A high-pressure syringe based on visual system monitoring

Abstract

The application discloses a high-pressure injector based on visual system monitoring and relates to medical devices. The application is proposed in view of the single monitoring means in the prior art. A first camera is installed outside a first bottle inserting structure and is used for acquiring state data of a contrast medium liquid bottle; a second camera is installed outside a second bottle inserting structure and is used for acquiring state data of a saline liquid bottle; a third camera is installed outside an output end of a liquid mixing device and is used for acquiring state data of mixed liquid; a fourth camera is installed outside a pressurizing mechanism and is used for acquiring state data of the pressurizing mechanism; and a fifth camera is installed outside an output end of the pressurizing mechanism and is used for acquiring state data of a pipeline liquid bubble. The application has the advantages that visual monitoring and electrical signal monitoring can work in parallel, the visual monitoring can be added without interfering with the original electrical signal monitoring, the injection state can be captured in all directions, and medical safety is improved.

Description

Technical Field

[0001] This invention relates to medical devices, and more particularly to a high-pressure injector based on vision system monitoring. Background Technology

[0002] like Figure 1 As shown, a conventional high-pressure injector includes the following essential components: a first insert structure for mounting a contrast agent bottle, a second insert structure for mounting a saline bottle, a mixing device for mixing the contrast agent and saline, a pressurizing mechanism for pressurizing the mixed solution, and an injection needle for injecting into a human artery.

[0003] Existing high-pressure injectors only obtain the injection status through electrical signal feedback and control the injection procedure through electrical signal output. The electrical control signal is direct and simple, but it is also prone to monitoring vulnerabilities or signal interference.

[0004] With the development of artificial intelligence, especially the deep application of vision systems, the need for comprehensive monitoring of the injection process through vision systems is becoming increasingly urgent. Summary of the Invention

[0005] The purpose of this invention is to provide a high-pressure injector based on vision system monitoring to solve the problems existing in the prior art.

[0006] The high-pressure injector based on vision system monitoring described in this invention includes: a first insert structure for installing a contrast agent bottle, a second insert structure for installing a saline bottle, a mixing device for mixing the contrast agent and saline, a pressurizing mechanism for pressurizing the mixed solution, and an injection needle for injecting into a human artery. A first camera is installed on the outside of the first bottle insertion structure to acquire status data of the contrast agent bottle; A second camera is installed on the outside of the second bottle insertion structure to acquire status data of the saline bottle; A third camera is installed on the outside of the output end of the mixing device to acquire the status data of the mixed solution; A fourth camera is installed on the outside of the pressurization mechanism to acquire status data of the pressurization mechanism; A fifth camera is installed on the outside of the output end of the pressurizing mechanism to acquire the status data of the liquid bubbles in the pipeline.

[0007] The status data of the contrast agent bottle includes: bottle installation status, bottle volume, and bottle label data.

[0008] The status data of the saline solution bottle includes: bottle installation status, bottle volume, and bottle label data.

[0009] The state data of the mixed drug solution includes: refractive index, transmittance, and color.

[0010] The status data of the pressurizing mechanism includes: actual injection speed or actual angular velocity; When the pressurization mechanism is a push-type mechanism, the status data is the actual push-in speed; When the pressurizing mechanism is a wheel-type mechanism, the state data is the actual angular velocity.

[0011] The status data of the pipeline bubbles include: the length of a single bubble, the number of bubbles per unit time, and the cumulative length of bubbles per unit time.

[0012] The monitoring process includes the following steps: S1. Acquire corresponding status data through the first and second cameras to monitor whether the liquid bottle meets the injection conditions; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning. S2. Obtain the corresponding status data through the fourth camera and monitor whether the actual speed meets the preset error with the preset speed of the program; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning. S3. Acquire corresponding status data through the third camera to monitor whether the mixing effect meets the injection conditions; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning. S4. Obtain the corresponding status data through the fifth camera to monitor whether the blister meets the injection conditions; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning.

[0013] The high-pressure injector based on vision system monitoring described in this invention has the advantage of enabling visual monitoring and electrical signal monitoring to work in parallel. Visual monitoring is added without interfering with the original electrical signal monitoring, which can capture the injection status from all angles and improve medical safety. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the structural principle of a high-pressure injector in the prior art.

[0015] Figure 2 This is a schematic diagram of the structural principle of the high-pressure injector described in this invention.

[0016] Figure 3 This is a schematic diagram of the monitoring process for the high-pressure injector described in this invention.

[0017] Figure label: 11-First bottle insertion mechanism, 12-Second bottle insertion mechanism; 21-Mixing apparatus; 31-Pressure mechanism; 41 - Injection needle; C1 - First camera, C2 - Second camera, C3 - Third camera, C4 - Fourth camera, C5 - Fifth camera. Detailed Implementation

[0018] like Figure 2 As shown, this invention discloses a high-pressure injector based on a vision system monitoring system. It boasts advantages such as full-process visual monitoring, real-time fault warning, and precise calibration of injection parameters. It effectively solves the technical defects of traditional high-pressure injectors, such as invisible drug-liquid ratios, lack of operational status monitoring, difficulty in detecting air bubbles in the tubing, and delayed alarms for injection faults, significantly improving the safety and accuracy of interventional angiography injection procedures. Specifically, the high-pressure injector includes: a first insert structure 11 for positioning and installing the contrast agent bottle; a second insert structure 12 for positioning and installing the saline bottle; a mixing device 21 for precisely mixing the contrast agent and saline to adjust the drug-liquid ratio; a pressurizing mechanism 31 for stabilizing and pressurizing the mixed drug solution to provide stable power for arterial injection; and an injection needle 41 for precisely piercing the human artery to complete targeted drug injection. These structures are sequentially connected to form a complete drug delivery, mixing, pressurization, and injection pathway. Simultaneously, with the addition of a multi-point visual monitoring component, intelligent monitoring of the entire injection process is achieved.

[0019] This invention innovatively incorporates a multi-camera distributed visual monitoring system. Monitoring cameras are deployed at key operating points of the equipment to comprehensively collect the equipment's operating status and drug parameters. The specific deployment scheme and monitoring functions are as follows: A first camera C1 is fixedly installed on the outside of the first bottle insertion structure 11. This camera is aimed at the installation area of ​​the contrast agent bottle and the overall structure of the bottle, continuously collecting various status data of the contrast agent bottle in real time, achieving pre-monitoring of the contrast agent supply end. A second camera C2 is fixedly installed on the outside of the second bottle insertion structure 12. This camera is directly facing the installation position of the saline bottle, comprehensively collecting the operating status data of the saline bottle, completing intelligent verification of the saline supply end.

[0020] A third camera C3 is installed on the outside of the drug output end of the mixing device 21 to focus on the area of ​​the mixed drug pipeline flowing out of the mixing device, and to collect optical state data of the mixed drug in real time, accurately judging the uniformity and quality of the drug mixture. A fourth camera C4 is installed on the outside of the main body of the pressurizing mechanism 31 to collect dynamic data of the pressurizing mechanism's operation, and to verify the power output accuracy of the pressurized injection in real time. A fifth camera C5 is installed on the outside of the drug output end of the pressurizing mechanism 31 to monitor the fluid state inside the delivery pipeline after pressurization, and to accurately collect various state data of the liquid bubbles inside the pipeline, avoiding medical risks caused by air injection into the human body.

[0021] The status data collected by each camera has clearly defined monitoring dimensions and judgment criteria. The specific parameter definitions are as follows: The contrast agent vial status data collected by the first camera C1 specifically includes three core parameters: vial installation status, remaining vial volume, and vial label data. The vial installation status determines whether the contrast agent vial is properly installed and whether there are any abnormalities such as misalignment, loosening, or detachment. The vial volume is used to monitor the remaining contrast agent capacity in real time, determining whether it meets the dosage requirements for this injection surgery and anticipating the risk of running out of medication. The vial label data uses visual recognition technology to read information such as the medication model, concentration, expiration date, and production batch number, verifying whether the currently used medication matches the pre-set surgical medication plan and preventing medication errors.

[0022] The data collected by the second camera C2 on the status of the saline solution bottle is consistent with the monitoring dimensions of the contrast agent bottle, including the bottle installation status, remaining liquid volume, and bottle label data. Visual monitoring enables fully automated verification of the compliance of saline solution installation, remaining liquid volume, and specifications, ensuring a compliant supply of saline solution for dilution and laying the foundation for accurate subsequent drug formulation.

[0023] The mixed drug solution status data collected by the third camera C3 includes three types of optical parameters: drug solution refractive index, drug solution transmittance, and drug solution color. Specifically, the drug solution refractive index accurately reflects the mixing ratio and concentration of the contrast agent and saline solution, determining whether the ratio meets surgical standards; the drug solution transmittance detects the presence of impurities, sediments, or unmixed drug solution clumps within the mixed solution, reflecting the purity and degree of mixing; and the drug solution color assists in verifying the drug solution mixing state, visually determining whether the mixing effect meets standards through color uniformity and standard color difference values, avoiding problems such as blurred imaging and poor injection effects caused by uneven drug solution mixing.

[0024] The status data of the pressurizing mechanism 31 collected by the fourth camera C4 is the actual operating speed parameter of the equipment. Data is collected differently according to the structural type of the pressurizing mechanism to accurately match the power output monitoring requirements of different structures. Specifically, when the pressurizing mechanism 31 adopts a push-type structure, its core operating parameter is the actual push-in speed. The real-time push-in rate is calculated by visually capturing the movement stroke and time of the push-in component. When the pressurizing mechanism 31 adopts a wheel-pressurizing structure, its core operating parameter is the actual angular velocity. The real-time operating parameters are collected by visually recognizing the number of rotations and rotation speed of the pressurizing roller, achieving full coverage and accurate monitoring of both mainstream pressurizing structures.

[0025] The fifth camera C5 collects data on the status of cavitation bubbles in the tubing, including three core parameters: the length of a single bubble, the number of bubbles generated per unit time, and the cumulative length of bubbles per unit time. By quantifying bubble size and generation frequency, the system accurately determines the amount of residual air inside the tubing, preventing serious medical accidents such as vascular embolism caused by incomplete venting of the tubing or air intake during equipment operation, which could result in air being injected into the human arteries with the medication. This ensures injection safety.

[0026] This invention addresses the five-channel monitoring cameras integrated into a device, constructing a complete local edge deployment solution for artificial intelligence. It overcomes the drawbacks of traditional cloud-based AI recognition, such as network dependence, high transmission latency, risk of medical image data leakage, and instability and malfunctions in operating room networks. The solution achieves a fully localized closed-loop operation encompassing image acquisition, preprocessing, intelligent recognition, parameter calculation, anomaly detection, and device linkage control. The entire AI system is deployed locally on the high-pressure injector device itself, requiring no external cloud server or network data transmission. It is suitable for medical application scenarios involving closed deployment in interventional operating rooms, complex electromagnetic environments, high real-time requirements, and high security. The system employs embedded edge hardware paired with lightweight dedicated visual AI models, utilizing a unified architecture of local preprocessing, multi-model time-sharing parallel scheduling, and local decision-making linkage to ensure the stability, accuracy, and timeliness of visual monitoring.

[0027] At the hardware deployment and data link level, the device incorporates an industrial-grade embedded edge AI computing chip as the sole inference carrier. This chip boasts low power consumption, high computing power, and strong resistance to electromagnetic interference, making it suitable for the high-frequency electromagnetic interference environment of operating rooms. Five cameras are directly connected to the edge AI chip via independent high-speed MIPI serial ports, creating five parallel, non-interfering local dedicated image transmission links. All real-time video streams and single-frame image data are transmitted, stored, and processed only within the device's internal hardware links, without transmission to the external network or cloud, and without external data sharing. This eliminates the risk of leakage of patient diagnosis data, device medication data, and surgical monitoring data at the hardware link level. Simultaneously, the edge AI chip and the device's main control microcontroller communicate in real-time via a high-speed bus. AI recognition results, anomaly detection commands, and parameter correction signals can be synchronized to the device's main control system in real time, forming a local closed-loop control system of "visual acquisition - AI analysis - device control."

[0028] At the level of model localization and deployment, for the five differentiated visual monitoring tasks of this device, the general deep learning visual model is customized and lightweighted, and optimized for scene adaptation. Through lightweight methods such as model structured pruning, 8-bit fixed-point quantization, redundant operator removal, and network layer fusion, the model size and computing power consumption are significantly reduced while ensuring that the recognition accuracy meets medical monitoring standards, adapting to the limited computing power resources of embedded edge chips. This solution specifically deploys five sets of independent lightweight local AI sub-models: a liquid bottle status recognition model for judging the installation status and remaining liquid volume of liquid bottles; a medicine label OCR recognition model for automatically reading and verifying the specifications and expiration date of medicines; a mixed medicine optical detection model for quantifying the refractive index, transmittance, and color parameters of medicines; a pressurization mechanism speed measurement model for high-precision calculation of equipment operating speed; and a pipeline bubble quantification statistical model for automatically identifying and counting the size and number of bubbles. Each sub-model adopts a time-sharing parallel scheduling mechanism, and all model parameters are permanently stored in the local chip flash memory, automatically loaded at boot, and run independently offline.

[0029] At the local inference and workflow level, the entire AI recognition process is executed entirely locally, without cloud intervention. Raw image frames captured by each camera are first fed into the chip's built-in preprocessing unit, automatically performing preprocessing such as image denoising, exposure correction, distortion correction, and region of interest cropping. This eliminates environmental interference from operating room lighting flicker, equipment obstruction, and uneven lighting, improving recognition stability. The preprocessed image data is then fed into the corresponding lightweight AI sub-model to complete feature extraction, state classification, parameter fitting, and quantization calculations, enabling millisecond-level local inference output of various monitoring parameters and state judgment results. For fault scenarios such as abnormal medication, equipment malfunctions, excessive air bubbles in tubing, and incorrect medication installation during monitoring, the local AI system can directly and autonomously output warning commands and equipment lock signals without cloud-based command interaction. The response speed is at the millisecond level, effectively avoiding surgical safety risks caused by cloud latency, perfectly matching the clinical needs of high-pressure injection—instantaneous, high-precision, and high-safety.

[0030] In terms of system operation and maintenance, data security and environment adaptation, this localized AI solution has the advantages of high stability and compliance. First, it supports local offline iterative upgrade, which can complete model fine-tuning, parameter calibration and version update through the device-specific local debugging port without network connection throughout the whole process, complying with the medical device network security control specifications and avoiding system vulnerabilities and data risks brought by networked upgrade. Second, it has the capability of all-time offline operation. No matter the operating room is in a state of network disconnection, weak network or network fluctuation, it will not affect the functions of visual monitoring, AI reasoning, abnormal warning and process verification, adapting to the complex network environments of various operating rooms. Third, it adopts a local temporary storage mechanism, which only caches the image data required for real-time operation, and automatically clears the temporary data after the single injection process is completed, with no local data retained, further ensuring the security of medical data. Meanwhile, the system has a built-in power consumption adaptive scheduling strategy, which automatically reduces the model computing power consumption under low-load working conditions and runs at full speed under high-load working conditions, reducing the overall power consumption of the device while ensuring the monitoring accuracy, and improving the operation stability and service life of the device.

[0031] Based on the above distributed visual monitoring system, the present invention correspondingly provides a high-pressure syringe monitoring process with full-process step-by-step verification, as shown in Figure 3 , this monitoring process adopts the control logic of step-by-step verification, immediate stop upon abnormality and fault warning. Each process can enter the next process only after passing the verification, so as to avoid various injection risks from the source, which specifically includes the following steps: S1. Pre-verification at the liquid medicine supply end: The first camera C1 and the second camera C2 synchronously collect full-dimensional state data of the contrast agent vial and the saline vial. The system comprehensively judges the installation firmness, remaining liquid sufficiency and liquid medicine label compliance of the two groups of vials, and comprehensively monitors whether the two groups of vials meet the basic supply conditions of the current injection operation. If all the monitored data meet the preset standards, it is determined that the vial supply state is qualified, and the system automatically enters the next monitoring step; if any abnormality such as uninstalled vial, loose installation, insufficient liquid volume, mismatched liquid medicine model, or expired liquid medicine occurs, the system immediately determines that the verification fails, synchronously outputs an acousto-optic error warning, and locks the injection function of the device at the same time, prohibiting the device from starting and running, waiting for manual troubleshooting and rectification.

[0032] S2. Pressurization Power Parameter Verification: After the equipment completes the drug supply verification, the actual operating speed data of the pressurization mechanism 31 is collected in real time through the fourth camera C4. Based on the structural type of the pressurization mechanism, the corresponding speed judgment standard is matched, and the actual injection speed or actual angular velocity is compared with the standard speed parameters preset by the system program. The deviation value is calculated to determine whether the actual operating speed is within the preset error allowable range. If the speed deviation meets the accuracy requirements, the pressurization power output is determined to be stable and qualified, and the next monitoring step is initiated. If the actual speed deviation exceeds the standard, or abnormalities such as excessively fast or slow speed, or stalling occur, the system immediately determines that the verification has failed, outputs an error warning, and suspends equipment operation to avoid problems such as unstable injection pressure, vascular damage, and angiography failure caused by abnormal pressurization speed.

[0033] S3. Drug Mixing Effect Verification: After the power parameters of the pressurizing mechanism are verified as qualified, the refractive index, transmittance, and color optical parameters of the mixed drug solution at the output end of the mixing device 21 are continuously collected by the third camera C3. The system compares the real-time collected optical data with the parameter range of the standard qualified drug solution to comprehensively determine whether the drug solution's proportioning accuracy, mixing uniformity, and purity meet the clinical injection standards. If all optical parameters are within the preset qualified range, the mixing effect is determined to be qualified, and the system proceeds to the next monitoring step. If abnormalities such as deviation in drug solution proportioning concentration, uneven mixing, or the presence of impurities or foreign matter occur, the system immediately triggers an error warning, terminates the injection process, and prevents unqualified drug solution from being injected into the human body.

[0034] S4. Tubing Bubble Safety Verification: After the drug mixing and pressurization parameters pass verification, the fifth camera (C5) collects real-time data on the bubble status within the drug delivery tubing. It calculates the length of a single bubble, the number of bubbles per unit time, and the cumulative bubble length. The measured data is compared with clinical injection safety thresholds to comprehensively monitor residual air within the tubing and determine if the bubble status meets safe injection conditions. If all bubble parameters are below the safety thresholds, the tubing is deemed safe, and the equipment can begin injection normally. If the bubble size, number, or cumulative length exceeds the limits, the system immediately identifies an air injection risk, outputs an error warning, locks the injection process, and reminds staff to vent the tubing to completely avoid the risk of arterial air embolism.

[0035] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

Claims

1. A high-pressure injector based on vision system monitoring, comprising: A first insert structure (11) for installing a contrast agent bottle, a second insert structure (12) for installing a saline bottle, a mixing device (21) for mixing the contrast agent and saline, a pressurizing mechanism (31) for pressurizing the mixed solution, and an injection needle (41) for injecting into a human artery. Its features are, A first camera (C1) is installed on the outside of the first bottle insertion structure (11) to acquire the status data of the contrast agent bottle; A second camera (C2) is installed on the outside of the second bottle insertion structure (12) to acquire the status data of the saline bottle; A third camera (C3) is installed on the outside of the output end of the mixing device (21) to acquire the status data of the mixed liquid; A fourth camera (C4) is installed on the outside of the pressurizing mechanism (31) to acquire status data of the pressurizing mechanism (31); A fifth camera (C5) is installed on the outside of the output end of the pressurizing mechanism (31) to obtain the status data of the liquid bubble in the pipeline.

2. The high-pressure injector based on vision system monitoring according to claim 1, characterized in that, The status data of the contrast agent bottle includes: bottle installation status, bottle volume, and bottle label data.

3. The high-pressure injector based on vision system monitoring according to claim 2, characterized in that, The status data of the saline solution bottle includes: bottle installation status, bottle volume, and bottle label data.

4. The high-pressure injector based on vision system monitoring according to claim 3, characterized in that, The state data of the mixed drug solution includes: refractive index, transmittance, and color.

5. A high-pressure injector based on vision system monitoring according to claim 4, characterized in that, The status data of the pressurizing mechanism (31) includes: actual injection speed or actual angular velocity; When the pressurizing mechanism (31) is a push-type mechanism, the status data is the actual push-in speed; When the pressurizing mechanism (31) is a wheel pressurizing mechanism, the state data is the actual angular velocity.

6. A high-pressure injector based on vision system monitoring according to claim 5, characterized in that, The status data of the pipeline bubbles include: the length of a single bubble, the number of bubbles per unit time, and the cumulative length of bubbles per unit time.

7. A high-pressure injector based on vision system monitoring according to claim 6, characterized in that, The monitoring process includes the following steps: S1. Obtain corresponding status data through the first camera (C1) and the second camera (C2) to monitor whether the liquid bottle meets the injection conditions; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning. S2. Obtain the corresponding status data through the fourth camera (C4) and monitor whether the actual speed meets the preset error with the preset speed of the program; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning. S3. Acquire the corresponding status data through the third camera (C3) to monitor whether the mixing effect meets the injection conditions; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning. S4. Obtain the corresponding status data through the fifth camera (C5) to monitor whether the blister meets the injection conditions; if the monitoring result is satisfactory, proceed to the next step; otherwise, output an error warning.

Images