Radiation detection early warning method and system for medical equipment

Abstract

The invention provides a radiation detection early warning method and system for medical equipment, and is applied to the technical field of data processing. The method comprises the following steps: performing preprocessing and feature extraction on radiation feature data in target medical equipment information to generate radiation feature time sequence data; processing the radiation characteristic time sequence data based on a convolutional neural network to generate radiation early warning characteristic parameters; carrying out abnormal value filtering processing on the radiation characteristic time sequence data based on an adaptive threshold filtering method to generate target radiation characteristic data; processing the target radiation characteristic data to generate a radiation intensity dynamic change curve and abnormal radiation risk characteristics; processing the radiation intensity dynamic change curve based on a preset dynamic early warning rule and the abnormal radiation risk characteristics to generate early warning decision data; and processing the early warning decision data and the radiation early warning characteristic parameters, generating radiation standard exceeding early warning information and an equipment abnormal shutdown instruction, and optimizing radiation protection energy consumption management.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a radiation detection and warning method and system for medical devices. Background Art

[0002] In the medical field, ensuring the safe operation of radiation medical devices is crucial for the health of patients and medical staff. With the continuous development of medical technology, more and more radiation medical devices are widely used in the diagnosis and treatment process, such as X-ray machines, magnetic resonance imaging devices, etc. While these devices provide important support for medical diagnosis and treatment, they also bring potential radiation safety risks. If the device fails or is operated improperly, it may lead to radiation leakage and cause serious harm to the human body.

[0003] The radiation safety of medical radiation devices (such as gamma knives, CT machines) is directly related to the health of patients and medical staff, and the risk threshold settings of existing environmental monitoring technologies cannot directly adapt to the safety baseline of the medical scenario (such as the strict fluctuation range of the treatment dose rate).

[0004] In medical radiation detection, not only should we focus on the operating status of the device itself, but also pay more attention to the direct impact of radiation on human health and factors such as the compliance of the medical detection process. Medical devices have different requirements for radiation protection energy consumption at different treatment stages (such as initialization, focusing, treatment). The existing technology lacks a dynamic energy consumption management mechanism combined with risk levels, making it difficult to achieve energy conservation while ensuring safety.

[0005] It should be noted that the information disclosed in the above background art section is only used to enhance the understanding of the background of the present disclosure, and thus may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present application is to provide a radiation detection and warning method and system for medical devices, which can at least overcome the problems existing in the prior art to a certain extent. Through multi-dimensional data processing and intelligent analysis, a radiation feature database and device information are obtained, time series data is generated through preprocessing, convolutional neural networks are used to extract features and filter out outliers, dynamic curves and risk features are generated, warning decision data is generated based on rules, and finally warning information, shutdown instructions are generated and energy consumption is optimized.

[0007] Other features and advantages of the present application will become apparent through the following detailed description, or will be partially learned through the practice of the present invention.

[0008] According to one aspect of the present application, there is provided a radiation detection and early warning method for medical devices, including: obtaining a multi-dimensional medical radiation feature database and target medical device information, where the multi-dimensional medical radiation feature database includes medical device operation radiation data and environmental background radiation data; preprocessing and feature extraction of the radiation feature data in the target medical device information to generate radiation feature time series data, where the radiation feature time series data is composed of a filter noise reduction result, a spatio-temporal feature normalization relationship, and a multi-source data fusion and enhancement result; processing the radiation feature time series data based on a convolutional neural network to generate radiation early warning feature parameters; performing outlier filtering processing on the radiation feature time series data based on an adaptive threshold filtering method to generate target radiation feature data; processing the target radiation feature data to generate a radiation intensity dynamic change curve and an abnormal radiation risk feature; processing the radiation intensity dynamic change curve based on a preset dynamic early warning rule and the abnormal radiation risk feature to generate early warning decision data; processing the early warning decision data and the radiation early warning feature parameters to generate a radiation over-limit early warning message, a device abnormal shutdown instruction, and optimize radiation protection energy consumption management.

[0009] In another aspect of the present application, a radiation detection and early warning device for medical devices is characterized by including: an acquisition module for obtaining a multi-dimensional medical radiation feature database and target medical device information, where the multi-dimensional medical radiation feature database includes medical device operation radiation data and environmental background radiation data; a processing module for preprocessing and feature extraction of the radiation feature data in the target medical device information to generate radiation feature time series data, where the radiation feature time series data is composed of a filter noise reduction result, a spatio-temporal feature normalization relationship, and a multi-source data fusion and enhancement result; processing the radiation feature time series data based on a convolutional neural network to generate radiation early warning feature parameters; performing outlier filtering processing on the radiation feature time series data based on an adaptive threshold filtering method to generate target radiation feature data; processing the target radiation feature data to generate a radiation intensity dynamic change curve and an abnormal radiation risk feature; processing the radiation intensity dynamic change curve based on a preset dynamic early warning rule and the abnormal radiation risk feature to generate early warning decision data; processing the early warning decision data and the radiation early warning feature parameters to generate a radiation over-limit early warning message, a device abnormal shutdown instruction, and optimize radiation protection energy consumption management.

[0010] According to still another aspect of the present application, an electronic device is characterized by including: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the radiation detection and early warning method for medical devices described above by executing the executable instructions.

[0011] According to another aspect of the present application, there is provided a computer-readable storage medium having stored thereon a computer program, which when executed by a second processor, implements the above-mentioned radiation detection and warning method for medical devices.

[0012] The radiation detection and warning method and system for medical devices provided by the present application achieve precise warning and energy consumption optimization through multi-dimensional data processing and intelligent analysis by the server. First, a radiation feature database containing device operation and environmental background data and device information is obtained, and time series data is generated through preprocessing such as filtering and noise reduction, spatio-temporal normalization, and multi-source fusion. Then, a convolutional neural network is used to extract deep features and combined with an adaptive threshold to filter out outliers, generating a dynamic curve of radiation intensity and risk features. Based on preset rules, warning decision data is generated, and finally, warning information, shutdown instructions are generated and energy consumption is optimized to comprehensively ensure medical radiation safety and optimize energy consumption.

[0013] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The flowchart showing a radiation detection and warning method for medical devices provided by an embodiment of the present application; Figure 2 The structural schematic diagram showing a radiation detection and warning device for medical devices provided by an embodiment of the present application. DETAILED DESCRIPTION

[0015] The following describes the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention.

[0016] The following combines Figure 1 to describe the radiation detection and warning method for medical devices according to an exemplary embodiment of the present application. It should be noted that the following application scenarios are only shown for the convenience of understanding the spirit and principle of the present application, and the embodiments of the present application are not limited in this regard. On the contrary, the embodiments of the present application are applicable to any applicable scenario.

[0017] In one embodiment, the present application also proposes a radiation detection and warning method and system for medical devices. Figure 1 Schematically shown is a flowchart of a radiation detection and warning method for medical devices according to an embodiment of the present application. As Figure 1 shown, it includes: S101, obtaining a multi-dimensional medical radiation feature database and target medical device information.

[0018] In one implementation, an example of the acquisition of radiation data during the operation of a medical device is as follows. The device type is the Nanjing Taikun GK-III gamma knife (a non-CT radiotherapy device). Radiation dose rate characteristics: In the static focusing mode, the dose rate at the isocenter is 1.8 Gy / min (the standard value during treatment); in the dynamic scanning mode, the dose rate fluctuates periodically as the target moves (the fluctuation range is ±5%). Energy spectrum distribution characteristics: The gamma-ray energy spectrum of the 60Co radiation source contains characteristic peaks of 1.17 MeV and 1.33 MeV, and the peak intensity ratio is approximately 1:1.15; during the treatment process, if the collimator shifts, the intensity of the 1.17 MeV peak may decrease by more than 10%. Device operation status parameters: Radiation source switch status (ON / OFF), treatment mode (such as single target / multi-target); focusing accuracy (normal range ≤ 0.3 mm, abnormal value such as 0.8 mm), treatment duration (typical value 15 - 30 minutes).

[0019] An example of the acquisition of environmental background radiation data is as follows. The acquisition scenario is the radiotherapy room where the gamma knife device is located and the surrounding area. Natural background radiation: The natural gamma radiation background value in the room is 0.12 μSv / h (contributed by natural radioactive elements such as rubidium and potassium); the additional background caused by the floor building materials (granite) is approximately 0.05 μSv / h. Surrounding equipment interference data: When the adjacent linear accelerator is operating, the X-ray leakage at the partition wall of the room is 0.03 μSv / h (at a distance of 10 meters); when the room air conditioner is started, the temperature drift effect of the temperature and humidity fluctuations (temperature ±2°C, humidity ±5%) on the radiation detector is ±2%.

[0020] The database architecture design is as follows: An example of the acquisition of target medical device information is as follows. The basic device information is device model: GK-III gamma knife; serial number: TK-GK-20230518, rated power: 150 kV·mA. The real-time operation information is as follows, treatment task ID: RT-20250529-001, current target coordinates (x = 12 mm, y = -8 mm, z = 5 mm); radiation source usage duration: cumulative 500 hours (the life threshold is 1000 hours), remaining activity: 3000 Ci.

[0021] S102, preprocess and extract features from the radiation characteristic data in the target medical device information to generate radiation characteristic time series data.

[0022] In one implementation, the radiation feature data in the target medical device information is preprocessed based on the multi-dimensional medical radiation feature database to generate a number of data subsets. Taking the γ knife treatment cycle (20 minutes) as a unit, the dose rate data is divided into 24,000 data subsets at a time interval of 50 ms, and the energy spectrum data and device status at the corresponding moment are extracted simultaneously. Subset 1: From 0 to 10 s after the treatment starts, the dose rate rises from 0 to 1.8 Gy / min, and the corresponding energy spectrum peak intensity is 1.17 MeV (4.2×10 4 cps), and the device status is "radioactive source turned on, static focusing mode"; Subset 2: At the 5th minute of the treatment, the dose rate is stable at 1.82 Gy / min, the energy spectrum peak intensity ratio is 1:1.15, and the device focusing accuracy is 0.2 mm (normal).

[0023] Based on the filter denoising algorithm, a number of data subsets are processed to generate denoised data. Wavelet transform (db4 wavelet, 3-layer decomposition) is used to denoise the dose rate data, and the detector electronic noise (high-frequency interference) and power supply ripple (50 Hz low-frequency interference) are removed. Comparing before and after denoising, for the original data: at the 10th minute of the treatment, the dose rate shows high-frequency fluctuations (1.82±0.15 Gy / min), and the actual value should be a stable value of 1.8 Gy / min; for the denoised data: the fluctuation range is narrowed to ±0.03 Gy / min, and the true dose rate changes (such as slow fluctuations caused by patient movement) are retained.

[0024] The denoised data is processed to generate spatio-temporal feature normalized data. The radiation data at different target positions is mapped to the isocenter coordinate system (the origin is the target center). For example: the dose rate of 1.75 Gy / min at the target coordinate (12, -8, 5) mm is normalized to 1.8 Gy / min after being mapped to the isocenter (considering the inverse square attenuation correction of distance); Time normalization: The treatment cycle (20 minutes) is linearly mapped to the [0, 1] interval to facilitate the comparison of data with different treatment durations (such as aligning the temporal characteristics of 15-minute treatment and 20-minute treatment).

[0025] The spatio-temporal feature normalized data is processed to generate multi-source fusion data. The weighted fusion formula F = 0.6D + 0.3S + 0.1E is used, where D is the dose rate, S is the device status score (0 - 1), and E is the environmental correction factor.

[0026] The specific data fusion dimensions are as follows: The dose rate, device status, and environmental background are fused according to weights to avoid misjudgment of a single data dimension (such as fluctuations in the environmental background do not affect the accurate extraction of the radiation characteristics of the device operation).

[0027] Process multi-source fusion data to generate temporal radiation feature data. Temporal data structure: Indexed by timestamp, store the fused multi-dimensional features. The following is an example snippet: In Python format, [{"timestamp":"2025-05-29 09:30:00","normalized_dose_rate":1.80, # Normalized dose rate (Gy / min)"spectrum_offset":0.02, # Energy spectrum offset index (deviation from the standard peak intensity ratio)"equipment_status":"static_focus","environment_factor":0.98 # Environmental correction factor}, {"timestamp":"2025-05-29 09:30:01","normalized_dose_rate":1.81,"spectrum_offset":0.01,"equipment_status":"static_focus","environment_factor":0.98}; # Data for subsequent time points...

[0028] Verify the temporal features. The fused temporal data can clearly reflect the evolution of radiation features during the entire treatment cycle. For example, the dose rate fluctuation in the static focus stage is ≤ ±1%, which is consistent with the normal operation baseline of the equipment and there is no conflict.

[0029] S103. Process the temporal radiation feature data based on a convolutional neural network to generate radiation warning feature parameters.

[0030] In one implementation, extract features from the temporal radiation feature data based on a convolutional neural network to generate a set of multi-dimensional feature vectors; the network architecture uses a 1D-CNN model, which includes 3 convolutional layers (kernel_size = 5, stride = 1) and 2 fully connected layers, and is used to extract the deep features of the temporal data of the gamma knife radiation. The following is an example of feature extraction: Input data: The fused temporal radiation feature data (including 5-dimensional features such as dose rate, energy spectrum offset, and equipment status, with a time step of 50 ms and a total of 2000 time points); Output of the convolutional layer: Extracted local features such as "dose rate mutation feature" (the rising slope of the dose rate is 0.08 Gy / min² from the 100th to 150th time steps) and "energy spectrum stability feature" (the peak intensity fluctuation range of 1.17 MeV is ±3%). Taking the Python format as an example, the multi-dimensional feature vector: # Feature vector extracted at a certain time step (dimension = 128) [0.72, 0.15, 0.31, 0.89,..., 0.45] # Includes features such as dose rate change, energy spectrum offset, and equipment status association.

[0031] By comparing the sets of feature vectors at adjacent time steps, the evolution law of radiation features over time is determined, and the information on the dynamic change trajectory of features is generated. The comparison of temporal features is as follows: calculate the Euclidean distance of the feature vectors at adjacent time steps (such as t and t + 1) to determine the change rate of radiation features. An example of generating the evolution trajectory is as follows: during the normal treatment stage (0 - 10 minutes), the distance of the dose rate feature vector from the mean is 0.12, and the distance of the energy spectrum feature vector from the mean is 0.08, and the trajectory shows a stable trend; during the abnormal stage (the 15th minute), the distance of the dose rate feature vector suddenly increases to 0.45 (standard value ≤ 0.2), and the distance of the energy spectrum feature vector is 0.32, and the trajectory shows an obvious mutation.

[0032] The visualized trajectory segments are as follows: The information on the dynamic change trajectory of features is compared with the preset normal radiation pattern library and abnormal radiation pattern library to generate the evaluation result of feature deviation degree. The normal pattern library stores the feature trajectories with a dose rate of 1.8 ± 0.05 Gy / min and an energy spectrum peak intensity ratio of 1:1.15 ± 5% under the static focusing mode of the gamma knife; the abnormal pattern library includes 6 abnormal patterns such as collimator offset (energy spectrum peak intensity decrease > 10%) and radiation source attenuation (dose rate continuously decreasing > 5%).

[0033] An example of deviation degree calculation is as follows: scenario: the matching degree of the feature trajectory at the 15th minute with the "collimator offset" pattern reaches 85%, and the matching degree with the normal pattern is only 30%; the evaluation result of deviation degree: the dose abnormal deviation degree = 0.75, and the energy spectrum abnormal deviation degree = 0.82 (0 is normal, 1 is severely abnormal).

[0034] Combined with the device operation status corresponding to the temporal data of radiation features and the spatio - temporal context information of environmental background radiation, the evaluation result of feature deviation degree is weighted to generate the radiation risk feature vector; the weighting processing logic is as follows: weight assignment: the weight of device operation status (focusing accuracy) is 0.4, the weight of environmental background (machine room interference) is 0.2, and the weight of feature deviation degree is 0.4. The focusing accuracy is 0.8 mm (abnormal, full score 1 point corresponding to 0.3 mm) and gets 0.3 points, and the environmental interference coefficient is 0.95 (low interference, full score 1 point) and gets 0.95 points. Risk feature vector calculation: risk feature vector = 0.4×[0.75, 0.82]+0.2×[0.3, 0.95]+0.4×[0.75, 0.82]=[0.63, 0.80] # corresponding to dose risk and energy spectrum risk respectively.

[0035] Normalize and comprehensively operate on the feature deviation degree evaluation results and radiation risk feature vectors to generate a set of radiation warning feature parameters including dose anomaly coefficients and energy spectrum offset indices. Map the deviation degree and risk feature vectors to the interval [0, 1]. For example, the original dose anomaly deviation degree of 0.75 → 0.75 after normalization; the original energy spectrum risk feature of 0.80 → 0.80 after normalization. The comprehensive operation formula is: dose anomaly coefficient = 0.6 × deviation degree + 0.4 × risk feature = 0.6 × 0.75 + 0.4 × 0.63 = 0.702; energy spectrum offset index = 0.7 × deviation degree + 0.3 × risk feature = 0.7 × 0.82 + 0.3 × 0.80 = 0.814.

[0036] Taking the warning parameter set in python format as an example: {"dose_anomaly_coeff":0.70, # Dose anomaly coefficient (after normalization) "spectrum_offset_index":0.81, # Energy spectrum offset index (after normalization) "risk_level":"medium" # Risk level}.

[0037] S104, perform outlier filtering on the radiation feature time series data based on the adaptive threshold filtering method to generate target radiation feature data.

[0038] In one implementation, perform preliminary outlier screening on the radiation feature time series data based on the adaptive threshold filtering method to generate a preliminary screening data set. Dynamically calculate the threshold based on historical treatment data (such as the dose rate data of the previous 100 gamma knife treatments). The formula is: threshold = mean + k × standard deviation, where k = 3 (covering 99.7% of the normal data under the Gaussian distribution). The normal dose rate mean = 1.8 Gy / min, the standard deviation = 0.05 Gy / min, and the dynamic threshold = 1.8 + 3 × 0.05 = 1.95 Gy / min; the preliminary screening rule is as follows: exclude data points with a dose rate > 1.95 Gy / min or < 1.65 Gy / min, and retain the dose rate of 1.82 Gy / min (normal) at the 15th minute of treatment, and exclude 2.5 Gy / min (abnormal) caused by detector failure.

[0039] Process the continuous abnormal data segments in the preliminary screening data set to generate a primary optimized data set. Determine the data segments that exceed the threshold for 5 consecutive time steps (250 ms) as "continuously abnormal". The processing example is as follows: abnormal segment: at the 5th minute of treatment, there are 10 consecutive time steps with a dose rate = 1.98 Gy / min (exceeding the threshold of 1.95 Gy / min), but it is actually a transient fluctuation caused by the device's focusing fine-tuning; optimization processing: use the moving window averaging method (window size = 10) to correct the abnormal segment to 1.85 Gy / min, removing the spike noise while retaining the fluctuation trend.

[0040] Process the environmental interference outliers in the initially optimized data set to generate a secondarily optimized data set. Identify interference anomalies by combining the environmental background data (such as the temperature drift effect of ±2% when the air conditioner starts) and the equipment status (such as the operation of adjacent linear accelerators). The example scenario is as follows: environmental interference outlier: when the air conditioner starts, the temperature drift of the detector causes the dose rate to falsely increase to 1.84 Gy / min (the normal fluctuation should be ≤1.82 Gy / min); correct the data according to the environmental interference coefficient (temperature drift effect of ±2%), and adjust 1.84 Gy / min to 1.84×0.98 = 1.80 Gy / min.

[0041] Perform time-series feature grouping processing on the secondarily optimized data set to generate a radiation feature sequence and an intermediate data set; the grouping strategy is as follows: divide the data segments according to the treatment stage (such as initialization, focusing, treatment, cooling), and extract features for each stage: initialization stage (0 - 2 minutes): the dose rate rises from 0 to 1.8 Gy / min, extract the rising slope; treatment stage (2 - 18 minutes): the stable segment of the dose rate, extract the mean and variance.

[0042] Taking the python format as an example for the feature sequence: [{"stage":"initialization","dose_rate_slope":0.15 Gy / min², # rising slope "normalized_dose":1.80 Gy / min},{"stage":"treatment","dose_rate_mean":1.81 Gy / min,"spectrum_stability":0.97 # energy spectrum stability index}].

[0043] Adjust the outlier filtering confidence based on the radiation feature sequence to generate the target radiation feature data. Dynamically adjust the filtering threshold according to the stability of the feature sequence. The formula is: adjusted threshold = base threshold × (1 ± confidence coefficient), where the confidence coefficient is determined by the variance of the feature sequence (the smaller the variance, the higher the confidence, and the stricter the threshold).

[0044] The dose rate variance in the treatment stage = 0.01 (stable), the confidence coefficient = -0.1 (increase the threshold strictness); adjusted threshold = 1.95×(1 - 0.1) = 1.755 Gy / min (original threshold 1.95 Gy / min). At this time, the dose rate of 1.82 Gy / min is retained, and the data below 1.755 Gy / min is filtered; Taking the python format as an example for the output of the target data: [{"timestamp":"2025-05-29 09:30:00","dose_rate":1.80Gy / min,"is_valid":1},{"timestamp":"2025-05-29 09:30:01","dose_rate":1.81Gy / min,"is_valid":1},# Filtered outlier data segment].

[0045] S105, process the target radiation characteristic data to generate a dynamic change curve of radiation intensity and abnormal radiation risk characteristics.

[0046] In one implementation, perform a time series trend analysis on the target radiation characteristic data to generate time series parameters of radiation intensity; use a sliding window (window size = 100 time steps, 5 seconds) to calculate the mean, variance, and rate of change of the dose rate. The specific parameters are as follows: the mean dose rate within the 10th minute of treatment in the window = 1.81 Gy / min, variance = 0.01, rising slope = 0.002 Gy / min²; the set of time series parameters: [1.81, 0.01, 0.002], corresponding to the mean, stability, and change trend respectively.

[0047] Obtain the energy spectrum distribution data of the target radiation characteristic data and the device operating condition information. Example of energy spectrum data is as follows: for the 60Co source, the peak intensity of the 1.17 MeV peak = 4.2×10 4 cps, the peak intensity of the 1.33 MeV peak = 4.9×10 4 cps, and the peak intensity ratio = 1:1.17 (standard value 1:1.15).

[0048] Example of operating condition information is as follows: device operating mode: static focusing, focusing accuracy = 0.8 mm (abnormal, normal ≤ 0.3 mm), treatment duration = 15 minutes (5 minutes remaining).

[0049] Process the energy spectrum distribution data to generate radiation energy characteristic parameters. The logic for feature extraction is that the peak intensity offset = |actual peak intensity ratio - standard peak intensity ratio| / standard peak intensity ratio; energy spectrum stability = 1 - (range of peak intensity fluctuation / average peak intensity). Specifically, the peak intensity ratio offset = |1.17 - 1.15| / 1.15 = 0.017; energy spectrum stability = 1 - (0.05×10 4 cps / 4.55×10 4 cps) = 0.89; energy characteristic parameters: [0.017, 0.89], corresponding to the offset and stability respectively.

[0050] Process the operating condition information of the equipment, statistically analyze the radiation characteristic distribution under different operating conditions, and generate the correlation baseline value between radiation and operating conditions, including the radiation dose rate fluctuation range during normal operation of the equipment and the characteristic deviation threshold under abnormal operating conditions. Normal condition baseline: In the static focusing mode, the dose rate fluctuation range = 1.8 ± 0.05 Gy / min, and the energy spectrum peak intensity ratio = 1.15 ± 0.05. Abnormal condition threshold: When the focusing accuracy > 0.5 mm, the allowable dose rate fluctuation range expands to 1.8 ± 0.1 Gy / min, and the peak intensity ratio deviation threshold = 0.1 (values exceeding this are determined as abnormal).

[0051] The correlation baseline is shown in the following Python format: {"normal_dose_range":[1.75,1.85],# Normal fluctuation range (Gy / min)"abnormal_spectrum_threshold":0.1,# Energy spectrum abnormality threshold"focus_accuracy_effect":{"0.3 - 0.5mm":{"dose_range":[1.7,1.9]},# Dose fluctuation range corresponding to different focusing accuracies">0.5mm":{"dose_range":[1.6,2.0]}}}。

[0052] Process the radiation intensity time series parameters, radiation energy characteristic parameters, and the correlation baseline value between radiation and operating conditions to generate the dynamic change curve of radiation intensity and the abnormal radiation risk characteristics. The dynamic curve is plotted with the x-axis as time (0 - 20 minutes) and the y-axis as the dose rate (Gy / min), marking the normal baseline (1.8 ± 0.05) and the abnormal threshold (1.8 ± 0.1 when the focusing accuracy is 0.8 mm). At the 15th minute, the dose rate = 1.92 Gy / min, which exceeds the normal baseline but does not exceed the abnormal threshold, and the curve is shown as the "yellow warning range".

[0053] The risk characteristics are generated as follows: Dose rate overrun = (1.92 - 1.85) / 0.05 = 1.4 (140% overrun within the normal baseline); Energy spectrum risk = Peak intensity offset 0.017 × Energy spectrum stability weight 0.6 + Focusing accuracy weight 0.4 × (0.8 - 0.3) / (1 - 0.3) = 0.017 × 0.6 + 0.4 × 0.71 = 0.31; Abnormal risk characteristic set: {"dose_overrun":1.4,"spectrum_risk":0.31,"risk_level":"medium"}。

[0054] S106, process the dynamic change curve of radiation intensity based on the preset dynamic warning rules and abnormal radiation risk characteristics to generate warning decision data.

[0055] In one implementation, a hierarchical parsing process is performed on the preset dynamic warning rules to generate warning rule weight coefficients. Example of rule levels: First-level rule: The dose rate continuously exceeds the standard by more than 10% and for more than 5 minutes (weight 0.5); Second-level rule: The energy spectrum peak intensity ratio deviates by more than 5% and the focusing accuracy > 0.5 mm (weight 0.3); Third-level rule: The dose rate rising slope > 0.05 Gy / min² (weight 0.2).

[0056] Taking the python format as an example for generating the weight coefficients: {"rule_weights":{"level1":0.5,"level2":0.3,"level3":0.2}}.

[0057] Obtain the abnormal radiation risk characteristic data and the real-time data of the dynamic change curve of the radiation intensity. Abnormal risk characteristic data: The dose rate exceeding standard amplitude = 1.4 (exceeding the standard by 140% within the normal baseline), the energy spectrum risk = 0.31, the risk level = medium. Real-time data of the dynamic curve: The dose rate at the 15th minute = 1.92 Gy / min, the dose rate rising slope within the past 5 minutes = 0.03 Gy / min², the peak exceeding standard times = 3 times (exceeding the normal baseline by 1.85 Gy / min).

[0058] Process the abnormal radiation risk characteristic data to generate risk characteristic quantization parameters. The quantization logic is as follows: Risk level mapping: medium → quantization value 0.6 (0 is low, 1 is high); Comprehensive risk parameter = dose exceeding standard amplitude × 0.6 + energy spectrum risk × 0.4 = 1.4 × 0.6 + 0.31 × 0.4 = 0.964.

[0059] Process the real-time data of the dynamic change curve of the radiation intensity, count the curve morphological characteristics and the threshold breakthrough frequency, and generate the curve characteristic baseline values, including the dose rate rising slope and the peak exceeding standard times. The morphological characteristics statistics are as follows: The dose rate rising slope = 0.03 Gy / min² (normal ≤ 0.02 Gy / min²); The peak exceeding standard frequency = 3 times / 5 minutes (allowable threshold = 1 time / 5 minutes); Taking the python format as an example for the baseline value set, {"slope_anomaly":0.03, # actual slope; "peak_overrun_count":3, # exceeding standard times; "normal_slope_threshold":0.02, # normal threshold; "allowable_overrun":1 # allowable frequency}.

[0060] Perform weighted fusion processing on the warning rule weight coefficient, risk feature quantization parameter, and curve feature baseline value to generate warning decision data. The fusion formula is: Warning decision value = Rule weight × (Risk quantization parameter × 0.6 + Curve feature deviation × 0.4); Curve feature deviation = (Slope deviation + Frequency deviation) / 2 = [(0.03 - 0.02) / 0.02 + (3 - 1) / 1] / 2 = (0.5 + 2) / 2 = 1.25.

[0061] Contribution of the first-level rule = 0.5 × (0.964 × 0.6 + 1.25 × 0.4) = 0.5 × (0.578 + 0.5) = 0.539; Contribution of the second-level rule = 0.3 × (0.964 × 0.6 + 1.25 × 0.4) = 0.3 × 1.078 = 0.323; Contribution of the third-level rule = 0.2 × (0.964 × 0.6 + 1.25 × 0.4) = 0.2 × 1.078 = 0.216; Taking the python format as an example for the warning decision data: {"warning_decision_value": 0.539 + 0.323 + 0.216 = 1.078, "risk_level": "high", # The decision value > 1.0 triggers a high risk "trigger_rules": ["level1", "level2", "level3"], "recommendation": "Immediately reduce the radiation power and check the focusing system"}.

[0062] The decision value of 1.078 triggers the "high" risk level, which matches the preset threshold (1.0), and the recommended measures (reduce power + check focus) are for the specific problem of abnormal focusing accuracy (0.8 mm) of the gamma knife.

[0063] S107. Process the warning decision data and radiation warning feature parameters to generate a radiation over-limit warning message, an equipment abnormal shutdown instruction, and optimize the radiation protection energy consumption management.

[0064] In one implementation, risk level analysis and processing are performed on the early warning decision data to generate a risk level assessment result. In the radiation monitoring and early warning system, the generated early warning decision data is an important basis for subsequent risk level judgment. The early warning decision data given here is a structured information set: "warning_decision_value": 1.078: This is a comprehensively calculated decision value, which is obtained by weighted fusion calculation of various radiation-related parameters, such as dose rate, energy spectrum characteristics, equipment operation status, etc., according to preset algorithms and rules. This value reflects the comprehensive degree to which the current radiation situation deviates from the normal state. "risk_level": "high": This is the preliminary risk level judgment result, marked as "high" (high risk) here, but further verification and analysis are required in combination with other conditions. "trigger_rules": ["level1", "level2", "level3"]: This is a list that records the set of rules triggering the current early warning decision. Here, it indicates that the rules triggering this early warning come from three different levels, corresponding to different types of radiation anomalies.

[0065] The decision value 1.078 is compared with the preset high-risk threshold of 1.0. The high-risk threshold is a key boundary value determined based on radiation safety standards, the normal operation parameter range of equipment, and long-term monitoring data. When the calculated decision value is greater than this threshold, it means that the current radiation situation has reached a dangerous level that requires high vigilance, so the highest risk level "high" is triggered. This indicates that there are relatively serious anomalies in the radiation state, which may pose a greater threat to personnel safety and the normal operation of equipment.

[0066] The triggering rules are as follows. First-level rule: Continuous dose rate exceeding the standard: The dose rate is an important indicator for measuring radiation intensity. In normal radiation therapy or monitoring scenarios, the dose rate should be maintained within a relatively stable range that meets safety standards. When the dose rate continuously exceeds this range, the first-level rule is triggered. This may be due to abnormal changes in the radiation source, equipment control system failures, etc., which will directly affect the effect and safety of radiation therapy, so it is listed as one of the most important triggering rules.

[0067] Second-level rule: Energy spectrum shift and focusing accuracy anomaly: The energy spectrum reflects the energy distribution of radiation, and the focusing accuracy is related to the precise delivery of radiation energy. When the energy spectrum shifts, it may mean that the state of the radiation source has changed or problems have occurred in components such as filters and collimators inside the equipment; while an abnormal focusing accuracy will cause the radiation energy not to act accurately on the target area, which may cause unnecessary radiation damage to surrounding normal tissues. The simultaneous occurrence of anomalies in both triggers the second-level rule, and its severity is second only to the continuous dose rate exceeding the standard.

[0068] Tertiary rule: Abnormal slope of dose rate increase: The slope of dose rate increase reflects the rate of change of dose rate over time. Under normal circumstances, the change of dose rate should be stable and controllable. If the slope of dose rate increase is abnormal, that is, the rate of change is too fast, it may indicate that the equipment is about to malfunction or there are signs of instability in the radiation source. Although it is less severe compared to the previous two cases when considered alone, it is still an abnormal situation that cannot be ignored. Therefore, it is involved in the warning trigger judgment as a tertiary rule.

[0069] Based on the radiation warning characteristic parameters, abnormal feature comparison and processing are carried out to generate the equipment abnormality determination result. The generated radiation warning characteristic parameters are the key basis for abnormality determination. The content of this parameter set is as follows: "dose_anomaly_coeff": 0.70: Dose anomaly coefficient, which is a quantitative index obtained through a series of calculation and processing of data related to radiation dose rate. This coefficient is used to measure the deviation degree of the current dose rate from the dose rate in the normal state. "spectrum_offset_index": 0.81: Energy spectrum offset index, which reflects the difference degree between the actual distribution of radiation energy spectrum and the normal standard energy spectrum distribution. The energy spectrum refers to the distribution of radiation energy in different energy segments. This index is of great significance for judging whether the working state of the radiation source and related equipment is normal. "risk_level": "medium": The initially given risk level is "medium" (medium risk), but this is only a preliminary assessment based on partial parameters. Further analysis in combination with other information is required to determine the final abnormality and risk level.

[0070] The dose anomaly coefficient 0.70 is compared with the normal threshold (≤0.5). The normal threshold is determined comprehensively based on a large amount of historical data, equipment standard operating parameters, and radiation safety specifications, etc. When the dose anomaly coefficient exceeds 0.5, it indicates that the current dose rate significantly deviates from the normal range. Abnormal dose rate may cause poor treatment effects on patients receiving radiation therapy or bring unnecessary radiation hazards to the surrounding environment and personnel. Association analysis of energy spectrum and focusing accuracy shows that the energy spectrum offset index 0.81 far exceeds the normal standard (≤0.3), which indicates that the radiation energy spectrum has shifted to a large extent.

[0071] At the same time, combined with the focusing accuracy of 0.8mm (in an abnormal state, the focusing accuracy should meet specific standards under normal circumstances, and 0.8mm here is beyond the normal range), it is comprehensively judged that the energy spectrum abnormality is caused by the collimator offset. The function of the collimator is to control and adjust the shape and direction of the radiation beam. When it is offset, the energy distribution of the radiation beam will change, which will lead to the energy spectrum offset. Based on the above analysis, the final result of the equipment abnormality is {"anomaly_type":"collimator_offset","severity":"high"}. That is, the abnormality type is collimator offset, and the severity is "high" (high risk). This is determined after comprehensively considering multiple factors such as dose abnormality, energy spectrum offset, and focusing accuracy. It shows that there are more serious problems with the current equipment and they need to be dealt with in time to avoid greater impact on the safety and effectiveness of radiation therapy.

[0072] The risk level assessment results and equipment abnormality determination results are processed to generate radiation excess warning information. Risk level: determined to be high risk (high), which is based on the previous analysis of the warning decision data. The high risk level means that there is a serious risk in the current radiation situation, which may pose a major threat to personnel safety and the normal operation of equipment, and requires immediate attention and corresponding measures. Abnormality type: determined to be collimator offset. The collimator plays a key role in controlling the shape and direction of the radiation beam in the radiation equipment. When the collimator is offset, it will cause abnormal energy distribution of the radiation beam, which in turn affects parameters such as dose rate and energy spectrum. This is the root cause of the current radiation abnormality.

[0073] The specific parameters are as follows: dose rate: 1.92Gy / min, 140% above the normal baseline. The normal baseline range is [1.75,1.85]Gy / min. The dose rate significantly exceeds the normal range, which directly reflects the abnormal increase in radiation intensity, which may cause excessive radiation to patients receiving radiation therapy and endanger their health. Energy spectrum peak intensity ratio deviation: 1.7%. Energy spectrum peak intensity ratio is an important indicator to measure the characteristics of radiation energy spectrum. Its deviation indicates that the energy distribution of radiation has changed, which is related to the change of radiation beam caused by collimator deviation. Focusing accuracy: 0.8mm. Focusing accuracy reflects whether the radiation can accurately act on the target area. The focusing accuracy of 0.8mm is in an abnormal state, which will cause the radiation to be unable to be accurately delivered, which will not only affect the treatment effect, but also may cause unnecessary radiation damage to surrounding normal tissues.

[0074] The warning information given is presented in JSON format, with a clear structure, facilitating system recognition and personnel understanding. warning_message: "

High-risk Warning

[0075] Obtain real-time information on the operating status of the equipment and energy consumption configuration information of the radiation protection system. The radiation source is currently in the "ON (during treatment)" state, indicating that the equipment is performing radiation therapy-related operations. The radiation source is the core component of the radiation equipment that generates radiation, and its status directly determines whether the equipment is in a working state. The remaining activity is 3000 Ci (Curie, the unit of radioactive activity). The activity of the radiation source decays over time, and the remaining activity reflects the current radioactive intensity of the radiation source. This parameter is of great significance for evaluating the service life of the radiation source, the stability of the treatment dose, etc. For example, if the remaining activity is low, it may be necessary to consider replacing the radiation source to ensure the consistency and stability of the treatment effect.

[0076] The protective shielding device, with a current power of 80% (full power is 100%), indicates that the protective shielding device is not operating at its maximum power. The function of the protective shielding device is to block and attenuate radiation, protecting the surrounding personnel from unnecessary radiation damage. The power level determines the strength of the shielding effect. The energy consumption is 20 kW, which is the power consumed by the current operation of the protective shielding device. Understanding the energy consumption situation helps to evaluate the operating cost of the equipment and the energy utilization efficiency. At the same time, the energy consumption data can also be used as a reference index to judge whether the protective shielding device is operating normally. If the energy consumption abnormally increases or decreases, it may mean that there are problems such as equipment failures or unreasonable operating parameter settings.

[0077] The energy consumption configuration information presented in JSON format contains the following key elements: shielding_power: 80: This represents that the current power setting of the protective shielding device is 80 (corresponding to the 80% power mentioned earlier). This is a relative value used to reflect the operating intensity of the shielding device. normal_power: 50: Represents the power setting when the protective shielding device is operating normally, which is 50 here. This value is the standard power value set after comprehensively considering factors such as radiation protection requirements and energy consumption under normal operating conditions of the equipment. shutdown_threshold: 90: Is the shutdown threshold of the protective shielding device. When certain relevant parameters (such as abnormal increase in radiation intensity, equipment failure, etc.) reach or exceed this threshold (90 here), the system may trigger a shutdown operation to ensure safety. energy_saving_mode: true: Indicates that the equipment is currently in the energy-saving mode. In the energy-saving mode, the equipment will reduce energy consumption and achieve the energy-saving goal by optimizing operating parameters and other means while meeting the basic radiation protection and equipment operation requirements. These energy consumption configuration information play an important guiding role in reasonably managing the energy consumption of the equipment, ensuring the safe and stable operation of the equipment, and adjusting parameters under different operating conditions.

[0078] Process the energy consumption configuration information based on the device operating status and radiation over - standard warning information, generate a device abnormal shutdown instruction, and optimize the radiation protection energy consumption management. Shutdown instruction generation: The condition trigger is when the risk level is determined to be "high" (high - risk), and the abnormal type is clearly a mechanical failure (here it is the collimator offset). When this occurs, the shutdown instruction generation mechanism is triggered. A high - risk level means that the current radiation abnormality is serious and may cause significant harm to personnel and equipment; while the collimator offset, as a mechanical failure, will cause the direction and energy distribution of the radiation beam to get out of control, and continued operation may lead to more serious consequences. Therefore, when these two conditions are met, shutdown processing is required. The generated instruction is presented in JSON format as {"shutdown_command":true,"delay_seconds":10,"reason":"collimator_offset_high_risk"}, indicating an automatic shutdown after 10 seconds. The 10 - second delay is considered comprehensively for multiple reasons: On the one hand, it gives on - site medical staff a certain amount of time for emergency handling, such as evacuating irrelevant personnel in the vicinity and making temporary arrangements for patients; on the other hand, it also leaves time for the system to perform some pre - shutdown preparation work, such as recording the current operating data and saving the device status. The reason is clearly marked in the instruction as "collimator offset high - risk", which facilitates subsequent maintenance personnel to quickly understand the shutdown background and fault type, improving the efficiency of fault troubleshooting and repair.

[0079] The energy consumption optimization logic is as follows: After being determined to be in a high - risk state, in order to prioritize safety and prevent radiation leakage from harming surrounding personnel, the power of the shielding device is increased from the current 80% to 100%, and the corresponding energy consumption increases from 20kW to 25kW. This is because in the case of radiation abnormality and high risk, it is necessary to enhance the shielding effect to ensure that the radiation is effectively blocked within the safe range. Although the energy consumption increases, it can minimize the risk of radiation leakage and ensure the safety of personnel and the environment.

[0080] After the device shuts down, it enters the standby mode. At this time, the energy consumption drops to the basic monitoring power of 5kW, which is only 25% of the energy consumption during normal operation. In the standby mode, the device retains basic monitoring functions to monitor the device status in real - time while significantly reducing energy consumption. This can not only respond in a timely manner when the device has an abnormality but also save energy in the non - working state, achieving the optimization of energy utilization.

[0081] The optimized energy consumption configuration is also presented in JSON format: shielding_power: 100: This indicates that the power of the shielding device has been increased to 100%, operating at the maximum power to provide the strongest radiation shielding ability. current_energy: 25: The current energy consumption is 25 kW, corresponding to the energy consumption level of the shielding device at 100% power. shutdown_status: "scheduled": This indicates that the device is in a scheduled shutdown state, i.e., shutting down according to the preset shutdown instructions and procedures, differentiating from an emergency shutdown caused by a sudden failure. standby_power: 5: The standby power is set to 5 kW, which is the energy consumption standard of the device in the standby mode, used to maintain basic monitoring and status keeping functions. These configuration information accurately reflect the energy consumption adjustment strategy and current status of the device when dealing with radiation anomalies, providing clear data basis for the operation management and energy consumption monitoring of the device.

[0082] In this application, the server realizes the precise early warning and energy consumption optimization of the radiation safety of medical devices through multi-dimensional data processing and intelligent analysis. First, it obtains a multi-dimensional medical radiation feature database (including device operation radiation data and environmental background radiation data) and target medical device information, laying a data foundation for subsequent analysis. Then, it preprocesses and extracts features from the radiation feature data. Through filtering and noise reduction, spatio-temporal normalization, and multi-source fusion, it generates radiation feature time series data to ensure the accuracy and availability of the data.

[0083] Then, it uses a convolutional neural network to extract deep features from the time series data, generating radiation early warning feature parameters. At the same time, it filters out outliers through an adaptive threshold to obtain target radiation feature data, providing high-quality data support for subsequent analysis. On this basis, it generates a dynamic change curve of radiation intensity and abnormal radiation risk features, intuitively showing the change trend of radiation intensity and potential risks.

[0084] Furthermore, it generates early warning decision data based on preset dynamic early warning rules and abnormal radiation risk features to realize the intelligent evaluation and early warning of radiation risks. Finally, according to the early warning decision data and radiation early warning feature parameters, it generates radiation exceeding the standard early warning information, device abnormal shutdown instructions, and optimizes the radiation protection energy consumption management, achieving reasonable optimization of energy consumption while ensuring safety. It realizes the comprehensive and precise monitoring and early warning of the radiation safety of medical devices, ensuring radiation safety during medical procedures and optimizing energy consumption management at the same time.

[0085] In one implementation, as Figure 2 shown, this application also provides a radiation detection and early warning device for medical devices, including: An acquisition module 301, configured to acquire a multi-dimensional medical radiation feature database and target medical device information, where the multi-dimensional medical radiation feature database includes medical device operation radiation data and environmental background radiation data; A processing module 302, configured to preprocess and extract feature data from the radiation feature data in the target medical device information to generate radiation feature time series data, where the radiation feature time series data is composed of a filtering and noise reduction result, a spatio-temporal feature normalization relationship, and a multi-source data fusion and enhancement result; process the radiation feature time series data based on a convolutional neural network to generate radiation warning feature parameters; perform outlier filtering on the radiation feature time series data based on an adaptive threshold filtering method to generate target radiation feature data; process the target radiation feature data to generate a radiation intensity dynamic change curve and an abnormal radiation risk feature; process the radiation intensity dynamic change curve based on a preset dynamic warning rule and the abnormal radiation risk feature to generate warning decision data; process the warning decision data and the radiation warning feature parameters to generate a radiation exceeding the standard warning message, a device abnormal shutdown instruction, and optimize radiation protection energy consumption management.

[0086] The computer-readable storage medium provided by the above embodiments of the present application and the radiation detection and warning method for medical devices provided by the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run, or implemented by the application programs stored therein.

[0087] Each embodiment in the present application is described in a related manner. The same or similar parts among the embodiments can be referred to each other, and the differences between each embodiment and other embodiments are emphasized. In particular, for the embodiments of the method, electronic device, electronic equipment, and readable storage medium for evaluating the radiation detection and warning method for medical devices, since they are basically similar to the embodiments of the radiation detection and warning method for medical devices described above, the description is relatively simple, and the relevant parts can be referred to the partial description of the embodiments of the radiation detection and warning method for medical devices described above.

Claims

1. A radiation detection and warning method for medical devices, characterized in that, Including: Obtain a multi-dimensional medical radiation feature database and target medical device information, where the multi-dimensional medical radiation feature database includes medical device operation radiation data and environmental background radiation data; Preprocess and extract features from the radiation feature data in the target medical device information to generate radiation feature time series data, where the radiation feature time series data consists of filter denoising results, spatio-temporal feature normalization relationships, multi-source data fusion and enhancement results; Process the radiation feature time series data based on a convolutional neural network to generate radiation warning feature parameters; Perform outlier filtering on the radiation feature time series data based on an adaptive threshold filtering method to generate target radiation feature data; Process the target radiation feature data to generate a radiation intensity dynamic change curve and abnormal radiation risk features; Process the radiation intensity dynamic change curve based on preset dynamic warning rules and abnormal radiation risk features to generate warning decision data; Process the warning decision data and radiation warning feature parameters to generate radiation over-standard warning information, device abnormal shutdown instructions and optimize radiation protection energy consumption management.

2. The method according to claim 1, wherein, Preprocess and extract features from the radiation feature data in the target medical device information to generate radiation feature time series data, including: Preprocess the radiation feature data in the target medical device information based on the multi-dimensional medical radiation feature database to generate several data subsets: Process the several data subsets based on a filter denoising algorithm to generate denoised data; Process the denoised data to generate spatio-temporal feature normalized data; Process the spatio-temporal feature normalized data to generate multi-source fusion data; Process the multi-source fusion data to generate radiation feature time series data.

3. The method according to claim 1, wherein Process the radiation feature time series data based on a convolutional neural network to generate radiation warning feature parameters, including: Extract features from the radiation feature time series data based on a convolutional neural network to generate a multi-dimensional feature vector set; Compare the feature vector sets at adjacent time steps to determine the evolution law of radiation features over time and generate feature dynamic change trajectory information; Compare the feature dynamic change trajectory information with a preset normal radiation pattern library and abnormal radiation pattern library to generate a feature deviation degree evaluation result; Combine the device operation status corresponding to the radiation feature time series data and the spatio-temporal context information of the environmental background radiation, and perform weighted processing on the feature deviation degree evaluation result to generate a radiation risk feature vector; Normalize and comprehensively calculate the feature deviation degree evaluation result and the radiation risk feature vector to generate a radiation warning feature parameter set including a dose anomaly coefficient and an energy spectrum offset index.

4. The method according to claim 1, characterized in that, Perform outlier filtering on the radiation feature time series data based on an adaptive threshold filtering method to generate target radiation feature data, including: Perform preliminary outlier screening on the radiation feature time series data based on an adaptive threshold filtering method to generate a preliminary screening data set; Process the continuous abnormal data segments in the preliminary screening data set to generate a first optimized data set; Process the environmental interference outliers in the first optimized data set to generate a second optimized data set; Perform time-series feature grouping processing on the secondary optimized data set to generate a radiation feature sequence and an intermediate data set; Adjust the outlier filtering confidence based on the radiation feature sequence to generate target radiation feature data.

5. The method according to claim 4, characterized in that Process the target radiation feature data to generate a radiation intensity dynamic change curve and an abnormal radiation risk feature, including: Perform time-series trend analysis processing on the target radiation feature data to generate radiation intensity time-series parameters; Obtain the energy spectrum distribution data and equipment operating condition information of the target radiation feature data; Process the energy spectrum distribution data to generate radiation energy feature parameters; Process the equipment operating condition information, statistically analyze the radiation feature distribution under different operating conditions, and generate a radiation and operating condition correlation baseline value, including the radiation dose rate fluctuation range during normal equipment operation and the feature deviation threshold under abnormal operating conditions; Process the radiation intensity time-series parameters, radiation energy feature parameters, and radiation and operating condition correlation baseline value to generate a radiation intensity dynamic change curve and an abnormal radiation risk feature.

6. The method according to claim 1, wherein Process the radiation intensity dynamic change curve based on the preset dynamic warning rule and the abnormal radiation risk feature to generate warning decision data, including: Perform hierarchical parsing processing on the preset dynamic warning rule to generate warning rule weight coefficients; Obtain the abnormal radiation risk feature data and the real-time data of the radiation intensity dynamic change curve; Process the abnormal radiation risk feature data to generate risk feature quantization parameters; Process the real-time data of the radiation intensity dynamic change curve, statistically analyze the curve shape feature and the threshold breakthrough frequency, and generate a curve feature baseline value, including the dose rate rising slope and the peak exceeding standard times; Perform weighted fusion processing on the warning rule weight coefficients, risk feature quantization parameters, and curve feature baseline value to generate warning decision data.

7. The method according to claim 6, wherein Process the warning decision data and the radiation warning feature parameters to generate a radiation exceeding standard warning message, an equipment abnormal shutdown instruction, and optimize the radiation protection energy consumption management, including: Perform risk level analysis processing on the warning decision data to generate a risk level assessment result; Perform abnormal feature comparison processing based on the radiation warning feature parameters to generate an equipment abnormality determination result; Process the risk level assessment result and the equipment abnormality determination result to generate a radiation exceeding standard warning message; Obtain the real-time information of the equipment operating state and the energy consumption configuration information of the radiation protection system; Process the energy consumption configuration information based on the equipment operating state and the radiation exceeding standard warning message to generate an equipment abnormal shutdown instruction and optimize the radiation protection energy consumption management.

8. A radiation detection and warning device for medical equipment, characterized in that, The device includes: An acquisition module for acquiring a multi-dimensional medical radiation feature database and target medical equipment information, where the multi-dimensional medical radiation feature database includes medical equipment operation radiation data and environmental background radiation data; A processing module is configured to preprocess and extract features from the radiation characteristic data in the target medical device information to generate radiation characteristic time series data, where the radiation characteristic time series data consists of a filtering and noise reduction result, a spatio-temporal feature normalization relationship, and a multi-source data fusion and enhancement result; process the radiation characteristic time series data based on a convolutional neural network to generate radiation warning characteristic parameters; perform outlier filtering on the radiation characteristic time series data based on an adaptive threshold filtering method to generate target radiation characteristic data; process the target radiation characteristic data to generate a radiation intensity dynamic change curve and an abnormal radiation risk characteristic; process the radiation intensity dynamic change curve based on a preset dynamic warning rule and the abnormal radiation risk characteristic to generate warning decision data; process the warning decision data and the radiation warning characteristic parameters to generate a radiation over-standard warning message, a device abnormal shutdown instruction, and optimize radiation protection energy consumption management.

9. An electronic device, characterized in that, It includes: A first processor; And a memory for storing executable instructions of the first processor; Wherein, the first processor is configured to execute the radiation detection and warning method for medical devices according to any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a second processor, it implements the radiation detection and warning method for medical devices according to any one of claims 1 to 7.

Images