Intelligent injection system for positron medicine

Abstract

The invention relates to the technical field of intelligent medical treatment, in particular to an intelligent injection system for positron medicine. The system is configured to: analyze global condition monitoring feature data; acquiring positron drug characteristic data; based on the global disease monitoring characteristic data and the positron medicine characteristic data, injection medicine flow diffusion simulation distribution analysis of the positron medicine is carried out, and injection medicine simulation distribution data are generated; performing injection medicine simulation distribution trend characteristic analysis according to the injection medicine simulation distribution data to generate injection medicine simulation distribution trend characteristic data; constructing and optimizing a positron medicine personalized injection control model based on the injection medicine simulation distribution trend characteristic data, and generating an optimized positron medicine personalized injection control model; and transmitting the injection medicine simulation distribution data to the optimized positron medicine personalized injection control model to analyze positron medicine personalized injection control parameters. According to the invention, intelligent positron medicine injection control is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and in particular to an intelligent injection system for positron drugs. Background Art

[0002] Positron emission tomography (PET) is a non-invasive medical imaging technology that is widely used in the diagnosis and treatment evaluation of tumors, neurological diseases, and cardiovascular diseases. PET scanning uses positron drugs with radioactive isotopes to detect the gamma rays they emit to obtain metabolic information of lesions in the body. With the continuous development of medical imaging technology, the application of PET technology in disease diagnosis is becoming more and more important. Although PET technology has made significant progress in accurately locating lesions and evaluating treatment effects, PET technology has certain limitations in the drug injection process, mainly in terms of the accuracy and personalized control of drug injection. However, the existing positron drug injection system mostly relies on manual operation and preset injection schemes for setting, lacking the ability to adjust and optimize in real time according to the individual characteristics of patients and changes in their condition, and in the automated injection process, factors such as the drug flow rate, targeted distribution, and metabolic rate are often not accurately controlled. Summary of the invention

[0003] Based on this, the present invention provides an intelligent injection system for positron drugs to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a smart injection system for positron drugs includes the following modules: A global symptom monitoring feature analysis module is used to obtain the patient's physiological sign data and the patient's historical lesion metabolic imaging data; perform global symptom monitoring feature analysis through the patient's physiological sign data and the patient's historical lesion metabolic imaging data to generate global symptom monitoring feature data; The injection drug simulation distribution analysis module is used to monitor the positron drug characteristics of positron drugs using radioactive detectors and generate positron drug characteristic data; based on the global disease monitoring feature data and positron drug characteristic data, the injection drug flow diffusion simulation distribution analysis of positron drugs is performed to generate injection drug simulation distribution data; The intelligent injection control model design module is used to analyze the injection drug simulation distribution trend characteristics according to the injection drug simulation distribution data, and generate the injection drug simulation distribution trend characteristic data; based on the injection drug simulation distribution trend characteristic data, a deep reinforcement learning model for personalized injection control of positron drugs is constructed to obtain a personalized injection control model for positron drugs; The positron drug intelligent injection control module is used to perform targeted tissue drug impact benefit analysis of the injection drug flow rate based on the injection drug simulation distribution data, and obtain the targeted tissue drug impact benefit data of the injection drug flow rate; perform model optimization processing on the positron drug personalized injection control model based on the targeted tissue drug impact benefit data of the injection drug flow rate, and generate an optimized positron drug personalized injection control model; transmit the injection drug simulation distribution data to the optimized positron drug personalized injection control model to perform personalized positron drug injection control analysis, and generate personalized positron drug injection control parameters; and perform positron drug intelligent injection operations based on the positron drug personalized injection control parameters.

[0005] Furthermore, the global symptom monitoring feature analysis module includes the following functions: Collecting and processing the patient's physiological signs in real time through a biosensor to generate the patient's physiological sign data, wherein the patient's physiological sign data includes body surface temperature data, heart rate variability data, and blood flow rate data; Obtain historical lesion metabolic imaging data of patients from the medical system; Perform lesion metabolic imaging feature analysis on the patient's historical lesion metabolic imaging data to generate lesion metabolic imaging feature data; Based on the lesion metabolic imaging feature data and the patient's physiological signs data, a global symptom monitoring feature analysis is performed to generate global symptom monitoring feature data.

[0006] Furthermore, the global disease monitoring feature analysis based on the lesion metabolic imaging feature data and the patient's physiological sign data includes: Perform voxel division processing on the lesion metabolic image feature data according to the lesion metabolic image feature data to generate voxelized lesion metabolic image feature data; The lesion metabolic imaging feature voxelized data and the patient's physiological sign data are aligned in time and space, and a global disease monitoring feature analysis is performed based on the lesion metabolic imaging feature voxelized data and the patient's physiological sign data after the time and space alignment to generate global disease monitoring feature data.

[0007] Furthermore, the injection drug simulation distribution analysis module includes the following functions: Using a radioactive detector to monitor the positron drug characteristics of the positron drug to generate positron drug characteristic data, wherein the positron drug characteristic data includes positron drug activity value data and positron drug decay rate data; Positron drug targeted tissue region analysis is performed through voxelized data of lesion metabolic imaging features of global disease monitoring feature data to generate positron drug targeted tissue region data; Based on the patient's physiological sign data of the global disease monitoring feature data and the positron drug characteristic data, a positron drug injection vascular domain simulation analysis is performed to generate positron drug injection vascular domain simulation data; Performing targeted tissue domain disease monitoring characteristic data analysis on the global disease monitoring characteristic data according to the positron drug targeted tissue region data to generate targeted tissue domain disease monitoring characteristic data; Perform positron drug targeted tissue domain diffusion analysis based on targeted tissue domain disease monitoring characteristic data and positron drug characteristic data to generate positron drug targeted tissue domain diffusion data; Based on the positron drug injection blood vessel domain simulation data and the positron drug targeted tissue domain diffusion data, the injection drug flow diffusion simulation distribution analysis of the positron drug is performed to generate the injection drug simulation distribution data.

[0008] Furthermore, the injection drug flow diffusion simulation distribution analysis of the positron drug based on the positron drug injection blood vessel domain simulation data and the positron drug targeted tissue domain diffusion data includes: Perform diffusion gradient analysis based on the positron drug targeted tissue domain diffusion data to generate positron drug targeted tissue domain diffusion gradient data; The positron drug targeted tissue domain transfer boundary conditions are set through the positron drug targeted tissue domain diffusion gradient data, and the positron drug targeted tissue domain transfer boundary conditions are used to perform positron drug injection simulation distribution analysis on the positron drug injection vascular domain simulation data and the positron drug targeted tissue domain diffusion data to generate injection drug simulation distribution data.

[0009] Furthermore, the intelligent injection control model design module includes the following functions: Perform injection drug simulation distribution trend feature analysis based on injection drug simulation distribution data to generate injection drug simulation distribution trend feature data; Perform positron drug injection state dimensional feature analysis based on injection drug simulation distribution trend feature data, generate positron drug injection state dimensional feature data, and set positron drug injection state space parameters based on the positron drug injection state dimensional feature data; Setting the positron drug injection action space parameters according to the preset positron drug injection control parameters; Design a reward function for positron injection; A deep reinforcement learning model for personalized positron drug injection control is constructed using the positron drug injection reward function, positron drug injection state space parameters, and positron drug injection action space parameters to obtain a personalized positron drug injection control model.

[0010] Furthermore, the reward function for positron drug injection is designed to include: Defining the injection drug simulation distribution stage based on the injection drug simulation distribution trend characteristic data to obtain the injection drug simulation distribution stage data; According to the injection drug simulation distribution stage data, the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage are analyzed, and the positron drug injection reward function is designed based on the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage.

[0011] Furthermore, the positron drug injection reward function in the positron drug injection reward function is designed based on the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage as follows: ; In the formula, It is represented by the positron drug injection control reward value, Expressed as the positron drug dose accuracy weight parameter, Represents the positron drug injection simulation dose data, Represents the target dose data of positron drug injection, It represents the maximum allowable dose deviation data of positron drug injection. Expressed as the positron drug radiation safety weight parameter, Expressed as a natural parameter, Represents the cumulative radiation data of the positron drug target area simulation, It is expressed as the radiation safety attenuation coefficient of the positron drug target area, It is expressed as the positron drug injection pressure stability weight parameter, It is expressed as the pressure fluctuation amplitude of positron drug injection.

[0012] Furthermore, the positron drug intelligent injection control module includes the following functions: Based on the injection drug simulation distribution data, the targeted tissue drug impact benefit analysis of the injection drug flow rate is performed to obtain the targeted tissue drug impact benefit data of the injection drug flow rate; Design an injection drug flow rate optimization strategy based on the targeted tissue drug impact benefit data of the injection drug flow rate, and optimize the personalized injection control model of the positron drug through the injection drug flow rate optimization strategy to generate an optimized personalized injection control model of the positron drug; The injection drug simulation distribution data is transmitted to the optimized positron drug personalized injection control model to perform positron drug personalized injection control analysis and generate positron drug personalized injection control parameters; The intelligent injection operation of positron drug is performed based on the personalized injection control parameters of positron drug.

[0013] Furthermore, the targeted tissue drug impact benefit analysis of the injection drug flow rate based on the injection drug simulation distribution data includes: Through the positron drug targeted tissue domain transfer boundary condition, the injection drug simulation distribution data is respectively analyzed for the injection drug flow rate and the targeted tissue boundary transfer rate correlation analysis and the injection drug flow rate and the targeted tissue boundary transfer absorption correlation analysis, so as to obtain the injection drug flow rate-targeted tissue boundary transfer rate correlation data and the injection drug flow rate and the targeted tissue boundary transfer absorption correlation data respectively; The targeted tissue drug impact benefit analysis of the injection drug flow rate is performed based on the injection drug flow rate-target tissue boundary transfer rate correlation data and the injection drug flow rate and targeted tissue boundary transfer absorption correlation data to obtain the targeted tissue drug impact benefit data of the injection drug flow rate.

[0014] The beneficial effect of the present application is that the present invention acquires the patient's physiological sign data and the patient's historical lesion metabolic imaging data, analyzes the patient's global symptom monitoring characteristics, and realizes personalized disease monitoring and analysis. By collecting the patient's key physiological data such as body surface temperature, heart rate variability and blood flow rate in real time through biosensors, the patient's physiological state can be dynamically tracked to improve the accuracy of symptom monitoring. By performing feature analysis on the patient's historical lesion metabolic imaging data and combining the physiological sign data for time-series spatial alignment, the global symptom monitoring feature analysis can accurately depict the metabolic characteristics of the lesion and its evolution trend. This method not only improves the accuracy of lesion positioning, but also provides more accurate personalized data support during the progression of the disease, which is helpful for the subsequent intelligent injection control of positron drugs. The positron drug is characteristically monitored by a radioactive detector, the drug activity value and decay rate data are obtained in real time, and the high-precision flow diffusion simulation distribution analysis of the positron drug is realized in combination with the global symptom monitoring feature data. The flow characteristics of the drug in the blood vessel and the diffusion process in the target tissue are effectively simulated to improve the injection accuracy of the positron drug. Through voxelized data analysis of lesion metabolic imaging features, the target tissue area of ​​positron drugs can be accurately locked, and vascular domain simulation analysis can be performed based on patient physiological signs data and drug property data to optimize the flow rate and distribution of drugs. In addition, through diffusion gradient analysis and targeted tissue domain transfer boundary condition setting, the diffusion path of drugs can be further optimized to ensure the precise delivery of drugs in the target area, improve the efficiency of disease diagnosis and treatment, and reduce the risk of misinjection and dose deviation. By designing a composite reward function (dose accuracy weight, radiation safety attenuation coefficient, pressure stability parameter), the system can trade-off and optimize dose accuracy, radiation safety and pressure stability, thereby improving the safety and effectiveness of the injection process. For example, the exponential decay term in the reward function can effectively suppress the risk of excessive radiation in the target area, and the quadratic penalty term reduces the mechanical damage to blood vessels caused by injection pressure fluctuations. Through the correlation analysis of injection flow rate and target area transfer rate / absorption efficiency (such as establishing a flow rate-absorption rate response curve), the control strategy is dynamically corrected. Based on drug property data and patient physiological characteristics, a gradient model of drug flow diffusion is established through vascular domain simulation, targeted tissue diffusion analysis and a composite reward function. The drug distribution is dynamically predicted using the transfer boundary conditions, and the injection parameters (such as flow rate and pressure) are adjusted in real time in combination with the deep reinforcement learning model to ensure efficient enrichment of the drug in the targeted area, reduce non-target tissue exposure, and improve treatment accuracy. The model is simulated and optimized through the transfer boundary conditions of the targeted tissue domain to ensure that the distribution of the drug meets the treatment needs, reduce drug dosage deviation, and improve the stability and safety of injection. The injection drug flow rate is correlated with the transfer rate and absorption characteristics of the targeted tissue boundary to more accurately evaluate the cumulative effect of the drug in the targeted tissue, thereby optimizing the injection flow rate and improving the drug utilization rate.According to the individual characteristics of the patient and the drug diffusion situation, the injection strategy is adjusted in real time to achieve precise intelligent injection operations. Ultimately, this method effectively improves the targeted delivery accuracy of positron drugs, reduces drug waste, reduces the risk of side effects, and provides a more intelligent solution for precision medicine.

[0015] Therefore, the intelligent injection system of the positron drug of the present invention analyzes the global disease monitoring characteristics based on the patient's physiological sign data and the lesion metabolism imaging data, that is, the intelligent comprehensive analysis of the patient's target tissue lesion characteristics combined with physiological signs, realizes the accurate identification of the patient's individual characteristics. It overcomes the deficiency of relying only on the preset injection scheme, so that the injection strategy can dynamically adapt to the patient's real-time physiological state and lesion metabolism characteristics, thereby improving the therapeutic effectiveness of the drug. Secondly, the key characteristic data of the positron drug is monitored by a radioactive detector, and a deep reinforcement learning model is constructed in combination with the drug simulation distribution trend analysis to realize personalized intelligent injection control, which can optimize the injection parameters in real time according to the flow and diffusion of the drug, ensure the uniform distribution of the drug in the target tissue, reduce errors, and improve image quality and therapeutic effects. And optimize the drug flow rate through simulation analysis, so that its transmission and absorption in the target tissue area are more efficient, thereby improving the utilization rate of the drug, reducing the risk of radiation exposure, and reducing drug waste, providing a more intelligent and adaptive injection scheme, and providing technical guarantee for precision medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of the module flow of an intelligent injection system for positron drugs according to the present invention; Figure 2 for Figure 1 Detailed functional flow chart of the injection drug simulation distribution analysis module; Figure 3 for Figure 1 Detailed functional flow chart of the intelligent injection control model design module; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0017] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0018] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0019] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0020] To achieve this, please refer to Figures 1 to 3 The present invention provides an intelligent injection system for positron drugs, comprising the following modules, referring to Figure 1 The above is a schematic diagram of the module flow of an intelligent injection system of a positron drug according to the present invention. In this embodiment, the intelligent injection system of a positron drug includes: S1: A global symptom monitoring feature analysis module, which is used to obtain the patient's physiological sign data and the patient's historical lesion metabolic imaging data; perform global symptom monitoring feature analysis through the patient's physiological sign data and the patient's historical lesion metabolic imaging data to generate global symptom monitoring feature data; In an embodiment of the present invention, a wearable device (such as a smart wristband or a medical-grade patch sensor) with integrated multiple physiological parameter acquisition functions is used to continuously monitor the patient's physiological signs. The patient's historical PET-CT or PET-MR image data is retrieved from the hospital P system. Image data from different periods (such as 1 month ago, 3 months ago, and 6 months ago) are selected for subsequent lesion change trend analysis. The image data is transmitted to the intelligent injection system through the DICOM protocol interface and subjected to standardized preprocessing, including noise filtering, contrast enhancement, and artifact removal, to improve image quality and subsequent analysis accuracy. The deep learning model based on the 3D U-Net structure automatically segments and extracts features from the patient's lesion metabolic images, normalizes the PET-CT images (SUV normalization range: [0, 1]), and uses Gaussian filtering to remove noise. The 3D U-Net model is used to identify lesion images and automatically segment lesions, and the Dice coefficient, an evaluation index for segmentation accuracy, reaches above 0.85. The segmented lesions are subjected to metabolic feature analysis, including lesion metabolic feature values ​​and 3D voxelized lesion metabolic activity distribution maps for subsequent analysis. The patient's physiological sign data and lesion metabolic imaging feature data are used for global disease monitoring feature analysis. The lesion metabolic imaging data are divided into 3D voxels based on the identified lesion segmentation area, and the corresponding SUV values ​​are stored in a three-dimensional matrix to form voxelized data. The physiological sign data are aligned using a time series interpolation algorithm (such as linear interpolation or spline interpolation) to match the time scale of the imaging data. For example, heart rate data is in seconds, and PET-CT scans are acquired in minutes, so time scale mapping is required. The correlation between lesion metabolic activity and physiological signs is calculated to obtain global disease monitoring feature data that combines lesion metabolic imaging feature data with patient physiological sign data to guide the intelligent injection strategy of positron drugs.

[0021] S2: injection drug simulation distribution analysis module, used to monitor the positron drug characteristics of positron drugs using radioactive detectors to generate positron drug characteristic data; based on the global disease monitoring feature data and positron drug characteristic data, perform injection drug flow diffusion simulation distribution analysis of positron drugs to generate injection drug simulation distribution data; In an embodiment of the present invention, a radioactive detector (such as a PET scanner) is used to monitor the characteristics of positron-emitting drugs in real time. The monitoring of the characteristics of positron-emitting drugs includes real-time acquisition of the activity value and decay rate of the drug. The injected positron-emitting drugs are detected using PET scanning technology, and the activity value is obtained by analyzing the intensity of the gamma rays released by the positron-emitting drugs. Assume that the decay rate constant λ of the drug is 0.05 min -1 , the initial positron drug activity value data is , the decay rate will be according to the formula Calculate the positron drug activity value data and update the decay state of the drug in real time. Use the voxelized data of the lesion metabolic image characteristics in the global disease monitoring feature data to analyze the positron drug targeted tissue area. Use image segmentation technology to analyze the lesion area and its metabolic characteristics, and use a threshold-based segmentation method (such as the Otsu algorithm) to extract the lesion area. Set the threshold to SUV ≥ 3, and select the metabolically active area as the target tissue area. According to the voxelized data of the metabolically active area, combined with the metabolic characteristics of the drug, determine the target tissue area that the drug should reach, and generate the targeted tissue area data. For example, if the SUV of the lesion area is 4.2 and meets the high metabolic activity, it will be delineated as a targeted tissue area, and the targeted drug will be concentrated in this area. Simulate the positron drug injection vascular domain based on the patient's physiological sign data and positron drug characteristic data of the global disease monitoring feature data. Use fluid mechanics models (such as the Navier-Stokes equation) to simulate the flow of drugs in blood vessels. The blood flow rate data of the patient (for example, the venous blood flow rate is 10 cm / s) is combined with the injection rate of the drug (such as 2 mL / min) to simulate the flow path of the drug in the blood vessel and generate injection blood vessel domain simulation data. According to the positron drug targeted tissue area data, the global disease monitoring feature data is further analyzed for the disease monitoring feature data of the targeted tissue domain. The disease data in the targeted tissue area are analyzed using a machine learning algorithm, and features such as metabolic rate changes and tissue blood flow are extracted. Analysis parameters such as the metabolic rate change range are set to ±5%, and adjusted according to individual differences of patients. The diffusion analysis of the positron drug targeted tissue domain is performed based on the disease monitoring feature data of the targeted tissue domain and the positron drug characteristic data. The distribution of the drug in the targeted area is predicted using a diffusion model (such as Fick's diffusion law). Assuming that the diffusion coefficient of the drug in the targeted tissue is D = 0.05 cm² / s, the diffusion rate of the drug is adjusted according to the density and metabolic characteristics of the lesion. Through numerical simulation, the diffusion data of the drug in the targeted tissue area is obtained. The flow diffusion simulation distribution analysis of the injected drug is performed based on the positron drug injection blood vessel domain simulation data and the targeted tissue domain diffusion data. Combining the flow and diffusion data of the vascular domain and the target tissue domain, the computational fluid dynamics (CFD) model is used to simulate how the drug diffuses from the blood vessels to the target tissue during the injection process. The distribution map of the flow and diffusion of the drug during the injection process is generated, and the concentration distribution of the drug in the target area and its trend over time are analyzed. In addition, the simulation analysis process sets different parameters to obtain different injection drug distribution data, which is used to optimize the drug injection plan.

[0022] S3: Intelligent injection control model design module, used to analyze the injection drug simulation distribution trend characteristics according to the injection drug simulation distribution data, and generate injection drug simulation distribution trend characteristic data; based on the injection drug simulation distribution trend characteristic data, a deep reinforcement learning model for personalized injection control of positron drugs is constructed to obtain a personalized injection control model for positron drugs; In an embodiment of the present invention, the obtained injection drug simulation distribution data is used for trend feature analysis. The drug distribution data is analyzed for time series changes to extract the concentration changes, diffusion speed and other characteristics of the drug at different time points. Time series analysis methods, such as smoothing filtering, Fourier transform, etc., are used to identify the change trend and periodic characteristics of drug distribution. Through trend analysis, the rate of change of drug concentration and the time point when the peak value appears (such as reaching the maximum concentration after 3 minutes) can be calculated, and this information is provided as trend feature data to subsequent steps. According to the injection drug simulation distribution trend feature data, the injection state dimension of the drug is analyzed, and according to the trend feature data, the various state dimensions in the drug injection process are analyzed, such as drug injection rate, drug concentration, diffusion effect of the target area, etc. A multidimensional feature analysis algorithm (such as principal component analysis PCA, cluster analysis, etc.) is used to extract the main features of the injection state. The action space of drug injection is defined according to the preset drug injection control parameters, and the preset drug injection control parameters are established by historical positron injection control. The action space of drug injection is set by a deep reinforcement learning model. The action space includes multiple dimensions such as injection rate, injection time, and injection concentration. The execution of each action corresponds to a specific injection scheme and drug distribution effect. The set action space parameters include: injection rate: 0.1 ml / s to 1 ml / s; drug injection time: 5 seconds to 60 seconds; drug concentration: 10μM to 100 μM; these parameters will be dynamically adjusted in the reinforcement learning model to optimize the injection scheme. The reward function is set to maximize the targeting accuracy of the drug, minimize the side effects of the drug, and ensure the safety of the drug while ensuring the breadth of drug distribution. The reward function can be designed based on the deviation between the drug concentration and the target concentration, the radiation dose, the safety threshold, etc. A personalized injection control model for positron drugs is constructed using the reward function, injection state space parameters, and injection action space parameters. Using reinforcement learning algorithms (such as DQN, PPO, etc.), through multiple training and iterations, the set reward function, state space, and action space are used to optimize the drug injection strategy. The model can automatically adjust various parameters in the injection process to achieve personalized precision treatment. During the training process, the learning rate is set to 0.001, the batch size is 32, and the number of training rounds is 1000. After each round of training, the model will evaluate the effectiveness of the strategy based on the reward function and adjust the parameters of drug injection to ensure the optimal injection process, so as to obtain a personalized injection control model for positron drugs.

[0023] S4: Positron drug intelligent injection control module, used to perform targeted tissue drug impact benefit analysis of injection drug flow rate based on injection drug simulation distribution data, and obtain targeted tissue drug impact benefit data of injection drug flow rate; perform model optimization processing on personalized positron drug injection control model based on targeted tissue drug impact benefit data of injection drug flow rate, and generate optimized personalized positron drug injection control model; transmit injection drug simulation distribution data to optimized personalized positron drug injection control model to perform personalized positron drug injection control analysis, and generate personalized positron drug injection control parameters; perform positron drug intelligent injection operation based on personalized positron drug injection control parameters.

[0024] In the embodiment of the present invention, based on the injection drug simulation distribution data, the drug flow rate and the distribution benefit of the targeted tissue drug are analyzed, and the drug transmission efficiency of different drug flow rates to the targeted tissue is determined, thereby providing a basis for the subsequent optimization of the injection drug flow rate. The technical means used include numerical simulation and optimization algorithms, which analyze the distribution effect of the drug in the targeted tissue at different injection flow rates and calculate its impact benefit data. According to the targeted tissue drug impact benefit data of the injection drug flow rate, the drug flow rate optimization strategy is designed. The goal of the optimization strategy is to select the optimal flow rate to ensure that the drug can be evenly and efficiently distributed to the targeted tissue, while avoiding the side effects caused by too fast or too slow injection of the drug. For example, too fast injection flow rate may cause excessive accumulation of drugs in the targeted area, while too slow flow rate may cause the drug to fail to achieve the expected therapeutic effect. When designing the optimization strategy, the flow rate parameters are tuned by genetic algorithms, simulated annealing algorithms or other optimization algorithms to ensure that the flow rate of the drug can meet the requirements of maximizing the drug benefit. The optimized flow rate parameters will be applied to the personalized injection control model of positron drugs to improve the injection effect. The output results of the optimization strategy will guide the subsequent injection flow rate selection. The optimized drug flow rate parameters and the simulated distribution data of the injected drug are transmitted to the optimized personalized injection control model of the positron drug for injection control analysis. The analysis will dynamically adjust the injection control scheme of the positron drug according to the optimized flow rate parameters, the distribution characteristics of the drug, and the physiological characteristics of the target area. Through the deep learning model or other machine learning algorithms (such as reinforcement learning), the analysis can generate the optimal personalized injection control parameters in real time under different treatment scenarios to ensure that the drug can reach the target area accurately and timely, and adjust according to the specific conditions of the patient. Based on the personalized injection control parameters, the intelligent injection operation of the positron drug is performed. This operation process is completed by an automated injection device. The system monitors various parameters during the injection process, such as flow rate, pressure, temperature, etc. in real time to ensure that the injection operation is performed according to the optimal control plan. The injection device will perform the injection operation by automatically adjusting parameters such as flow rate and pressure. The system will continuously compare with the patient's real-time physiological data and adjust the parameters during the injection process to ensure that the drug is accurately injected into the target tissue to achieve the expected therapeutic effect. The injection device adjusts the flow rate, pressure, injection time and injection node in real time through sensors and feedback systems. During the injection process, the system will dynamically adjust the injection control parameters based on real-time feedback data (such as blood flow rate, tissue reaction, etc.) to ensure accurate injection of the drug.

[0025] Furthermore, the global symptom monitoring feature analysis module includes the following functions: Collecting and processing the patient's physiological signs in real time through a biosensor to generate the patient's physiological sign data, wherein the patient's physiological sign data includes body surface temperature data, heart rate variability data, and blood flow rate data; Obtain historical lesion metabolic imaging data of patients from the medical system; Perform lesion metabolic imaging feature analysis on the patient's historical lesion metabolic imaging data to generate lesion metabolic imaging feature data; Based on the lesion metabolic imaging feature data and the patient's physiological signs data, a global symptom monitoring feature analysis is performed to generate global symptom monitoring feature data.

[0026] In an embodiment of the present invention, a wearable device (such as a smart wristband or a medical-grade patch sensor) with integrated multiple physiological parameter acquisition functions is used to continuously monitor the patient's physiological signs. Among them, the parameters of the sensor are set as follows: a high-precision temperature sensor (measurement range: 30°C-42°C, accuracy: ±0.1°C) is used to record the patient's body surface temperature in real time. Photoplethysmography (PPG) and electrocardiogram (ECG) sensors are used to detect heart rate fluctuations and calculate heart rate variability parameters such as SDNN (standard deviation), RMSSD (root mean square of adjacent heartbeat intervals), and LF / HF (low frequency / high frequency power ratio). The blood flow velocity of a patient's specific blood vessels (such as the carotid artery or radial artery) is measured by an ultrasonic Doppler sensor (measurement range: 0.1 cm / s - 200 cm / s, accuracy: ±0.5 cm / s) to reflect the blood circulation situation. The data acquisition frequency is set to 5 times per second and transmitted to the patient monitoring system in real time to ensure the continuity and stability of the data. The patient's historical PET-CT or PET-MR imaging data were retrieved from the hospital's PACS (Picture Archiving and Communication System) or EMR (Electronic Medical Record) system. These imaging data contain the following key information: DICOM (Digital Imaging and Communications in Medicine) standard imaging format to ensure that the data can be parsed by medical imaging processing software. And set indicators such as SUV (standard uptake value), TBR (target tissue / background ratio), MTV (metabolic tumor volume). Select imaging data from different periods (such as 1 month ago, 3 months ago, and 6 months ago) for subsequent lesion change trend analysis. The imaging data was transmitted to the intelligent injection system through the DICOM protocol interface and subjected to standardized preprocessing, including noise filtering, contrast enhancement, and artifact removal, to improve image quality and subsequent analysis accuracy. The deep learning model based on the 3D U-Net structure automatically segmented and extracted features from the patient's lesion metabolic images, normalized the PET-CT images (SUV normalization range: [0,1]), and used Gaussian filtering to remove noise. The 3D U-Net model was used to identify lesion images and automatically segment lesions, with the Dice coefficient, an evaluation index for segmentation accuracy, reaching above 0.85. Metabolic characteristics of the segmented lesions were analyzed, including: maximum standardized uptake value (SUV_max), mean standardized uptake value (SUV_mean), metabolic volume (MTV), total tumor metabolic activity (TLG), and metabolic boundary gradient (MBG).Using the B-Spline-based registration algorithm, the current image is aligned with the historical image to achieve cross-time lesion change analysis and generate lesion metabolic image feature data, including the above-mentioned feature values ​​and 3D voxelized lesion metabolic activity distribution map for subsequent analysis. The patient's physiological sign data and lesion metabolic image feature data are used for global disease monitoring feature analysis. The lesion metabolic image data is divided into 3D voxels based on the identified lesion segmentation area, and the corresponding SUV values ​​are stored in a three-dimensional matrix to form voxelized data. For example, when SUV_max=8.5, the area with SUV>5 is set as a high metabolic active area, SUV 3-5 is a moderate metabolic area, and SUV<3 is a low metabolic area. The physiological sign data is timestamped and aligned with the acquisition time of the image data through the ICU time synchronization standard (such as the HL7 protocol). The physiological sign data is aligned using a time series interpolation algorithm (such as linear interpolation or spline interpolation) to match the time scale of the image data. For example, heart rate data is in seconds, and PET-CT scans are acquired in minutes, so time scale mapping is required. The correlation between lesion metabolic activity and physiological signs is calculated. For example, the Pearson correlation coefficient between SUV_max and heart rate variability (HRV) is 0.72, indicating that high metabolic tumor areas may cause abnormal heart rate fluctuations. The Spearman correlation coefficient between blood flow rate and lesion metabolic volume (MTV) is -0.58, indicating that reduced blood flow affects tumor metabolic activity. Finally, global disease monitoring feature data that combines lesion metabolic imaging feature data with patient physiological sign data is obtained to guide the intelligent injection strategy of positron drugs.

[0027] Furthermore, the global disease monitoring feature analysis based on the lesion metabolic imaging feature data and the patient's physiological sign data includes: Perform voxel division processing on the lesion metabolic image feature data according to the lesion metabolic image feature data to generate voxelized lesion metabolic image feature data; The lesion metabolic imaging feature voxelized data and the patient's physiological sign data are aligned in time and space, and a global disease monitoring feature analysis is performed based on the lesion metabolic imaging feature voxelized data and the patient's physiological sign data after the time and space alignment to generate global disease monitoring feature data.

[0028] In an embodiment of the present invention, the lesion metabolic imaging feature data is divided into 3D voxels, and the PET-CT scan data is preprocessed using medical image processing software (such as 3D Slicer or ITK-Snap), and the identified lesion area is converted into voxelized data. In this process, the spatial resolution of each voxel is set to 0.5mm × 0.5mm × 0.5mm to ensure the accurate expression of the details of the lesion area. The standard uptake value (SUV) threshold of the metabolically active area is set to SUV ≥ 3 as a high metabolic area, SUV 1.5-3 as a moderate metabolic area, and SUV < 1.5 as a low metabolic area. After voxelization, each lesion area will be divided into several 3D voxels, and each voxel stores its corresponding SUV value to generate lesion metabolic imaging feature voxelized data. The data contains information such as metabolically active areas and metabolically inactive areas, which can more intuitively reflect the metabolic status of the lesion. Temporal and spatial alignment is performed on the lesion metabolic imaging feature voxelized data and the patient's physiological sign data. The frequency of physiological sign data collection is once per second. The data provided by the sensor include heart rate variability (HRV), surface temperature, blood flow rate, etc., and these data need to be aligned with the time axis of the PET scan. The specific steps are: Use interpolation methods (such as linear interpolation or spline interpolation) to map the physiological sign data to the time node of the PET image to ensure that the data is compared at the same time point. Match the 3D coordinate system of the lesion metabolic imaging data with the spatial information of the patient's physiological data to ensure the spatial consistency of the physiological sign data and the corresponding lesion metabolic area. For example, if the patient's PET scan is obtained after 2 hours, and the heart rate data at this time is 70bpm, the data is mapped to the 2-hour moment of the PET image through interpolation technology to ensure the synchronization of the data. After the temporal and spatial alignment, the lesion metabolic imaging feature voxelized data is combined with the physiological sign data to perform global disease monitoring feature analysis. For example, by calculating the Pearson correlation coefficient between the SUV value and the heart rate variability (HRV), the correlation is 0.65, indicating that the high metabolic lesion area may be associated with greater heart rate variability. The relationship between body temperature, blood flow rate and lesion metabolic volume (MTV) was comprehensively analyzed, and the change trend of the lesion was modeled using regression analysis. It was found that the increase in metabolically active areas of certain lesions may be closely related to the decrease in blood flow rate, so as to obtain the patient's global disease monitoring characteristic data. These data not only reflect the metabolic status of the lesion, but also include the impact of physiological signs on lesion activity. For example, a multidimensional data set including voxelized lesion area data, lesion area metabolic parameters, heart rate variability, blood flow rate, body temperature and other characteristics is generated and stored in the patient's electronic health record system for the subsequent application of personalized positron drug injection control.

[0029] For further reference, Figure 2 As shown, Figure 1 The functional flow diagram of the injection drug simulation distribution analysis module in this example includes: S21: monitoring the positron drug characteristics of the positron drug using a radioactive detector to generate positron drug characteristic data, wherein the positron drug characteristic data includes positron drug activity value data and positron drug decay rate data; In an embodiment of the present invention, a radioactive detector (such as a PET scanner) is used to monitor the characteristics of positron drugs in real time. The monitoring of the characteristics of positron drugs includes real-time acquisition of the activity value and decay rate of the drug. The injected positron drug is detected using PET scanning technology, and the activity value is obtained by analyzing the intensity of the gamma rays released by the positron drug. Taking [18F]FDG as an example, the detection sensitivity of the radioactive detector is set to 5000 cps (counts persecond), and the activity value of the positron drug (such as 1000 kBq / mL) is obtained by comparing with the preset standard activity model. Based on the radioactive decay law of the drug, the decay rate is calculated using an exponential decay model by observing the trend of the drug activity over time. Assume that the decay rate constant λ of the drug is 0.05 min -1 , the initial positron drug activity value data is , the decay rate will be according to the formula Calculate and obtain the positron drug activity value data, and update the drug decay state in real time.

[0030] S22: positron drug targeted tissue region analysis is performed using voxelized data of lesion metabolic imaging features of global disease monitoring feature data to generate positron drug targeted tissue region data; In an embodiment of the present invention, positron drug targeted tissue region analysis is performed through voxelized data of lesion metabolic imaging features in global disease monitoring feature data. Image segmentation technology is used to analyze the lesion area and its metabolic characteristics, and a threshold-based segmentation method (such as the Otsu algorithm) is used to extract the lesion area. The threshold is set to SUV ≥ 3, and the metabolically active area is selected as the target tissue area. Based on the voxelized data of the metabolically active area and the metabolic characteristics of the drug, the target tissue area that the drug should reach is determined, and the target tissue area data is generated. For example, if the SUV of the lesion area is 4.2 and meets the high metabolic activity, it is delineated as a target tissue area, and the targeted drug will be concentrated and diffused in this area.

[0031] S23: performing simulation analysis of the positron drug injection vascular domain based on the patient's physiological sign data and the positron drug characteristic data of the global disease monitoring characteristic data, and generating positron drug injection vascular domain simulation data; In an embodiment of the present invention, a simulation analysis of the positron drug injection vascular domain is performed based on the patient's physiological sign data and the positron drug characteristic data of the global disease monitoring characteristic data. The flow of drugs in the blood vessels is simulated using a fluid mechanics model (such as the Navier-Stokes equation). The flow path of the drug in the blood vessels is simulated by combining the patient's blood flow rate data (for example, the venous blood flow rate is 10 cm / s) with the drug injection rate (such as 2 mL / min) to generate injection vascular domain simulation data. The stability and distribution of drug flow in the vascular domain are analyzed to ensure that the drug does not diffuse or deposit excessively during the injection process.

[0032] S24: performing targeted tissue domain disease monitoring characteristic data analysis on the global disease monitoring characteristic data according to the positron drug targeted tissue region data to generate targeted tissue domain disease monitoring characteristic data; In an embodiment of the present invention, the global symptom monitoring feature data is further analyzed for the targeted tissue domain symptom monitoring feature data based on the positron drug targeted tissue region data. A machine learning algorithm (such as a random forest or support vector machine and corresponding historical targeted tissue medical data to establish a supervised learning model) is used to analyze the symptom data in the targeted tissue region and extract features, such as metabolic rate changes, tissue blood flow, etc. Analysis parameters such as the metabolic rate variation range of ±5% are set and adjusted according to individual differences in patients. By comparing the symptom data of different targeted tissue regions, targeted tissue domain symptom monitoring feature data is generated to provide support for subsequent drug diffusion analysis.

[0033] S25: performing positron drug targeted tissue domain diffusion analysis according to the targeted tissue domain disease monitoring characteristic data and the positron drug characteristic data, and generating positron drug targeted tissue domain diffusion data; In an embodiment of the present invention, diffusion analysis of the positron drug targeted tissue domain is performed based on the targeted tissue domain disease monitoring characteristic data and the positron drug characteristic data. The distribution of the drug in the targeted area is predicted using a diffusion model (such as Fick's diffusion law). Assuming that the diffusion coefficient of the drug in the targeted tissue is D = 0.05 cm² / s, the diffusion rate of the drug is adjusted according to the density and metabolic characteristics of the lesion. Through numerical simulation, the diffusion data of the drug in the targeted tissue area (such as a diffusion radius of 1.2 cm) is obtained, and the drug distribution strategy is adjusted accordingly.

[0034] S26: performing injection drug flow diffusion simulation distribution analysis of the positron drug based on the positron drug injection blood vessel domain simulation data and the positron drug targeted tissue domain diffusion data to generate injection drug simulation distribution data.

[0035] In an embodiment of the present invention, flow diffusion simulation distribution analysis of injected drugs is performed based on positron drug injection vascular domain simulation data and targeted tissue domain diffusion data. Combining the flow and diffusion data of the vascular domain and the targeted tissue domain, a computational fluid dynamics (CFD) model is used to simulate how the drug diffuses from the blood vessels to the targeted tissue during the injection process. A distribution diagram of the flow and diffusion of the drug during the injection process is generated, and the concentration distribution of the drug in the targeted area and its trend over time are analyzed. The simulation analysis process sets different parameters to obtain different injection drug distribution data for optimizing the drug injection scheme. For example, the simulation analysis results show that after 3 minutes, the concentration of the drug in the targeted area reaches a maximum value and stabilizes in this area, which provides a basis for subsequent injection control and dosage adjustment.

[0036] Furthermore, the injection drug flow diffusion simulation distribution analysis of the positron drug based on the positron drug injection blood vessel domain simulation data and the positron drug targeted tissue domain diffusion data includes: Perform diffusion gradient analysis based on the positron drug targeted tissue domain diffusion data to generate positron drug targeted tissue domain diffusion gradient data; The positron drug targeted tissue domain transfer boundary conditions are set through the positron drug targeted tissue domain diffusion gradient data, and the positron drug targeted tissue domain transfer boundary conditions are used to perform positron drug injection simulation distribution analysis on the positron drug injection vascular domain simulation data and the positron drug targeted tissue domain diffusion data to generate injection drug simulation distribution data.

[0037] In the embodiment of the present invention, diffusion gradient analysis is performed based on the diffusion data of the positron drug in the targeted tissue domain, and the concentration gradient of the drug in the targeted tissue is analyzed by numerical calculation methods. The diffusion rate of the drug in the tissue is calculated by solving the diffusion equation (such as Fick's first law). A three-dimensional coordinate system is established in the targeted tissue region, the drug concentration C (x, y, z) is set, and time t is used as a variable to construct a concentration dynamic change model, and the diffusion gradient is calculated based on the change in the concentration. , and obtain the diffusion gradient data of the positron drug targeted tissue domain. For example, assuming that the drug diffusion coefficient D is 0.05cm² / s, and the drug concentration C(x, y, z) changes with time t. Diffusion gradient data are generated by simulating different tissue densities and drug concentration distributions. Specifically, in a specific targeted area, the initial drug concentration at the center point is 20 μM. As time goes by, the drug gradually diffuses to the edge, and the concentration in the edge area eventually stabilizes at 5 μM. Based on the changes in concentration in space and time, the drug diffusion flux J is calculated by numerical integration and other calculation methods. Based on Fick's first law, its expression is: , and thus calculate the diffusion gradient , in order to obtain the diffusion gradient data of positron drug targeted tissue domain, which can provide key basis for subsequent boundary condition setting, drug metabolism kinetics research and efficacy evaluation. According to the diffusion gradient data of positron drug targeted tissue domain, mathematical models (such as Neumann boundary conditions or Dirichlet boundary conditions) are used to set appropriate boundary conditions for the delivery of targeted tissue domain. When setting boundary conditions, it is assumed that the drug concentration at the outer boundary (outside of the tissue) of the targeted area is C_ext = 5 μM, the inner boundary concentration is C_int = 20 μM, and the drug delivery rate is 0.1 μM / s. Based on these parameters, the boundary conditions of the transfer equation are set to simulate the distribution and transfer of drugs in the targeted area. According to the set targeted tissue domain delivery boundary conditions, the flow and diffusion analysis of the positron drug injection vascular domain simulation data and the targeted tissue domain diffusion data are performed to generate the final injection drug simulation distribution data. The computational fluid dynamics (CFD) model is used to simulate the process of drug injection from blood vessels to targeted tissues in combination with the set delivery boundary conditions. The simulation analysis takes into account the speed of drug flow in blood vessels, the effect of diffusion to targeted tissues, and the diffusion and delivery process of drugs in tissues. For example, assuming that the flow rate of the drug in the blood vessel is 10 cm / s, the permeability coefficient of the target tissue is 0.02 cm / s, and the diffusion coefficient of the tissue is 0.05 cm² / s. The simulation results show that the drug flows from the blood vessel into the target tissue within 3 minutes, the diffusion of the drug concentration in the target area gradually reaches a stable state, and the concentration of the drug in the target tissue reaches a peak value (about 50 μM) after 3 minutes. The generated injection drug simulation distribution data will show the concentration distribution of the drug in the vascular domain and the target tissue domain, ensuring that the flow rate, targeted distribution and metabolic rate of the drug meet the expected goals.

[0038] For further reference, Figure 3 As shown, Figure 1 The functional flow diagram of the intelligent injection control model design module in this example includes: S31: performing injection drug simulation distribution trend characteristic analysis according to injection drug simulation distribution data to generate injection drug simulation distribution trend characteristic data; In an embodiment of the present invention, the obtained simulated distribution data of injected drugs is used to perform trend feature analysis. The drug distribution data is analyzed for time series changes to extract characteristics such as drug concentration changes and diffusion speed at different time points. Time series analysis methods, such as smoothing filtering and Fourier transform, are used to identify the changing trends and periodic characteristics of drug distribution. Assume that the distribution concentration data of the drug in the target tissue at different time points is [20 μM, 35 μM, 50 μM, 45 μM]. Through trend analysis, the rate of change of drug concentration and the time point when the peak value appears (such as reaching the maximum concentration after 3 minutes) can be calculated, and this information can be provided as trend feature data to subsequent steps.

[0039] S32: performing positron drug injection state dimensional feature analysis according to the injection drug simulation distribution trend feature data, generating positron drug injection state dimensional feature data, and setting positron drug injection state space parameters according to the positron drug injection state dimensional feature data; In an embodiment of the present invention, the injection state dimension of the drug is analyzed according to the simulated distribution trend characteristic data of the injected drug, and the various state dimensions in the drug injection process are analyzed according to the trend characteristic data, such as the drug injection rate, drug concentration, diffusion effect of the targeted area, etc. A multidimensional feature analysis algorithm (such as principal component analysis PCA, cluster analysis, etc.) is used to extract the main features of the injection state. For example, when analyzing the drug injection process, it is assumed that the drug injection rate (ml / s) is 0.5 ml / s and the drug concentration change rate is 5 μM per minute. By analyzing these data, the main dimensional feature data of the drug injection process are obtained, including the injection speed, the time-varying characteristics of the drug concentration, etc.

[0040] S33: setting the positron drug injection action space parameters according to the preset positron drug injection control parameters; In an embodiment of the present invention, the action space of drug injection is defined according to preset drug injection control parameters, and the preset drug injection control parameters are established through historical positron injection control. Through a deep reinforcement learning model, the action space of drug injection is set. The action space includes multiple dimensions such as injection rate, injection time, and injection concentration. The execution of each action corresponds to a specific injection scheme and drug distribution effect. The set action space parameters include: injection rate: 0.1 ml / s to 1 ml / s; drug injection time: 5 seconds to 60 seconds; drug concentration: 10 μM to 100 μM; these parameters will be dynamically adjusted in the reinforcement learning model to optimize the injection scheme.

[0041] S34: Design reward function for positron drug injection; In the embodiment of the present invention, a reward function is set to maximize the targeting accuracy of the drug, minimize the side effects of the drug, and ensure the safety of the drug while ensuring the breadth of drug distribution. The reward function can be designed based on the deviation between the drug concentration and the target concentration, the radiation dose, the safety threshold, etc.

[0042] S35: A deep reinforcement learning model for personalized positron drug injection control is constructed using the positron drug injection reward function, the positron drug injection state space parameters, and the positron drug injection action space parameters to obtain a personalized positron drug injection control model.

[0043] In an embodiment of the present invention, a personalized injection control model of positron drugs is constructed using a reward function, injection state space parameters, and injection action space parameters. A reinforcement learning algorithm (such as DQN, PPO, etc.) is used to optimize the drug injection strategy through multiple training and iterations using the set reward function, state space, and action space. The model can automatically adjust various parameters during the injection process to achieve personalized precision treatment. During the training process, the learning rate is set to 0.001, the batch size is 32, and the number of training rounds is 1000. After each round of training, the model will evaluate the effectiveness of the strategy based on the reward function, adjust the parameters of the drug injection, and ensure that the injection process is optimal to obtain a personalized injection control model for positron drugs.

[0044] Furthermore, the reward function for positron drug injection is designed to include: Defining the injection drug simulation distribution stage based on the injection drug simulation distribution trend characteristic data to obtain the injection drug simulation distribution stage data; According to the injection drug simulation distribution stage data, the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage are analyzed, and the positron drug injection reward function is designed based on the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage.

[0045] In an embodiment of the present invention, different stages of drug distribution are defined by analyzing the characteristic data of the simulated distribution trend of the injected drug. The implementation method is as follows: According to the drug concentration change trend in the simulated distribution data, combined with the distribution characteristics of the drug at different time points, the drug distribution process is divided into several stages. Common stage divisions may include the initial stage of drug injection, the drug diffusion stage, the stable stage, etc. Assume that the distribution data of the drug at different time points are as follows: Time point 1: The concentration is 20 μM, the drug diffusion is less, defined as the initial stage of injection, and the injection pressure is large; Time point 2: The concentration reaches 60 μM, the drug begins to diffuse significantly, defined as the diffusion stage, and the associated information of drug diffusion and drug injection is analyzed based on the diffusion characteristics; Time point 3: The concentration reaches the maximum value, the diffusion tends to be stable, defined as the stable stage, and the associated information of drug concentration and radiation residue is analyzed. Based on these data, the drug distribution process can be divided into three stages: the initial stage, the diffusion stage, and the stable stage. Each stage has different drug concentrations, diffusion characteristics and metabolic characteristics. The dose, radiation characteristics and pressure characteristics of each drug distribution stage are analyzed to further provide data support for the design of the reward function. Through simulation data, extract the drug injection dose data of each stage, such as the drug concentration change rate and total dose in each stage. For example, in the diffusion stage, the drug concentration changes from 50 μM to 80 μM, and the drug dose increases by 20%. Combined with the radiation characteristics of the drug, analyze the radiation intensity data of the drug in each stage. Radiation data is usually related to the distribution of the drug in the target area. In the diffusion stage, the radiation intensity is high, while in the stable stage, the radiation intensity tends to be stable. During the injection process, the pressure change of the drug will also affect the injection effect. In each stage, the pressure change trend is calculated according to the simulation data, and the pressure fluctuations in different stages are analyzed. For example, in the initial stage, the pressure change is small, while in the diffusion stage, the pressure change is large. In the diffusion stage, the drug injection dose is 50 μM, the radiation intensity is 500 mCi, and the pressure fluctuation is ±5 Pa. Through these data, different drug characteristics can be defined for each stage respectively, and provide a basis for the subsequent reward function design. Combined with the dose, radiation and pressure data of each stage, design the reward function for positron drug injection. The design of the reward function needs to comprehensively consider the dose accuracy, radiation safety and pressure stability of drug injection. The goal is to optimize each stage of the injection process so that the drug reaches the target area accurately while ensuring radiation safety and pressure stability. Through a large number of trial and error iterations, the policy network parameters are optimized so that the positron drug injection control reward value gradually converges to the maximum value, and finally the positron drug injection reward function is constructed. For example, for cancer patients, the positron drug dose accuracy weight parameter is increased to prioritize dose accuracy; for children or patients with sensitive organs, the positron drug radiation safety weight parameter is increased to enhance radiation safety.And the weights are adjusted according to different injection stages. In the initial stage, the weight parameter of positron drug injection pressure stability is increased to ensure injection stability; in the diffusion stage, the weight parameter of positron drug dose accuracy is increased to optimize dose accuracy; in the stable stage, the weight parameter of positron drug radiation safety is increased to reduce residual radiation.

[0046] Furthermore, the positron drug injection reward function in the positron drug injection reward function is designed based on the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage as follows: ; In the formula, It is represented by the positron drug injection control reward value, Expressed as the positron drug dose accuracy weight parameter, Represents the positron drug injection simulation dose data, Represents the target dose data of positron drug injection, It represents the maximum allowable dose deviation data of positron drug injection. Expressed as the positron drug radiation safety weight parameter, Expressed as a natural parameter, Represents the cumulative radiation data of the positron drug target area simulation, It is expressed as the radiation safety attenuation coefficient of the positron drug target area, It is expressed as the positron drug injection pressure stability weight parameter, It is expressed as the pressure fluctuation amplitude of positron drug injection.

[0047] The positron drug injection reward function of the present invention can accurately adjust the drug injection process by optimizing multiple key factors (drug dosage, radiation safety and pressure stability), thereby improving the personalized injection control effect of positron drugs. The first term in the reward function By measuring actual drug dosage With target dose The difference between , ensuring that the dose of drug injected is as close to the target dose as possible, thereby improving the dose accuracy and avoiding overdose or underdose. For example, assuming the target dose is 75 μM, while the simulated drug dose The maximum allowable dose deviation is 70 μM. The reward value is 10 μM, and the system will continuously adjust the injection control and gradually adjust the actual dose to the target dose to minimize the dose deviation. Specially used for radiation safety control. By evaluating the radiation data difference in the target area of ​​positron drug and the radiation attenuation coefficient By comparing the radiation safety during drug injection, we can effectively control the safety of radiation during drug injection and avoid safety risks caused by excessive radiation. 10mCi, radiation safety attenuation coefficient 2 mCi, substitute it into the formula. In this way, when the radiation difference is large, the system will significantly reduce the reward value, thereby prompting the system to adjust the drug distribution to ensure that the radiation is controlled within a safe range. The third term in the reward function The focus is on the stability of the injection pressure. Pressure fluctuation amplitude It is a factor that cannot be ignored during the injection process, especially in terms of the impact of injected drugs on human tissues. Excessive pressure fluctuations may cause tissue damage or uneven drug absorption. Maintaining pressure stability is an important goal. By negatively adjusting the square term of pressure fluctuations, the system will punish injection processes with large pressure fluctuations and encourage them to maintain stable pressure. The larger the pressure fluctuation amplitude, the smaller the reward value, which prompts the system to adjust the injection pressure to reduce pressure fluctuations and ensure a more stable injection process. The reward function can optimize the positron drug injection process from multiple angles by integrating three factors: dose accuracy, radiation safety, and pressure stability. The weight parameters of each factor can be adjusted according to different injection stages and corresponding patient information. For example, for some patients, more attention is paid to radiation safety, while for other patients, more attention may be paid to the precise control of drug doses. By adaptively adjusting these parameters, the system can perform personalized optimization according to the actual conditions of different patients. The design of the reward function provides a clear optimization goal for the training of the deep reinforcement learning (DRL) model, enabling the system to gradually optimize the drug injection process through multiple iterative learning. After each drug injection, the system adjusts the control strategy according to the actual dose, radiation, and pressure data, thereby continuously improving the accuracy and safety of drug injection. This process enables drug injection control to not only meet current treatment needs, but also adapt to dynamic changes in the patient's condition.

[0048] Furthermore, the positron drug intelligent injection control module includes the following functions: Based on the injection drug simulation distribution data, the targeted tissue drug impact benefit analysis of the injection drug flow rate is performed to obtain the targeted tissue drug impact benefit data of the injection drug flow rate; Design an injection drug flow rate optimization strategy based on the targeted tissue drug impact benefit data of the injection drug flow rate, and optimize the personalized injection control model of the positron drug through the injection drug flow rate optimization strategy to generate an optimized personalized injection control model of the positron drug; The injection drug simulation distribution data is transmitted to the optimized positron drug personalized injection control model to perform positron drug personalized injection control analysis and generate positron drug personalized injection control parameters; The intelligent injection operation of positron drug is performed based on the personalized injection control parameters of positron drug.

[0049] In the embodiment of the present invention, based on the injection drug simulation distribution data, the drug flow rate and the distribution benefit of the targeted tissue drug are first analyzed. The purpose of this analysis is to determine the drug transmission efficiency of different drug flow rates to the targeted tissue, and then provide a basis for the subsequent optimization of the injection drug flow rate. The technical means used include numerical simulation and optimization algorithms to analyze the distribution effect of the drug in the targeted tissue at different injection flow rates and calculate its impact benefit data. For example, assuming that the range of the drug in the targeted tissue is 0.5 mL / s to 2.0 mL / s, the drug distribution at different flow rates is simulated within this range to obtain the drug concentration distribution of the drug in the targeted tissue at each flow rate. Assuming that when the drug flow rate is 1.0 mL / s, the drug concentration distribution is optimal, and the benefits of the injection flow rate in the targeted tissue diffusion are collected accordingly, while when the flow rate is 1.5 mL / s, the drug diffusion range is too large, and the drug concentration in the targeted area is reduced, resulting in a decrease in the drug effect. This result will serve as the basis for the targeted tissue drug impact benefit data and provide a basis for the subsequent optimization strategy design. Design a drug flow rate optimization strategy based on the targeted tissue drug impact benefit data of the injection drug flow rate. The goal of the optimization strategy is to select the optimal flow rate to ensure that the drug can be evenly and efficiently distributed to the target tissue, while avoiding the side effects of too fast or too slow injection of the drug. For example, too fast injection flow rate may lead to excessive accumulation of drugs in the target area, while too slow flow rate may cause the drug to fail to achieve the expected therapeutic effect. When designing the optimization strategy, the flow rate parameters are tuned by genetic algorithm, simulated annealing algorithm or other optimization algorithms to ensure that the flow rate of the drug can meet the needs of maximizing the drug benefit. The optimized flow rate parameters will be applied to the personalized injection control model of positron drugs to improve the injection effect. Assume that the optimal flow rate found in the simulation is 1.0 mL / s, at which the drug can be optimally distributed in the target area, and the concentration of the drug in the target tissue reaches the target therapeutic concentration. The output results of the optimization strategy will guide the subsequent injection flow rate selection. The optimized drug flow rate parameters and the injection drug simulation distribution data are transmitted to the optimized personalized injection control model of positron drugs for injection control analysis. This analysis will dynamically adjust the injection control scheme of positron drugs according to the optimized flow rate parameters, the distribution characteristics of the drug and the physiological characteristics of the target area. Through deep learning models or other machine learning algorithms (such as reinforcement learning), this analysis can generate optimal personalized injection control parameters in real time under different treatment scenarios to ensure that the drug can reach the target area accurately and timely, and adjust according to the patient's specific conditions. Assume that the generated personalized injection control parameters include drug flow rate (1.0 mL / s), injection pressure (30 Pa) and injection duration (10 minutes), as well as the targeted tissue target of the drug after injection, which are automatically adjusted through the machine learning model to meet the patient's treatment needs. Based on the personalized injection control parameters, perform intelligent injection operations of positron drugs.This operation process is completed by an automated injection device. The system monitors various parameters during the injection process, such as flow rate, pressure, temperature, etc. in real time to ensure that the injection operation is performed according to the optimal control plan. The injection device will perform the injection operation by automatically adjusting parameters such as flow rate and pressure. The system will constantly compare with the patient's real-time physiological data and adjust the parameters during the injection process to ensure that the drug is accurately injected into the target tissue to achieve the desired therapeutic effect. The injection device adjusts the flow rate, pressure, injection time and injection node in real time through sensors and feedback systems. During the injection process, the system will dynamically adjust the injection control parameters based on real-time feedback data (such as blood flow rate, tissue response, etc.) to ensure accurate injection of the drug.

[0050] Furthermore, the targeted tissue drug impact benefit analysis of the injection drug flow rate based on the injection drug simulation distribution data includes: Through the positron drug targeted tissue domain transfer boundary condition, the injection drug simulation distribution data is respectively analyzed for the injection drug flow rate and the targeted tissue boundary transfer rate correlation analysis and the injection drug flow rate and the targeted tissue boundary transfer absorption correlation analysis, so as to obtain the injection drug flow rate-targeted tissue boundary transfer rate correlation data and the injection drug flow rate and the targeted tissue boundary transfer absorption correlation data respectively; The targeted tissue drug impact benefit analysis of the injection drug flow rate is performed based on the injection drug flow rate-target tissue boundary transfer rate correlation data and the injection drug flow rate and targeted tissue boundary transfer absorption correlation data to obtain the targeted tissue drug impact benefit data of the injection drug flow rate.

[0051] In an embodiment of the present invention, the simulated distribution data of the injected drug is analyzed according to the transfer boundary conditions of the positron drug targeted tissue domain. The transfer boundary conditions include the transmission efficiency of the drug through the blood vessel wall to the targeted tissue, and the ability of the targeted tissue to receive the drug. These boundary conditions are modeled by clinical imaging data and the biophysical properties of the lesions. Numerical simulation methods, such as finite difference method (FDM) or finite element analysis (FEA), are used to simulate the flow and transfer process of the drug between the blood vessel and the targeted tissue. The transfer rate of the drug in the targeted tissue area is calculated by the injection drug flow rate (flow rate: 0.5 mL / s to 2.0 mL / s). The transfer rate can be defined as the dose flow rate of the drug passing through the targeted tissue area per unit time, usually in mL / s or μg / min. For example, at a flow rate of 1.0 mL / s, the transfer rate of the drug through the targeted tissue boundary is 2.5 μg / min, and the transfer attenuation rate is A, while at 1.5 mL / s, the transfer rate increases to 3.2 μg / min, and the transfer attenuation rate is B. The absorption of drugs is evaluated by adjusting the drug flow rate and according to the physiological characteristics of the target tissue. The absorption capacity of the target tissue depends on the blood supply of the tissue, the molecular characteristics of the drug, and the permeability of the target area. The relationship between the injection drug flow rate and the absorption rate is crucial, because too fast a drug flow rate may result in the drug not being completely absorbed by the target tissue, while too slow a flow rate may result in excessive drug diffusion, reducing the efficacy, and the overall drug delivery rate in the target tissue is analyzed by the transfer decay rate. The absorption process of the drug is described using a biotransfer model (such as Fick's law) or a pharmacokinetic (PK) model. These models usually consider the absorption constant (K_a) of the tissue and the bioavailability (F) of the drug. By simulating the absorption rate of the drug at different flow rates, the correlation data between the flow rate and the absorption rate are established. Assume that when the flow rate is 1.0 mL / s, the targeted absorption rate of the drug is 75%, and when the flow rate is 1.5 mL / s, the absorption rate increases to 80%. This shows that an increase in flow rate helps to improve the absorption efficiency of the drug. Based on the injection drug flow rate-target tissue boundary transfer rate correlation data and the injection drug flow rate and target tissue boundary transfer absorption correlation data, a comprehensive analysis of the injection drug flow rate on the drug effect benefit of the target tissue is performed. The goal of this analysis is to evaluate the efficacy and side effects of the drug in the target area at different flow rates to ensure the optimal concentration distribution of the drug in the target area. Combining the flow rate-transfer rate correlation data and the flow rate-absorption correlation data, the drug effect benefit is quantified by weighted average method, linear regression or machine learning model. The evaluation indicators of drug effect benefit usually include the concentration of the drug in the target area, the therapeutic effect, the drug metabolism rate and the possible side effects.For example, at a flow rate of 1.0 mL / s, the drug's impact-benefit score was 85% (highly effective with fewer side effects), while at a flow rate of 1.5 mL / s, the drug's impact-benefit score was 78%, indicating that a higher flow rate may cause the drug to be distributed too quickly, thus affecting efficacy.

[0052] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0053] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An intelligent injection system for positron drugs, characterized in that: Includes the following modules: A global symptom monitoring feature analysis module is used to obtain the patient's physiological sign data and the patient's historical lesion metabolic imaging data; perform global symptom monitoring feature analysis through the patient's physiological sign data and the patient's historical lesion metabolic imaging data to generate global symptom monitoring feature data; The injection drug simulation distribution analysis module is used to monitor the positron drug characteristics of positron drugs using radioactive detectors and generate positron drug characteristic data; Based on the global disease monitoring characteristic data and the positron drug characteristic data, the injection drug flow diffusion simulation distribution analysis of the positron drug is performed to generate the injection drug simulation distribution data; An intelligent injection control model design module is used to analyze the injection drug simulation distribution trend characteristics according to the injection drug simulation distribution data, and generate the injection drug simulation distribution trend characteristic data; A deep reinforcement learning model for personalized injection control of positron drugs is constructed based on the simulated distribution trend characteristic data of injected drugs to obtain a personalized injection control model for positron drugs; The positron drug intelligent injection control module is used to perform targeted tissue drug impact benefit analysis of the injection drug flow rate based on the injection drug simulation distribution data, and obtain the targeted tissue drug impact benefit data of the injection drug flow rate; perform model optimization processing on the positron drug personalized injection control model based on the targeted tissue drug impact benefit data of the injection drug flow rate, and generate an optimized positron drug personalized injection control model; transmit the injection drug simulation distribution data to the optimized positron drug personalized injection control model to perform personalized positron drug injection control analysis, and generate personalized positron drug injection control parameters; and perform positron drug intelligent injection operations based on the positron drug personalized injection control parameters.

2. The intelligent injection system for positron-emitting drugs according to claim 1, characterized in that: The global symptom monitoring feature analysis module includes the following functions: Collecting and processing the patient's physiological signs in real time through a biosensor to generate the patient's physiological sign data, wherein the patient's physiological sign data includes body surface temperature data, heart rate variability data, and blood flow rate data; Obtain historical lesion metabolic imaging data of patients from the medical system; Perform lesion metabolic imaging feature analysis on the patient's historical lesion metabolic imaging data to generate lesion metabolic imaging feature data; Based on the lesion metabolic imaging feature data and the patient's physiological signs data, a global symptom monitoring feature analysis is performed to generate global symptom monitoring feature data.

3. The intelligent injection system of positron drug according to claim 2, characterized in that: The global disease monitoring feature analysis based on lesion metabolic imaging feature data and patient physiological sign data includes: Perform voxel division processing on the lesion metabolic image feature data according to the lesion metabolic image feature data to generate voxelized lesion metabolic image feature data; The lesion metabolic imaging feature voxelized data and the patient's physiological sign data are temporally and spatially aligned, and a global disease monitoring feature analysis is performed based on the temporally and spatially aligned lesion metabolic imaging feature voxelized data and the patient's physiological sign data to generate global disease monitoring feature data.

4. The intelligent injection system of positron drug according to claim 1, characterized in that: The injection drug simulation distribution analysis module includes the following functions: Using a radioactive detector to monitor the positron drug characteristics of the positron drug to generate positron drug characteristic data, wherein the positron drug characteristic data includes positron drug activity value data and positron drug decay rate data; Positron drug targeted tissue region analysis is performed through voxelized data of lesion metabolic imaging features of global disease monitoring feature data to generate positron drug targeted tissue region data; Based on the patient's physiological sign data of the global disease monitoring feature data and the positron drug characteristic data, a positron drug injection vascular domain simulation analysis is performed to generate positron drug injection vascular domain simulation data; Performing targeted tissue domain disease monitoring characteristic data analysis on the global disease monitoring characteristic data according to the positron drug targeted tissue region data to generate targeted tissue domain disease monitoring characteristic data; Perform positron drug targeted tissue domain diffusion analysis based on targeted tissue domain disease monitoring characteristic data and positron drug characteristic data to generate positron drug targeted tissue domain diffusion data; Based on the positron drug injection blood vessel domain simulation data and the positron drug targeted tissue domain diffusion data, the injection drug flow diffusion simulation distribution analysis of the positron drug is performed to generate the injection drug simulation distribution data.

5. The intelligent injection system of positron drug according to claim 4, characterized in that: The injection drug flow diffusion simulation distribution analysis based on the positron drug injection blood vessel domain simulation data and the positron drug targeted tissue domain diffusion data includes: Perform diffusion gradient analysis based on the positron drug targeted tissue domain diffusion data to generate positron drug targeted tissue domain diffusion gradient data; The positron drug targeted tissue domain transfer boundary conditions are set through the positron drug targeted tissue domain diffusion gradient data, and the positron drug targeted tissue domain transfer boundary conditions are used to perform positron drug injection simulation distribution analysis on the positron drug injection vascular domain simulation data and the positron drug targeted tissue domain diffusion data to generate injection drug simulation distribution data.

6. The intelligent injection system of positron drug according to claim 1, characterized in that: The intelligent injection control model design module includes the following functions: Perform injection drug simulation distribution trend feature analysis based on injection drug simulation distribution data to generate injection drug simulation distribution trend feature data; Perform positron drug injection state dimensional feature analysis based on injection drug simulation distribution trend feature data, generate positron drug injection state dimensional feature data, and set positron drug injection state space parameters based on the positron drug injection state dimensional feature data; Setting the positron drug injection action space parameters according to the preset positron drug injection control parameters; Design a reward function for positron injection; A deep reinforcement learning model for personalized positron drug injection control is constructed using the positron drug injection reward function, positron drug injection state space parameters, and positron drug injection action space parameters to obtain a personalized positron drug injection control model.

7. The intelligent injection system of positron drug according to claim 6, characterized in that: Designing a positron injection reward function involves: Defining the injection drug simulation distribution stage based on the injection drug simulation distribution trend characteristic data to obtain the injection drug simulation distribution stage data; According to the injection drug simulation distribution stage data, the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage are analyzed, and the positron drug injection reward function is designed based on the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage.

8. The intelligent injection system of positron drug according to claim 7, characterized in that: The positron drug injection reward function is designed based on the injection drug simulation dose characteristic data, injection drug simulation radiation characteristic data and injection drug simulation pressure data of each stage. The positron drug injection reward function is as follows: ； In the formula, It is represented by the positron drug injection control reward value, Expressed as the positron drug dose accuracy weight parameter, Represents the positron drug injection simulation dose data, Represents the target dose data of positron drug injection, It represents the maximum allowable dose deviation data of positron drug injection. Expressed as the positron drug radiation safety weight parameter, Expressed as a natural parameter, Represents the cumulative radiation data of the positron drug target area simulation, It is expressed as the radiation safety attenuation coefficient of the positron drug target area, It is expressed as the positron drug injection pressure stability weight parameter, It is expressed as the pressure fluctuation amplitude of positron drug injection.

9. The intelligent injection system of positron drug according to claim 5, characterized in that: The positron drug intelligent injection control module includes the following functions: Based on the injection drug simulation distribution data, the targeted tissue drug impact benefit analysis of the injection drug flow rate is performed to obtain the targeted tissue drug impact benefit data of the injection drug flow rate; Design an injection drug flow rate optimization strategy based on the targeted tissue drug impact benefit data of the injection drug flow rate, and optimize the personalized injection control model of the positron drug through the injection drug flow rate optimization strategy to generate an optimized personalized injection control model of the positron drug; The injection drug simulation distribution data is transmitted to the optimized positron drug personalized injection control model to perform positron drug personalized injection control analysis and generate positron drug personalized injection control parameters; The intelligent injection operation of positron drug is performed based on the personalized injection control parameters of positron drug.

10. The intelligent injection system of positron drug according to claim 9, characterized in that: The targeted tissue drug impact benefit analysis of the injection drug flow rate based on the injection drug simulation distribution data includes: Through the positron drug targeted tissue domain transfer boundary condition, the injection drug simulation distribution data is respectively analyzed for the injection drug flow rate and the targeted tissue boundary transfer rate correlation analysis and the injection drug flow rate and the targeted tissue boundary transfer absorption correlation analysis, so as to obtain the injection drug flow rate-targeted tissue boundary transfer rate correlation data and the injection drug flow rate and the targeted tissue boundary transfer absorption correlation data respectively; The targeted tissue drug impact benefit analysis of the injection drug flow rate is performed based on the injection drug flow rate-target tissue boundary transfer rate correlation data and the injection drug flow rate and targeted tissue boundary transfer absorption correlation data to obtain the targeted tissue drug impact benefit data of the injection drug flow rate.

Images