X-ray imaging equipment for real-time monitoring in radiology department

Abstract

The invention provides X-ray imaging equipment for real-time monitoring in the radiology department, and relates to the technical field of medical instruments, the X-ray imaging equipment comprises an imaging execution module, a data fusion acquisition module, a monitoring decision module and a feedback execution module, the imaging execution module generates original projection data; the data fusion acquisition module generates a comprehensive data set; the monitoring decision module receives and processes the comprehensive data set, judges a risk level through a risk assessment model, and generates a corresponding decision instruction; and the feedback execution module receives the decision instruction, performs real-time regulation and control on the original projection data according to the decision instruction, and outputs visual monitoring and early warning information to the operation interface at the same time. The problems that in the prior art, due to the split relation between the imaging function and the monitoring function, intelligent linkage monitoring cannot be carried out based on imaging operation, the physiological state of a subject is abnormal, and equipment operation risk identification lags behind are solved.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, and in particular to an X-ray imaging device for real-time monitoring in radiology departments. Background Technology

[0002] In existing technologies, imaging systems focus on image acquisition and reconstruction, while monitoring functions rely on external devices such as electrocardiogram monitors and pulse oximeters. Judgments are made through manual observation or simple alarm thresholds, resulting in a separation between imaging and monitoring functions. This separation leads to a temporal and spatial disconnect between monitoring information and the imaging process, making it impossible to achieve intelligent linkage monitoring based on imaging operations. For example, in dynamic fluoroscopy, the examinee may experience physiological fluctuations due to discomfort in body position or equipment movement, but existing technologies struggle to correlate these fluctuations with specific imaging events, thus delaying the identification of real risks. On the other hand, the monitoring of key parameters such as the X-ray tube temperature and the stability of the high-voltage generator is often independent of the image quality assessment system, making it impossible to warn of potential equipment failures before image artifacts appear.

[0003] Therefore, there is an urgent need to provide an X-ray imaging device that can deeply integrate and intelligently analyze the imaging process and monitoring information. Summary of the Invention

[0004] The purpose of this invention is to provide an X-ray imaging device for real-time monitoring in radiology departments, which solves the problem of the disconnect between imaging and monitoring functions in the prior art, which makes it impossible to perform intelligent linkage monitoring based on imaging operations, resulting in a lag in the identification of abnormal physiological states of examinees and risks in equipment operation.

[0005] The technical solution of the present invention:

[0006] This invention provides an X-ray imaging device for real-time monitoring in radiology departments, comprising:

[0007] The imaging execution module performs X-ray generation, collimation, penetration of the subject, and detection and reception, and generates raw projection data;

[0008] The data fusion acquisition module simultaneously acquires imaging process data, subject physiological monitoring data, and equipment operating status data, and generates a comprehensive dataset.

[0009] The monitoring decision module receives and processes a comprehensive dataset, determines the risk level through a risk assessment model, and generates corresponding decision instructions.

[0010] The feedback execution module receives decision instructions and performs real-time control on the original projection data accordingly, while outputting visualized monitoring and early warning information to the operation interface.

[0011] In one possible implementation, the data fusion acquisition module includes an imaging process acquisition unit, a physiological data acquisition unit, an equipment status acquisition unit, and a data processing unit. The imaging process acquisition unit is directly coupled to the control bus and data bus of the imaging execution module. The physiological data acquisition unit is integrated into the contact area of ​​the examination bed or robotic arm. The equipment status acquisition unit is deployed inside the X-ray tube heat dissipation duct, high-voltage generator, and detector. The data processing unit has a built-in high-precision clock source.

[0012] In one possible implementation, the imaging process acquisition unit captures and records in real time the time when each exposure command is issued, the preset tube voltage value, tube current value, exposure duration, the current anatomical site identifier, and the spatial pose coordinates of the robotic arm, forming an imaging process dataset.

[0013] The physiological data acquisition unit includes a biosensor component, which acquires electrocardiogram signals, respiratory waveforms, local body motion acceleration, and body surface temperature data from the subject's body surface to form a physiological dataset.

[0014] The equipment status acquisition unit collects data on the anode target surface temperature of the X-ray tube, the high-voltage output ripple coefficient, the detector dark current noise level, and the coolant circulation pressure to form an equipment status dataset.

[0015] The data processing unit receives imaging process datasets, physiological datasets, and device status datasets, and adds a microsecond-level timestamp to each frame data packet based on the same time base. It also encapsulates the multi-source heterogeneous data into a continuous comprehensive dataset according to a preset data structure.

[0016] In one possible implementation, the monitoring decision module includes a feature extraction unit, a correlation analysis unit, a risk assessment model, and a decision generation unit, wherein:

[0017] The feature extraction unit performs real-time processing of the feature data on the input dataset and forms a feature dataset;

[0018] The correlation analysis unit receives the raw projection data and feature dataset, and establishes a causal correlation map between imaging operation events and physiological and device responses;

[0019] The risk assessment model consists of an equipment risk sub-model, a physiological risk sub-model, and a cross-risk sub-model. It receives and correlates data from the analysis unit and generates corresponding equipment risk scores, physiological risk scores, and coupling risk scores.

[0020] The decision generation unit receives the device risk score, physiological risk score, and coupling risk score, calculates the comprehensive risk level according to a preset weighted fusion strategy, and generates a decision instruction.

[0021] In one possible implementation, the feature extraction unit extracts cumulative exposure dose, mechanical motion speed and acceleration features from the imaging process dataset, heart rate variability, respiratory rhythm disorder, body movement amplitude and frequency features from the physiological dataset, and temperature change rate, voltage stability index and noise spectrum features from the equipment status dataset.

[0022] In one possible implementation, the equipment risk sub-model uses a pre-trained time-series prediction algorithm to predict the probability of performance degradation or failure of key components of the equipment within a specified future time period, and generates an equipment risk score.

[0023] The physiological risk sub-model assesses the risk of subjects experiencing physiological stress, discomfort, or potential abnormal vital signs, and generates a physiological risk score.

[0024] The cross-risk sub-model calculates the cross-influence weights of the device risk sub-model and the physiological risk sub-model through an attention mechanism network, and outputs a coupled risk score.

[0025] In one possible implementation, the decision generation unit is configured with a decision instruction mapping table, which generates corresponding decision instructions when the overall risk level reaches a preset threshold.

[0026] In one possible implementation, the decision instruction mapping table contains at least four risk levels, wherein:

[0027] Level 1 is the attention level, which only triggers an on-screen prompt.

[0028] Level 2 is the warning level, which triggers an audible and visual alarm and suggests adjusting the imaging plan.

[0029] Level 3 is the intervention level, which automatically performs parameter adjustments to reduce exposure dose or limit mechanical movement speed;

[0030] Level 4 is an emergency level, requiring immediate cessation of exposure and reverting the equipment to a safe state.

[0031] In one possible implementation, the feedback execution module includes a parameter control unit, a motion control unit, and a human-computer interaction unit, wherein:

[0032] The parameter control unit directly sends digital adjustment signals to the imaging execution module and dynamically modulates them in real time.

[0033] The motion control unit sends motion correction commands to the imaging execution module;

[0034] The human-computer interaction unit performs centralized visual presentation on the operation interface and triggers audible and visual alarm signals.

[0035] Based on the above technical features, the beneficial effects of the present invention are as follows:

[0036] (1) The X-ray imaging device provided by the present invention, by configuring a data fusion acquisition module, can realize millisecond-level synchronization and fusion of imaging process data, subject physiological data and device operating status data, thereby breaking the information silos in traditional systems and devices, and providing a data foundation for accurate and timely risk response.

[0037] (2) The X-ray imaging device provided by the present invention can causally correlate discrete monitoring signals with specific imaging operation events through the correlation analysis unit and risk assessment model configured in the monitoring decision module, and can quantitatively assess the risks of the device itself, the physiological risks of the examinee and their cross-coupling risks, so that risk identification is upgraded from passive threshold alarm to proactive and predictive comprehensive assessment.

[0038] (3) The X-ray imaging device provided by the present invention can continuously and cyclically execute a closed-loop process of data acquisition, feature extraction, risk assessment and collaborative control to ensure the timeliness of intervention measures. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the architecture of the present invention;

[0040] Figure 2 This is a schematic diagram illustrating the principle of the monitoring decision module in this invention. Detailed Implementation

[0041] The following will be based on embodiments of the present invention. Figures 1-2 The technical solutions in the embodiments of the present invention will be clearly and completely described together. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0042] Example

[0043] An embodiment of the present invention provides an X-ray imaging device for real-time monitoring in a radiology department, the device comprising:

[0044] The imaging execution module performs X-ray generation, collimation, penetration of the subject, and detection and reception, and generates raw projection data;

[0045] The data fusion acquisition module simultaneously acquires imaging process data, subject physiological monitoring data, and equipment operating status data, and generates a comprehensive dataset.

[0046] The monitoring decision module receives and processes the comprehensive dataset, determines the risk level through a risk assessment model, and generates corresponding decision instructions.

[0047] The feedback execution module receives decision instructions and performs real-time adjustments to the original projection data accordingly, while simultaneously outputting visualized monitoring and early warning information to the operation interface.

[0048] To further clarify the present invention, it is described in detail below:

[0049] First, the imaging execution module performs X-ray generation, collimation, penetration of the subject, and detection and reception, and generates raw projection data. In detail, the imaging execution module includes an X-ray tube, a high-voltage generator, an adjustable collimator, a robotic arm drive mechanism, and a flat panel detector. For example, in some typical chest X-ray examinations, medical staff set the anatomical location as the chest through a human-computer interface, select the imaging mode as static radiography, and input preset tube voltage values ​​of 120 kV, tube current values ​​of 400 mA, and exposure duration of 0.1 seconds. After receiving the above instructions, the imaging execution module uses the robotic arm drive mechanism to move the X-ray tube and the flat panel detector to preset spatial pose coordinates, ensuring that the central X-ray is perpendicular to the midline of the patient's chest. Then, the high-voltage generator applies a preset high-voltage electric field to the X-ray tube, exciting the electron beam to bombard the anode target surface, generating an X-ray beam with a specific energy spectrum distribution. After the X-ray beam is limited by the adjustable collimator to the irradiation field range, it penetrates the patient's body cavity and is finally received by the flat panel detector and converted into raw projection data. The key operating parameters and state variables throughout the process are recorded in real time, forming raw projection data that is input into the data fusion acquisition module for analysis.

[0050] Next, the data fusion acquisition module simultaneously acquires imaging process data, subject physiological monitoring data, and equipment operating status data, and generates a comprehensive dataset. The data fusion acquisition module includes an imaging process acquisition unit, a physiological data acquisition unit, an equipment status acquisition unit, and a data processing unit, specifically:

[0051] First, the imaging process acquisition unit is directly coupled to the control bus and data bus of the imaging execution module, capturing the complete context information of each exposure event with microsecond-level sampling accuracy. Specifically, it captures and records in real time the time when each exposure command is issued, the preset tube voltage value, tube current value, exposure duration, the current anatomical site identification, and the spatial pose coordinates of the robotic arm. The above data, combined with the original projection data, forms the imaging process dataset, which is temporarily stored in the local buffer in a structured message format, waiting for synchronous processing.

[0052] Secondly, the physiological data acquisition unit is integrated into the contact area of ​​the examination bed or robotic arm. This unit includes a biosensor component, which is encapsulated using flexible printed circuits and biocompatible materials. The biosensor component acquires electrocardiogram (ECG) signals, respiratory waveforms, local body motion acceleration, and body surface temperature data from the subject's body surface in a non-invasive manner. Optionally, the biosensor component includes ECG electrode pairs, a piezoelectric respiratory sensor, a triaxial microelectromechanical system (MEMS) accelerometer, and an infrared thermistor. In some embodiments, the ECG electrode pairs are connected to the shoulder and hip areas of the examination bed, enabling continuous acquisition of standard lead I ECG data. The sampling frequency is preferably 500 Hz. In some embodiments, a piezoelectric respiratory sensor is embedded in a chest band support structure to generate a respiratory waveform signal by detecting minute pressure changes caused by chest wall expansion, with a sampling frequency preferably 100 Hz. In some embodiments, a triaxial microelectromechanical system accelerometer is installed at the contact point on the back of the subject's torso to monitor the amplitude and direction of local body movements in real time and output three-channel acceleration data, with a sampling frequency preferably 200 Hz. In some embodiments, an infrared thermistor is attached to the skin surface of the subject's forehead or back of the hand, preferably measuring the body surface temperature at a frequency of once per second with an accuracy of ±0.1 degrees Celsius. All physiological signals acquired by the biosensing components are converted from analog to digital to form a physiological dataset, which is then input to the data processing unit after being appended with a local timestamp.

[0053] Thirdly, the equipment status acquisition unit is deployed inside the X-ray tube's cooling duct, high-voltage generator, and detector to monitor the operational health of the core components. It collects data on the tube's anode target surface temperature, high-voltage output ripple coefficient, detector dark current noise level, and coolant circulation pressure to form an equipment status dataset. Specifically, a high-precision thermocouple is connected at the tube's cooling duct outlet to measure the anode target surface cooling airflow temperature in real time, indirectly reflecting the target surface's thermal load. A high-frequency voltage probe is connected to the high-voltage generator's output port to acquire the high-voltage output waveform and calculate its ripple coefficient using a fast Fourier transform, which is then used to assess power supply stability. Inside the flat panel detector, a dark current monitoring pixel array is connected to periodically read the background noise level under X-ray-free conditions. A pressure sensor is connected in the tube's coolant circulation loop to monitor the cooling pump's operating status and the risk of pipe blockage. By acquiring various equipment status data at a frequency of one frame per second and marking the acquisition time, an equipment status dataset is formed.

[0054] Fourth, the data processing unit has a built-in high-precision clock source based on a temperature-controlled crystal oscillator, with a frequency stability better than ±0.1ppm. The data processing unit receives imaging process datasets, physiological datasets, and device status datasets, and adds a microsecond-level timestamp to each frame data packet based on the same time base. It then encapsulates the multi-source heterogeneous data into a continuous comprehensive dataset according to a preset data structure. The specific process is as follows: First, time alignment correction is performed to compensate for the transmission delay between the imaging process acquisition unit, physiological data acquisition unit, and device status acquisition unit based on network time protocol or hardware trigger signal, ensuring that the timestamps of all data frames are based on the same physical clock. Then, according to the preset fusion data structure, the received imaging process dataset, physiological dataset, and device status dataset are interleaved and encapsulated in chronological order to form a continuous comprehensive dataset. The basic frame structure of the comprehensive dataset includes a frame header, timestamp field, imaging data block, physiological data block, device status data block, and check code. The generation period of each frame is driven by the system master clock, with a typical value of 10 milliseconds, that is, 100 frames of fusion data stream are generated per second, thus meeting the timeliness requirements of real-time monitoring.

[0055] Then, the monitoring decision-making module receives and processes the comprehensive dataset, determines the risk level through a risk assessment model, and generates corresponding decision instructions. The monitoring decision-making module includes a feature extraction unit, a correlation analysis unit, a risk assessment model, and a decision generation unit. Specifically:

[0056] Firstly, the feature extraction unit performs real-time processing of the input dataset, specifically real-time parsing and feature engineering, thereby forming a feature dataset. The feature extraction unit extracts cumulative exposure dose, mechanical motion speed, and acceleration features from the imaging process dataset. Cumulative exposure dose is obtained by weighted integration of tube voltage, tube current, exposure time, and irradiation field area for each exposure. Mechanical motion speed and acceleration features are based on a continuous sequence of changes in the robotic arm's pose coordinates; numerical differentiation is used to calculate the current motion speed and acceleration, and the maximum instantaneous acceleration is extracted as an impact indicator. Secondly, the feature extraction unit extracts heart rate variability, respiratory rhythm disorder, and body movement amplitude and frequency features from the physiological dataset. The feature extraction unit first processes the electrocardiogram signal... R-wave detection is performed, and the standard deviation of adjacent RR intervals is calculated to characterize heart rate variability. Next, zero-crossing analysis and spectral decomposition are performed on the respiratory waveform to calculate the respiratory rhythm disorder, defined as the root mean square deviation between the actual respiratory cycle and the ideal cycle. Then, envelope detection and frequency domain analysis are performed on the acceleration signal to extract body motion amplitude and dominant frequency. Finally, the slope of change within a sliding window is calculated for the body surface temperature sequence. The feature extraction unit extracts temperature change rate, voltage stability index, and noise spectrum features from the equipment status dataset. Specifically, it calculates the rate of change of the X-ray tube temperature over the past 10 seconds, the moving average of the high-voltage ripple coefficient, the peak power spectral density of dark current noise, and the fluctuation amplitude of cooling hydraulic pressure. The feature extraction unit integrates the above data to form a feature dataset and inputs it into the correlation analysis unit.

[0057] Secondly, the correlation analysis unit receives the raw projection data and feature dataset, and establishes a causal correlation graph between imaging operation events and physiological and equipment responses. Specifically, the correlation analysis unit receives the feature dataset, extracts its feature vectors, and combines them with the timestamps and anatomical site identifiers in the raw projection data to construct a dynamic causal correlation graph. The causal correlation graph maintains a sliding event context window with a length of 30 seconds. Specifically, when an exposure event or a large-scale movement event of the robotic arm is detected, the correlation analysis unit automatically activates the correlation analysis process. In some embodiments, an exposure event is defined as the time when the exposure command is issued. As an anchor point, a single large-amplitude robotic arm movement event can be selected with an acceleration exceeding 500 mm / s² and a duration greater than 0.5 seconds. In some embodiments, taking the event occurrence time as the center, a time series of all physiological characteristics and equipment status characteristics within this time window is extracted by looking back 10 seconds and forward 20 seconds. In some embodiments, for each characteristic, its relative deviation from the average value of the baseline period 5 seconds before the event is calculated, and the Pearson correlation coefficient between the characteristic and the imaging event type is calculated as an indicator of correlation strength. All deviations and correlation coefficients constitute a context-related feature vector, which is input into the risk assessment model. The formula for the standardized deviation is as follows:

[0058]

[0059] In the formula, Let be the standardized deviation of the i-th feature. This is the instantaneous value of the feature within the time window following the event. This is the mean of the feature at the baseline period. Let be the standard deviation of this characteristic at the baseline period. To analyze window length;

[0060] Third, the risk assessment model consists of equipment risk sub-models, physiological risk sub-models, and cross-risk sub-models. It receives and correlates data from the analysis units and generates corresponding equipment risk scores, physiological risk scores, and coupling risk scores. The equipment risk sub-models, physiological risk sub-models, and cross-risk sub-models run in parallel. Specifically:

[0061] The equipment risk sub-model employs a long short-term memory (LSTM) network architecture. Based on equipment state characteristics and their changing trends, the sub-model uses a pre-trained time-series prediction algorithm to predict the probability of performance degradation or failure of key equipment components within a specified future timeframe, and generates an equipment risk score. Specifically, the input to the equipment risk sub-model is a dataset of equipment state information obtained over the past 60 consecutive seconds at a sampling frequency of once per second. Equipment state feature sequences are extracted from this dataset, including tube temperature, high-voltage ripple coefficient, and dark current noise level. The LTM network architecture contains two hidden layers, each with 128 memory units. After offline training, it learns long-term dependency patterns in equipment state evolution. The network output layer predicts the probability of tube overheating within the next 30 seconds. High voltage flashover probability Detector performance degradation probability Based on the probability of X-ray tube overheating High voltage flashover probability Detector performance degradation probability Obtain the equipment risk score. Through linear weighted synthesis:

[0062]

[0063] In the formula, , , These are all weighting coefficients, which are determined based on historical fault statistics. In some embodiments, , , ;

[0064] The physiological risk sub-model assesses the risk of subjects experiencing physiological stress, discomfort, or potential vital sign abnormalities, and generates a physiological risk score. The detailed physiological risk score assessment process is as follows: First, the physiological risk sub-model establishes a baseline database of basic physiological characteristics for subjects under different anatomical locations and imaging modes based on historical normal examination data. When performing real-time assessment, it retrieves the current physiological characteristics from the feature extraction unit and obtains the current imaging event type and its correlation strength with the physiological characteristics from the correlation analysis unit. Next, it calculates the deviation of the current physiological characteristics from the corresponding baseline. Finally, it combines the deviation, correlation strength, and deviation... The duration is used as input to a trained 3-layer neural network classifier. The output layer nodes of the neural network classifier correspond to different stress risk levels. The physiological risk score at the current moment is calculated through forward propagation. In some embodiments, the input layer of the neural network classifier contains 9 nodes, the hidden layer contains 32 nodes, and the output layer contains 4 nodes, corresponding to four levels: "no risk," "low risk," "medium risk," and "high risk," respectively. After forward propagation and softmax activation, the output layer obtains the probability distribution of each category. The value assigned to the level corresponding to the highest probability is taken as the physiological risk score. ;

[0065] The cross-risk sub-model handles coupled scenarios where equipment malfunctions induce physiological risks or physiological malfunctions exacerbate equipment risks. The input consists of intermediate output features from the equipment risk sub-model and the physiological risk sub-model. An attention mechanism network calculates the cross-influence weights between the two sub-models and outputs a coupled risk score. Specifically, the feature vectors K and V from the last hidden state of the equipment risk sub-model are used as the key and value vectors, respectively. The feature vector Q from the hidden layer of the physiological risk sub-model is used as the query vector. First, Q is mapped to a 128-dimensional space using linear projection to obtain Q′. Then, the attention score is calculated. This yields a normalized attention weight vector, which is then used to perform a weighted summation of the value vector V to generate a context vector. This vector represents the device state factor that contributes most to the current physiological risk. Finally, C is concatenated with the original physiological feature vector and input into a fully connected layer containing 64 neurons. After Sigmoid activation, the coupling risk score is output. .

[0066] Fourth, the decision generation unit receives the risk score from the receiving device. Physiological risk score Coupling risk score The system calculates the overall risk level according to a preset weighted fusion strategy and generates decision instructions. The decision generation unit is equipped with a decision instruction mapping table that strictly corresponds to different risk levels. When the overall risk level reaches a preset threshold, the control actions corresponding to each level of risk are strictly defined, and corresponding decision instructions are generated. The specific process is as follows: First, the physiological risk score is... Normalization to Within the interval, then calculate the weighted sum:

[0067]

[0068] In the formula, ,according to The value is mapped to a level 4 risk level: when At level 1, only an on-screen prompt will be triggered to indicate the level of attention. The current level is Level 2, which is a warning level. It triggers an audible and visual alarm and suggests adjusting the imaging plan. At level 3, which is the intervention level, parameters are automatically adjusted to reduce exposure dose or limit mechanical movement speed. The current level is Level 4, which is an emergency level. Immediately stop the exposure and control the equipment to return to a safe state.

[0069] Finally, the feedback execution module receives the decision command and performs real-time adjustment and feedback operations on the original projection data accordingly. Simultaneously, it outputs visualized monitoring and early warning information to the operation interface. The feedback execution module includes a parameter adjustment unit, a motion control unit, and a human-computer interaction unit, specifically:

[0070] Firstly, the parameter control unit directly sends digital adjustment signals to the imaging execution module through the digital communication interface and dynamically modulates them in real time. That is, it sends tube voltage adjustment commands to the high voltage generator and tube current adjustment commands to the filament current controller. In some embodiments, when the risk is determined to be level 3 and the main risk source is the subject's sudden body movement, the parameter control unit immediately sends a command to reduce the tube current of the current exposure by 50%, or if the exposure has not yet started, it delays the start until the body movement amplitude falls below the safe threshold.

[0071] Secondly, the motion control unit sends motion correction commands to the imaging execution module. The motion control unit sends speed limit commands or position retraction commands to the motors of each joint of the robotic arm through the servo driver interface. In some embodiments, when the risk score of the device layer indicates that the X-ray tube temperature rises too quickly, the motion control unit injects motion damping into the robotic arm control system to limit the maximum motion speed to 60% of the original set value. If it is determined to be a level 4 emergency risk, all motion commands are immediately interrupted and the robotic arm is driven to return to the initial parking position along the preset safety trajectory.

[0072] Third, the human-computer interaction unit performs centralized visualization on the operation interface and triggers audible and visual alarm signals. The human-computer interaction unit will graphically overlay the comprehensive risk level, main risk sources, related imaging events and corresponding physiological data segments on the main control screen. Optionally, the presentation methods include, but are not limited to: displaying a colored risk level icon in the upper right corner of the real-time fluoroscopic image; drawing heart rate and respiratory trend curves below the image and marking abnormal intervals with red dotted lines; when the risk level reaches level 2 or above, the audible and visual alarm device will be triggered simultaneously, emitting buzzers and flashing warning lights at different frequencies.

[0073] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the scope of the invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0074] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. An X-ray imaging apparatus for radiology real-time monitoring, characterized in that, The system comprises: an imaging execution module for generating, collimating, penetrating the subject and detecting the received X-rays, and generating raw projection data; a data fusion acquisition module for synchronously acquiring imaging process data, subject physiological monitoring data and equipment operating state data, and generating a comprehensive data set; a monitoring decision module for receiving and processing the comprehensive data set, determining the risk level through a risk assessment model, and generating corresponding decision instructions; a feedback execution module for receiving the decision instructions, and performing real-time regulation on the raw projection data accordingly, while outputting visual monitoring warning information to the operation interface.

2. The apparatus of claim 1, wherein, The data fusion acquisition module comprises an imaging process acquisition unit, a physiological data acquisition unit, an equipment state acquisition unit and a data processing unit. The imaging process acquisition unit is directly coupled to the control bus and data bus of the imaging execution module. The physiological data acquisition unit is integrated in the contact part of the examination bed or mechanical arm. The equipment state acquisition unit is disposed in the X-ray tube heat dissipation air duct, high-voltage generator and detector. The data processing unit is built-in high-precision clock source.

3. The apparatus of claim 2, wherein, The imaging process acquisition unit captures and records the time of issuing each exposure instruction, the preset tube voltage value, the tube current value, the exposure duration, the current imaging anatomical site identifier and the spatial pose coordinates of the mechanical arm in real time, forming an imaging process data set. The physiological data acquisition unit contains a biological sensing component, which acquires the ECG signal, respiratory waveform, local body acceleration and body surface temperature data of the subject, forming a physiological data set. The equipment state acquisition unit acquires the anode target surface temperature of the tube, the high-voltage output ripple coefficient, the detector dark current noise level and the cooling liquid circulation pressure data, forming an equipment state data set. The data processing unit receives the imaging process data set, physiological data set and equipment state data set, and stamps each frame of data packet with a microsecond-level time stamp based on the same time reference, and encapsulates the multi-source heterogeneous data into a continuous comprehensive data set according to the preset data structure.

4. The apparatus of claim 3, wherein, The monitoring decision module comprises a feature extraction unit, a correlation analysis unit, a risk assessment model and a decision generation unit, wherein: The feature extraction unit performs real-time processing of the feature data of the input data set, and forms a feature data set; The correlation analysis unit receives the raw projection data and the feature data set, and establishes a causal correlation graph between the imaging operation events and the physiological and equipment responses; The risk assessment model is composed of a device risk sub-model, a physiological risk sub-model and a cross-risk sub-model, which receives and correlates the data of the analysis unit and generates the device risk score, the physiological risk score and the coupled risk score accordingly; The decision generation unit receives the device risk score, the physiological risk score and the coupled risk score, calculates the comprehensive risk level according to the preset weighted fusion strategy, and generates the decision instructions.

5. The apparatus of claim 4, wherein, The feature extraction unit extracts the cumulative exposure dose, mechanical motion speed and acceleration features from the imaging process data set, extracts the heart rate variability, respiratory rhythm disorder degree, body motion amplitude and frequency features from the physiological data set, and extracts the temperature change rate, voltage stability index and noise spectrum features from the equipment state data set.

6. The apparatus of claim 4, wherein, The equipment risk sub-model adopts a pre-trained time series prediction algorithm to predict the probability of performance degradation or failure of key components of the equipment within a specified future time, and generates an equipment risk score; The physiological risk sub-model assesses the risk of physiological stress, discomfort or potential abnormal vital signs in the subject, and generates a physiological risk score; The cross-risk sub-model calculates the cross-influence weight of the equipment risk sub-model and the physiological risk sub-model through an attention mechanism network, and outputs a coupled risk score.

7. The apparatus of claim 6, wherein, The decision generation unit is configured with a decision instruction mapping table, which generates corresponding decision instructions when the comprehensive risk level reaches a preset threshold.

8. The apparatus of claim 7, wherein, The decision instruction mapping table includes at least 4 risk levels, wherein: The first level is the attention level, which only triggers the interface prompt; The second level is the warning level, which triggers the sound and light alarm and suggests adjusting the imaging plan; The third level is the intervention level, which automatically executes parameter regulation to reduce exposure dose or limit mechanical motion speed; The fourth level is the emergency level, which immediately suspends exposure and controls the equipment to return to a safe state.

9. The apparatus of claim 4, wherein, The feedback execution module includes a parameter regulation unit, a motion control unit and a human-computer interaction unit, wherein: The parameter regulation unit directly sends digital adjustment signals to the imaging execution module and dynamically adjusts in real time; The motion control unit sends motion correction instructions to the imaging execution module; The human-computer interaction unit can be visually presented in the operation interface execution set and trigger the sound and light alarm signal.

Images