, and historical databases, and formulates dose plans based on correlation analysis, avoiding reliance on experience and improving the scientificity and reliability of the plan; through dynamic monitoring and deviation correction, it ensures that the dose is implemented as expected during treatment, improving the safety and stability of treatment, and providing strong support for the development of radiotherapy technology. Attached Figure Description

[0018] Figure 1 A flowchart illustrating the method of the deep learning-based radiation dose adjustment system provided by this invention. Figure 2 The block diagram of the deep learning-based radiation dose adjustment system provided by this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1 Please see Figure 1 This invention provides a technical solution: a deep learning-based radiation dose adjustment system. This system includes an image analysis module, a clinical parameter analysis module, a dose decision module, and a dose implementation monitoring module that work sequentially and collaboratively. Each module forms a complete closed loop for radiation dose adjustment through data interaction. The specific functions and data flow relationships are as follows: Image analysis module: After receiving medical image data input, it first reduces image noise through Gaussian smoothing to optimize the data basis, then uses Canny edge detection technology to accurately distinguish the boundary between tumor and normal tissue and refine the size and shape of the tumor, then extracts key features in the image through contrast enhancement algorithm, and finally summarizes the above processing results to generate tumor feature data, and transmits the tumor feature data to the clinical parameter analysis module. Clinical parameter analysis module: After receiving tumor feature data transmitted from the image analysis module, it integrates three types of clinical information: patient position information, tumor type, and previous radiation response records. Through the collaborative analysis of tumor feature data and clinical information, it calculates the tumor's radiosensitivity index, forms a collaborative clinical parameter result, and transmits the collaborative clinical parameter result to the dose decision module. Dosage Decision Module: After receiving the collaborative results of clinical parameters transmitted by the clinical parameter analysis module, it compares them with the existing radiation dose database, analyzes the correlation between pathological features and clinical data to clarify the basis for dose adjustment, adjusts individualized dose allocation based on this basis and refines the matching degree and adaptability of the dose, generates a dose adjustment plan, and transmits the dose adjustment plan to the dose implementation monitoring module. Dosage Implementation Monitoring Module: After receiving the dose adjustment plan transmitted by the dose decision module, it uses this as a benchmark to monitor the dose implementation effect in real time during radiotherapy, synchronously records the dynamic data of treatment, makes timely adjustments for treatment dose deviations found in the monitoring, and finally generates an implementation effect report that includes dose implementation effect, dynamic data and deviation adjustment status.

[0021] The implementation method of this system includes the following steps: S1: The patient's original images are acquired through CT and MRI. After noise reduction by Gaussian kernel smoothing, Canny edge detection is used to clarify the boundary between the tumor and normal tissue. Then, the contrast is enhanced to extract key features, and deep learning weights are combined to generate quantitative tumor feature data. S2: Receives tumor characteristic data, simultaneously collects patient treatment position angles, enters tumor type and previous radiation response records, and calculates radiosensitivity indicators through multiple rounds of parameter integration and coefficient adjustment to form clinical parameter synergy results; S3: Match the clinical parameters with the radiation dose database, screen similar cases, calculate and standardize the correlation score, clarify the degree of correlation through the Pearson correlation coefficient, and determine the basis for dose adjustment; S4: Based on the standard baseline dose, and taking into account factors such as correlation coefficients and individual differences, optimize the final dose adjustment plan to balance efficacy and tolerability. S5: The actual dose is captured in real time by the sensor according to the dosage plan, the dynamic dose data is recorded and generated, the planned dose is compared and the deviation exceeding the threshold is automatically corrected, and finally the information is integrated to generate an effect report, while backtracking and optimizing the parameters of the preceding modules.

[0022] The Gaussian kernel in the image analysis module is designed as a two-dimensional normal distribution function. The kernel size can be dynamically adjusted according to the resolution and noise level of the medical image. The kernel size range is 4×4. The body position information in the clinical parameter analysis module is the angle information of the patient during treatment, in degrees, with a value range of 90°. It is collected in real time by medical positioning equipment and transmitted to the system.

[0023] The dose decision module utilizes existing radiation dose databases from 450 clinical radiotherapy cases. This data covers dose regimens and treatment efficacy records corresponding to different tumor types, stages, and individual patient characteristics. The database supports regular updates and data expansion. The dose implementation monitoring module employs a real-time dose monitoring sensor with a sampling frequency ≥1Hz, capable of capturing instantaneous changes in radiation dose in real time and transmitting the data to the system for dynamic analysis and deviation correction. The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 2 Please see Figure 2 This invention provides a technical solution: a method for a radiation dose adjustment system based on deep learning, comprising the following steps: S1: Medical Imaging Data Acquisition and Tumor Feature Extraction Raw patient image data is acquired using medical equipment such as CT and MRI and transmitted to the image analysis module. A two-dimensional normally distributed Gaussian kernel, with a kernel size of 4×4, is dynamically adjusted according to image resolution and noise to Gaussian smooth the raw images, reducing noise interference and preserving core tumor features. Based on the smoothed images, Canny edge detection technology is used to generate an edge enhancement map by calculating multi-directional gradients and synthesizing gradient amplitudes, accurately distinguishing the boundary between tumor and normal tissue, and refining the tumor size and shape. The edge enhancement map is then contrast-enhanced to extract key features such as tumor edge roughness and internal structural differences. Weighted summation of feature weights trained by deep learning is used to generate quantified tumor feature data.

[0025] S2: Clinical parameter integration and radiosensitivity analysis The clinical parameter analysis module receives tumor feature data output from the image analysis module and simultaneously collects the patient's treatment position angle at 180° using medical positioning equipment (combined with clinical data, the current rotation angle allows for better contact with the affected area), and records the tumor type and previous radiation response records. First, the tumor feature data and position information are integrated, and a moderating coefficient is introduced to balance their contributions, generating an initial comprehensive dataset. Then, combined with the tumor type quantification value, the influence of the initial dataset and tumor type is adjusted through weighting coefficients to generate an intermediate dataset. Referring to previous radiation response records, a sensitivity moderating coefficient and threshold are introduced to calculate the radiosensitivity index, forming a synergistic clinical parameter result.

[0026] S3: Historical data comparison and determination of dosage adjustment basis The dose decision module matches the results of clinical parameters with the existing radiation dose database to screen similar clinical cases; calculates an initial correlation score; standardizes the score and then calculates the Pearson correlation coefficient to clarify the degree of linear correlation between pathological features and clinical data, and to determine the basis for dose adjustment.

[0027] S4: Generation of Individualized Dosing Adjustment Protocols Based on the standard clinical dose, a preliminary dose adjustment value is calculated by combining the correlation coefficient and the adjustment coefficient. The individual difference adjustment coefficient and the quantitative value of the patient's side effect intensity are introduced to further optimize the preliminary dose adjustment value and generate a final dose adjustment plan that takes into account both the therapeutic effect and the patient's tolerance.

[0028] S5: Treatment process monitoring and outcome report generation The dosage implementation monitoring module uses the dosage adjustment plan as a benchmark and employs a sensor with a sampling frequency of 2Hz to capture the actual dose administered in real time. It records the actual dose data by combining weighted parameters such as the equipment deviation ratio and subtle changes in the patient's body position. It continuously records the actual dose and duration of each stage of treatment, introduces a time-weighted factor, and accumulates dynamic dose data to reflect the dose accumulation throughout the treatment. It compares the dynamic dose data with the planned dose, calculates the deviation ratio, and automatically corrects it if the deviation exceeds the threshold. It integrates the dosage implementation effect, dynamic data, and deviation adjustment status to generate an implementation effect report, while also retrospectively optimizing the parameters of the preceding modules.

Claims

1. A deep learning based radiation dose adjustment system, characterized in that, The system comprises image analysis module, clinical parameter analysis module, dose decision module and dose implementation monitoring module which work in sequence, and each module forms a complete radiation dose adjustment closed loop through data interaction, and the specific functions and data flow relationship are as follows: The image analysis module: after receiving the medical image data input, the image noise is first reduced through Gaussian smoothing processing to optimize the data basis, then the Canny edge detection technology is used to accurately distinguish the boundary between tumor and normal tissue and refine the size and shape of the tumor, then the contrast enhancement algorithm is used to extract the key features in the image, finally the tumor feature data is generated by summarizing the above processing results, and the tumor feature data is transmitted to the clinical parameter analysis module; The clinical parameter analysis module: after receiving the tumor feature data transmitted by the image analysis module, the patient's body position information, tumor type and previous radiation reaction record are integrated, the radiation sensitivity index of the tumor is calculated through the cooperative analysis of the tumor feature data and the clinical information, the clinical parameter cooperative result is formed, and the clinical parameter cooperative result is transmitted to the dose decision module; The dose decision module: after receiving the clinical parameter cooperative result transmitted by the clinical parameter analysis module, it is compared with the existing radiation dose database, the correlation between pathological features and clinical data is analyzed to determine the dose adjustment basis, the individualized dose allocation is adjusted based on the basis, the matching degree and adaptability of the dose are refined, the dose adjustment scheme is generated, and the dose adjustment scheme is transmitted to the dose implementation monitoring module; The dose implementation monitoring module: after receiving the dose adjustment scheme transmitted by the dose decision module, the dose implementation effect in the radiotherapy process is monitored in real time based on the dose adjustment scheme, the dynamic data of the treatment is recorded synchronously, the treatment dose deviation found in the monitoring is adjusted in time, and finally the implementation effect report containing the dose implementation effect, dynamic data and deviation adjustment situation is generated; The implementation method of the system comprises the following steps: S1: Collect the original image of the patient by CT and MRI, reduce the noise after Gaussian kernel smoothing processing, use Canny edge detection to determine the boundary between tumor and normal tissue, enhance the contrast to extract key features, and generate quantitative tumor feature data combined with deep learning weight; S2: Receive the tumor feature data, synchronously collect the patient's treatment position angle, input the tumor type and previous radiation reaction record, integrate multiple parameters and adjust the coefficients, calculate the radiation sensitivity index, and form the clinical parameter cooperative result; S3: Match the clinical parameter cooperative result with the radiation dose database, calculate and standardize the correlation score after screening similar cases, determine the correlation degree through Pearson correlation coefficient, and determine the dose adjustment basis; S4: Based on the conventional basic dose, combined with factors such as correlation coefficient and individual difference, the final dose adjustment scheme considering efficacy and tolerance is optimized; S5: Real-time capture actual dose according to dose scheme with sensor, record and generate dynamic dose data, compare planned dose and automatically correct threshold deviation, finally integrate information to generate effect report, and backtrack to optimize previous module parameters.

2. The deep learning-based radiation dose adjustment system of claim 1, wherein, The Gaussian kernel in the image analysis module is designed as a two-dimensional normal distribution function, and the kernel size can be dynamically adjusted according to the resolution and noise level of the medical image, and the kernel size is in the range of 3*3 to 11*11.

3. The deep learning-based radiation dose adjustment system of claim 1, wherein, The body position information in the clinical parameter analysis module is the angle information of the patient during treatment, with a unit of degrees and a value range of 0° to 360°, wherein the value range is collected and transmitted to the system in real time by a medical positioning device.

4. The deep learning-based radiation dose adjustment system of claim 1, wherein, The existing radiation dose database in the dose decision module contains 1000 clinical radiotherapy data, covering dose plans and treatment effect records corresponding to different tumor types, stages and patient individual characteristics, and the database supports regular updating and data expansion.

5. The deep learning-based radiation dose adjustment system of claim 1, wherein, The dose implementation monitoring module uses a real-time dose monitoring sensor with a sampling frequency of ≥1 Hz, which can capture the instantaneous changes of the radiation dose in real time and transmit the data to the system for dynamic analysis and deviation correction.

6. A method of using the deep learning based radiation dose adjustment system according to any one of claims 1-5, characterized in that: The method comprises the following steps: S1: Medical image data acquisition and tumor feature extraction The original image data of the patient is collected by CT, MRI and other medical devices and transmitted to the image analysis module; a Gaussian kernel of two-dimensional normal distribution is used, wherein the kernel size is 3*3-11*11, and the kernel size is dynamically adjusted according to the image resolution and noise, the original image is subjected to Gaussian smoothing processing to reduce noise interference and retain the core features of the tumor; on the basis of the smoothed image, the Canny edge detection technology is used, the gradient in multiple directions is calculated and combined to generate an edge enhancement map, the boundary between the tumor and the normal tissue is accurately distinguished, and the size and shape of the tumor are refined; the edge enhancement map is subjected to contrast enhancement, the key features such as tumor edge roughness and internal structure difference are extracted, the feature weight trained by deep learning is used for weighted summation, and quantitative tumor feature data is generated; S2: Clinical parameter integration and radiosensitivity analysis The tumor feature data output by the image analysis module is received by the clinical parameter analysis module, and the treatment body position angle (0°-360°) of the patient is collected by a medical positioning device, the tumor type and the previous radiation reaction record are recorded; the tumor feature data and the body position information are integrated, the adjustment coefficient is introduced to balance the contributions of the two, and the initial comprehensive data set is generated; then, the weight coefficient is adjusted to balance the influence of the initial data set and the tumor type according to the quantitative value of the tumor type, and the intermediate data set is generated; the radiosensitivity index is calculated by introducing the sensitivity adjustment coefficient and the threshold value according to the previous radiation reaction record, and the clinical parameter coordination result is formed; S3: Historical data comparison and dose adjustment basis determination The clinical parameter coordination result is matched with the existing radiation dose database in the dose decision module, and similar clinical cases are screened; the initial correlation score is calculated; the score is standardized, and then the Pearson correlation coefficient is calculated to determine the linear correlation degree between the pathological characteristics and the clinical data, and the dose adjustment basis is determined; S4: Generation of individualized dose adjustment scheme Based on the clinical routine dose, combined with the correlation coefficient and the adjustment coefficient, the preliminary dose adjustment value is calculated; the individual difference adjustment coefficient and the patient side effect intensity quantification value are introduced to further optimize the preliminary dose adjustment value, and the final dose adjustment scheme is generated, which takes into account the treatment effect and the patient tolerance; S5: treatment process monitoring and effect report generation The dose implementation monitoring module takes the dose adjustment scheme as the basis, uses a sensor with a sampling frequency of ≥1 Hz to capture the actual dose in real time, combines with the weight parameters such as the equipment deviation ratio and the patient body position slight change, and records the actual dose data. The actual dose and the time length of each stage of treatment are recorded continuously, a time weighting factor is introduced, and dynamic dose data is accumulated to reflect the whole dose accumulation. The deviation ratio is calculated by comparing the dynamic dose data with the planned dose, and the deviation is automatically corrected if the deviation exceeds the threshold. The implementation effect report is generated by integrating the dose implementation effect, the dynamic data, and the deviation adjustment situation, while the parameters of the previous optimization sequence module are backtracked.

Images