Ultrasonic vein puncture system integrating image recognition and data analysis

Abstract

The invention relates to the technical field of medical intelligent image recognition data processing, and discloses an image recognition and data analysis fused ultrasonic venipuncture system which comprises a multi-modal image acquisition module, an intelligent analysis processing module, a real-time navigation execution module and a complication early warning module. By arranging a multi-mode fusion sensing end, when vein puncture real-time navigation is carried out, through ultrasonic image, thermodynamic distribution and optical characteristic three-mode data collaborative registration, the consistency of deep blood vessel recognition is guaranteed, meanwhile, a three-dimensional topological model containing blood vessel elastic parameters is dynamically constructed, blood vessel position deviation can be calibrated in real time in the puncture process, and the accuracy of vein puncture is improved. The accuracy of blood vessel positioning is guaranteed, puncture positioning errors of complex cases are further reduced, whether angle or depth deviation occurs in a puncture path or not is judged in real time by arranging a dynamic navigation end, a compensation path can be planned in real time through a blood vessel elastic characteristic matrix, and it is guaranteed that the angle errors are reduced when a needle body enters a blood vessel.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing for medical intelligent image recognition, and in particular to an ultrasonic venipuncture system integrating image recognition and data analysis. Background Art

[0002] Ultrasound examination is non-invasive, repeatable, and accurate, and has been widely used clinically. It can not only detect lesions of solid organs such as the liver, spleen, pancreas, and kidneys at an early stage, but also has obvious advantages in guiding interventional treatment and vascular puncture. Venous puncture is a medical term for patients who require long-term infusion but whose peripheral veins are difficult to puncture due to hardening or collapse, those who require total parenteral nutrition, and emergency treatment of critically ill patients and patients with difficulty in blood collection, as well as central venous pressure measurement.

[0003] There are three major technical bottlenecks in existing assistive technologies: 1. Image fusion defects: Although pure ultrasound navigation can display vascular cross-sections, it cannot simultaneously obtain hemodynamic data, resulting in the neglect of the risk of vascular spasm during puncture; although visible light and infrared imaging are integrated, there is a lack of spatiotemporal registration mechanism, and the vascular thermal map deviates greatly from the visible light vascular position, making it impossible to construct an accurate three-dimensional model.

[0004] 2. Lack of dynamic monitoring. The existing system only statically plans the path before puncture, without considering that the blood vessels are compressed and deformed. When the puncture needle contacts the skin, the average displacement of the superficial veins is large, and the blood flow changes in real time, resulting in the failure of the puncture path.

[0005] 3. Complication warning is delayed. Traditional technology relies on postoperative observation and cannot detect blood vessel wall penetration in real time during the puncture process. Existing image recognition algorithms have delays in detecting interstitial bleeding, missing the optimal treatment window.

[0006] Therefore, the present invention provides an ultrasonic venipuncture system that integrates image recognition and data analysis to fuse multimodal images, track dynamic changes in blood vessels in real time, and have the ability to provide immediate early warning during surgery. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides an ultrasonic venipuncture system that integrates image recognition and data analysis, which solves the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solutions: an ultrasonic venipuncture system integrating image recognition and data analysis, the system comprising a multimodal image acquisition module, an intelligent analysis and processing module, a real-time navigation execution module, and a complication warning module; The multimodal image acquisition terminal sets a blood vessel scanning area and formulates multimodal registration standard parameters, deploys ultrasonic sensor equipment in multiple locations in the target area, performs real-time spatiotemporal registration fusion on the images collected by the multi-source sensors, determines whether there is any modal data deviation in blood vessel positioning, and performs real-time registration correction if data deviation occurs; The intelligent analysis and processing terminal analyzes the vascular topology in real time and obtains the puncture path plan for the corresponding type of blood vessel based on the puncture safety standard parameters. It calculates the matching degree between the current vascular structure characteristics and the standard model to determine whether there is feature deviation in the vascular identification, and reconstructs the 3D model in real time if the vascular characteristics are offset. The real-time navigation execution terminal dynamically tracks the spatial position of the puncture needle, combines a tactile feedback device with an augmented reality display device, and detects the multi-dimensional deviation values ​​of the needle body's motion trajectory in real time, and generates navigation correction instructions in real time according to the deviation values ​​of different dimensions; The complication warning terminal performs graded assessment of puncture risk by real-time monitoring of vascular deformation parameters and hemodynamic indicators, detects vascular collapse rate and bleeding diffusion rate in real time, and triggers a graded blocking mechanism in real time according to different risk levels.

[0009] Preferably, the system comprises the following steps: S1. Synchronously acquire ultrasonic image data, infrared thermal imaging data, and visible light image data of the target area through a multimodal image acquisition device; S2. Performing blood vessel boundary enhancement processing on the ultrasonic image data to generate blood vessel structure enhancement data, and simultaneously performing blood flow heat distribution analysis on the infrared thermal imaging data to generate blood vessel thermal map data; S3, performing spatiotemporal registration and fusion of the vascular structure enhancement data, vascular thermal map data, and visible light image data to construct a multimodal vascular three-dimensional topological model; S4. Extracting the central axis of the target blood vessel in the three-dimensional topological model based on a deep learning blood vessel segmentation algorithm, and generating puncture path planning data in combination with the calculation of the elastic modulus of the blood vessel wall; S5. Real-time collection of puncture needle spatial posture data, dynamic deviation analysis of the data and the puncture path planning data, and generation of puncture navigation correction instructions; S6. Executing the puncture navigation correction instruction through the tactile feedback device and the augmented reality display device, and outputting a puncture angle adjustment parameter and a depth warning signal; S7. During the puncture process, vascular deformation parameters and hemodynamic indicators are continuously monitored. When vascular collapse and blood extravasation are detected, the puncture termination instruction is triggered and a complication warning report is generated.

[0010] Preferably, the S1 includes: S11. Use a 40MHz high-frequency linear ultrasound probe to obtain cross-sectional images of blood vessels within a depth of 5cm below the skin, with a sampling frequency of more than 30 frames per second; S12, collect 14μm band infrared radiation data of the target area through a macro thermal imager, with a spatial resolution of 0.05℃; S13. Use a polarization hyperspectral camera to capture 700-band visible light images and eliminate interference from skin surface reflections.

[0011] Preferably, the S2 includes: S21. Apply the improved U-Net network to segment blood vessels in ultrasound images. The network input layer integrates the prior information of blood vessel orientation preprocessed by Hough transform. S22. Based on the blood flow heat conduction equation Calculate the vascular thermal diffusion characteristics, where is the thermal diffusivity of the tissue, is the blood perfusion rate, is the heat source per unit volume; S23. Correlate the echo intensity of the vascular wall with thermodynamic parameters to construct a vascular elasticity characteristic matrix in It represents the deformation rate of the i-th blood vessel under the j-th pressure level.

[0012] Preferably, the S3 includes: S31. Establish a three-dimensional coordinate system with the puncture point as the origin, and achieve spatial alignment of multi-source images through a feature point matching algorithm, with an alignment error of less than 0.1 mm; S32. Use voxel fusion technology to integrate vascular structure data, thermodynamic data, and epidermal texture data into a unified coordinate system to generate a vascular model with multi-layer physical parameters. in is a multimodal fusion vascular model. is the blood vessel geometric feature set, is the thermodynamic characteristic set, is the optical feature set.

[0013] Preferably, the S4 includes: S41. Calculate the vascular punctureability score in is the blood vessel depth, is the pipe wall elasticity index, is the blood flow velocity, For power Weight coefficient; S42. Select the target blood vessel based on the scoring results and generate the optimal puncture path using B-spline curve fitting in is the control point, is the basis function.

[0014] Preferably, the S5 includes: S51, obtain the six-degree-of-freedom position data of the puncture needle through the needle body embedded MEMS sensor is the three-dimensional rectangular coordinate of the needle tip, The rotation angle of the needle body around the X, Y, and Z axes; S52. Dynamically adjust path deviation using the egret foraging optimization algorithm: S521, initialize the Bailong population position is the search radius; S522, exploration phase simulates egret stride search, position update formula in is the current position of the i-th candidate solution, 、 are the path attraction and group cooperation coefficients, is the current optimal solution position; S523, Development stage simulation precision pecking, position update formula in is the local search intensity factor; S524, output the posture correction value corresponding to the minimum deviation value .

[0015] Preferably, the S7 includes: S71. Monitoring the rate of change of blood vessel diameter by ultrasound Doppler Determine blood vessels collapse; S72, based on the YOLOv5s architecture training blood seepage detection model, when the interstitial fluid dark area expansion rate is identified When triggering an early warning; S73. Establish a complication risk matrix in It represents the strength of association between the jth type of signs and the jth type of complications.

[0016] Preferably, it also includes: S8. Build an adaptive learning system: S81, storing path planning data, real-time navigation data, and puncture result data for each puncture; S82. Use the Transformer network to analyze historical data and generate a correlation map between vascular characteristics and puncture success rate; S83. When the amount of new data reaches a threshold value N, the weight parameters of the path planning algorithm are automatically updated.

[0017] Compared with the existing technology, the present invention provides an ultrasonic venipuncture system that integrates image recognition and data analysis, which has the following beneficial effects: 1. In the present invention, by setting up a multimodal fusion sensing terminal, when performing real-time navigation for venipuncture, the consistency of deep blood vessel identification is ensured through the coordinated registration of three-modal data: ultrasound imaging, thermodynamic distribution, and optical characteristics. At the same time, a three-dimensional topological model containing vascular elastic parameters is dynamically constructed, which can calibrate the vascular position offset in real time during the puncture process, ensure the accuracy of vascular positioning, and further reduce the puncture positioning error in complex cases. 2. In the present invention, by setting a dynamic navigation terminal, when performing real-time venipuncture operations, the needle posture deviation value is analyzed based on the Egret optimization algorithm, and it is judged in real time whether the puncture path has angle or depth deviation, so that the system can immediately adjust the tactile and visual dual-mode feedback signals, and when the path deviates, the compensation path can be planned in real time through the vascular elasticity characteristic matrix to ensure that the angle error of the needle entering the blood vessel is reduced. 3. In the present invention, by setting up a risk warning terminal, when monitoring the entire process of venipuncture, the blood vessel deformation and bleeding signs are automatically detected in a graded manner at the key stage of puncture, and the multi-level complication risk coefficient is output in real time. The graded blocking mechanism is dynamically activated according to the collapse rate and bleeding expansion rate, so that the system can implement differentiated warning measures for different risk levels, reduce the missed reporting rate of puncture complications, avoid false alarms and interruptions of operations, and further improve the success rate and safety of complex blood vessel puncture. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the overall system architecture of the present invention.

[0019] Figure 2 Flow chart of the method of the present invention.

[0020] Figure 3 This is a flow chart of the subsequent steps of S1 of the present invention.

[0021] Figure 4 This is a flow chart of the subsequent steps of S2 of the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] See also Figures 1 to 4 The ultrasonic venipuncture system integrates image recognition and data analysis. The system includes a multimodal image acquisition module, an intelligent analysis and processing module, a real-time navigation execution module, and a complication warning module. The multimodal image acquisition terminal sets a blood vessel scanning area and formulates multimodal registration standard parameters, deploys ultrasonic sensor equipment in multiple locations in the target area, performs real-time spatiotemporal registration fusion on the images collected by the multi-source sensors, determines whether there is any modal data deviation in blood vessel positioning, and performs real-time registration correction if data deviation occurs; The intelligent analysis and processing terminal analyzes the vascular topology in real time and obtains the puncture path plan for the corresponding type of blood vessel based on the puncture safety standard parameters. It calculates the matching degree between the current vascular structure characteristics and the standard model to determine whether there is feature deviation in the vascular identification, and reconstructs the 3D model in real time if the vascular characteristics are offset. The real-time navigation execution terminal dynamically tracks the spatial position of the puncture needle, combines a tactile feedback device with an augmented reality display device, and detects the multi-dimensional deviation values ​​of the needle body's motion trajectory in real time, and generates navigation correction instructions in real time according to the deviation values ​​of different dimensions; The complication warning terminal performs graded assessment of puncture risk by real-time monitoring of vascular deformation parameters and hemodynamic indicators, detects vascular collapse rate and bleeding diffusion rate in real time, and triggers a graded blocking mechanism in real time according to different risk levels.

[0024] The AR puncture navigation glasses use waveguide optical display technology to superimpose the display in the field of view: Semi-transparent vascular 3D model; Real-time puncture needle trajectory; Safety needle insertion depth warning ring; Vascular elasticity thermogram.

[0025] The system includes the following steps: S1. Synchronously acquire ultrasonic image data, infrared thermal imaging data, and visible light image data of the target area through a multimodal image acquisition device; S2. Performing blood vessel boundary enhancement processing on the ultrasonic image data to generate blood vessel structure enhancement data, and simultaneously performing blood flow heat distribution analysis on the infrared thermal imaging data to generate blood vessel thermal map data; S3, performing spatiotemporal registration and fusion of the vascular structure enhancement data, vascular thermal map data, and visible light image data to construct a multimodal vascular three-dimensional topological model; S4. Extracting the central axis of the target blood vessel in the three-dimensional topological model based on a deep learning blood vessel segmentation algorithm, and generating puncture path planning data in combination with the calculation of the elastic modulus of the blood vessel wall; S5. Real-time collection of puncture needle spatial posture data, dynamic deviation analysis of the data and the puncture path planning data, and generation of puncture navigation correction instructions; S6. Executing the puncture navigation correction instruction through the tactile feedback device and the augmented reality display device, and outputting a puncture angle adjustment parameter and a depth warning signal; S7. During the puncture process, vascular deformation parameters and hemodynamic indicators are continuously monitored. When vascular collapse and blood extravasation are detected, the puncture termination instruction is triggered and a complication warning report is generated.

[0026] S1 includes: S11. Use a 40MHz high-frequency linear ultrasound probe to obtain cross-sectional images of blood vessels within a depth of 5cm below the skin, with a sampling frequency of more than 30 frames per second; S12, collect 14μm band infrared radiation data of the target area through a macro thermal imager, with a spatial resolution of 0.05℃; S13. Use a polarization hyperspectral camera to capture 700-band visible light images and eliminate interference from skin surface reflections.

[0027] S2 includes: S21. Apply the improved U-Net network to segment blood vessels in ultrasound images. The network input layer integrates the prior information of blood vessel orientation preprocessed by Hough transform. S22. Based on the blood flow heat conduction equation Calculate the vascular thermal diffusion characteristics, where is the thermal diffusivity of the tissue, is the blood perfusion rate, is the heat source per unit volume; S23. Correlate the echo intensity of the vascular wall with thermodynamic parameters to construct a vascular elasticity characteristic matrix in It represents the deformation rate of the i-th blood vessel under the j-th pressure level.

[0028] S3 includes: S31. Establish a three-dimensional coordinate system with the puncture point as the origin, and achieve spatial alignment of multi-source images through a feature point matching algorithm, with an alignment error of less than 0.1 mm; S32. Use voxel fusion technology to integrate vascular structure data, thermodynamic data, and epidermal texture data into a unified coordinate system to generate a vascular model with multi-layer physical parameters. in is a multimodal fusion vascular model. is the blood vessel geometric feature set, is the thermodynamic characteristic set, is the optical feature set.

[0029] S4 includes: S41. Calculate the vascular punctureability score in is the blood vessel depth, is the pipe wall elasticity index, is the blood flow velocity, For power Weight coefficient; S42. Select the target blood vessel based on the scoring results and generate the optimal puncture path using B-spline curve fitting in is the control point, is the basis function.

[0030] S5 includes: S51, obtain the six-degree-of-freedom position data of the puncture needle through the needle body embedded MEMS sensor is the three-dimensional rectangular coordinate of the needle tip, The rotation angle of the needle body around the X, Y, and Z axes; S52. Dynamically adjust path deviation using the egret foraging optimization algorithm: S521, initialize the Bailong population position is the search radius; S522, exploration phase simulates egret stride search, position update formula in is the current position of the i-th candidate solution, 、 are the path attraction and group cooperation coefficients, is the current optimal solution position; S523, Development stage simulation precision pecking, position update formula in is the local search intensity factor; S524, output the posture correction value corresponding to the minimum deviation value .

[0031] S7 includes: S71. Monitoring the rate of change of blood vessel diameter by ultrasound Doppler Determine blood vessels collapse; S72, based on the YOLOv5s architecture training blood seepage detection model, when the interstitial fluid dark area expansion rate is identified When triggering an early warning; S73. Establish a complication risk matrix in It represents the strength of association between the jth type of signs and the jth type of complications.

[0032] Also includes: S8. Build an adaptive learning system: S81, storing path planning data, real-time navigation data, and puncture result data for each puncture; S82. Use the Transformer network to analyze historical data and generate a correlation map between vascular characteristics and puncture success rate; S83. When the amount of new data reaches a threshold value N, the weight parameters of the path planning algorithm are automatically updated. First embodiment

[0033] S1 multimodal imaging data acquisition The operator secures the patient's forearm to a constant-temperature operating platform and activates the multimodal acquisition system. A high-frequency ultrasound probe contacts the skin surface at a 30-degree angle, transmitting 40MHz ultrasound waves through a sterile coupling agent to capture real-time cross-sectional images of blood vessels at a depth of 0-5cm. A synchronously activated macro thermal imager collects infrared radiation in the 8-14μm band at a distance of 20cm, generating a thermal distribution map with a temperature resolution of 0.03°C. A polarized hyperspectral camera vertically focuses on the target area, collecting 128 bands of data at 5nm intervals within the 470-1000nm spectral range, automatically eliminating interference from skin surface reflections. The three devices use a hardware synchronization module to ensure temporal and spatial consistency, generating 30 registration data packets per second.

[0034] S2 intelligent processing of vascular features The data processing platform performs multi-level analysis on the collected information. Ultrasound images are first enhanced with a Frangi filter to enhance vascular edge features. These features are then fed into a modified U-Net network for 3D segmentation. This network incorporates a spatial attention mechanism to improve the accuracy of small vessel identification. Thermodynamic data are then used to calculate blood flow distribution using the biological heat conduction equation, and a model is established to model the relationship between the temperature field and vascular elastic strain. Hyperspectral data are then decomposed using characteristic spectra to extract the characteristic hemoglobin absorption peak at 525 nm and calculate blood oxygen saturation parameters. The final output is a 12-dimensional vascular feature vector containing vessel diameter, elastic modulus, and blood flow velocity.

[0035] S3 multi-source fusion 3D modeling The system establishes a three-dimensional coordinate system centered on the puncture target and spatially aligns multi-source data using a feature point matching algorithm. Voxel fusion technology integrates ultrasound geometry, thermodynamic distribution, and optical features into a unified model, constructing a dynamic vascular topology with physical properties. This model incorporates a layered update mechanism: the geometric layer updates vascular morphology at a 10Hz frequency, the thermodynamic layer updates temperature distribution at 5Hz, and the optical layer adjusts color characteristics at 2Hz. The real-time modeling process automatically compensates for tissue displacement caused by breathing and muscle micro-movements.

[0036] S4 dynamic puncture path planning The intelligent decision-making system assesses puncture feasibility based on vascular characteristics. Parameters such as target vessel depth, wall elastic modulus, and real-time blood flow velocity are extracted, and a weighted scoring algorithm is used to calculate the puncture safety factor. If the score exceeds the safety threshold, a B-spline curve algorithm is used to generate a three-dimensional puncture path, planning the coordinates of the epidermal entry point and the target vessel. The path planning results are projected into the operator's field of view through augmented reality glasses, visually displaying a green safe channel and a red dangerous zone. The safe channel diameter dynamically adjusts to the vessel's elasticity.

[0037] S5 real-time navigation execution control The puncture is performed using a specially designed navigation needle embedded with a six-dimensional MEMS sensor that provides real-time spatial position feedback. The system dynamically corrects path deviations using an egret foraging optimization algorithm. During the exploration phase, it simulates a wide-area search to rapidly approach the target path; during the development phase, it simulates precise pecking movements for millimeter-level fine-tuning. The tactile feedback glove triggers different vibration patterns depending on the type of deviation. The thumb area generates low-frequency continuous vibrations during horizontal yaw, the index finger joint receives high-frequency pulses when the needle is inserted too quickly, and a three-beat warning is emitted from the palm of the hand when approaching the blood vessel wall.

[0038] S6 Complications Active Defense During the operation, a multi-channel monitoring mechanism is established, and the ultrasonic Doppler continuously tracks the rate of change of blood vessel diameter. When the instantaneous collapse exceeds the safety threshold, a level one alarm is triggered. The bleeding recognition model based on the YOLOv5s architecture analyzes the expansion characteristics of the dark area of ​​the interstitial fluid in real time. When abnormal diffusion is detected, the operation is immediately terminated. The risk warning system implements a graded response: low-risk events are visually warned through the AR interface; high-risk conditions simultaneously trigger the tactile glove operation lock and the system buzzer alarm, forming a physical blocking mechanism.

[0039] S7 system self-evolution mechanism After each puncture is completed, the full-cycle operation data is automatically archived to the knowledge base; when the case accumulation reaches the set threshold, the system starts the adaptive learning engine; through the deep neural network analysis of the mapping relationship between vascular characteristics and operation results, the path planning algorithm weight parameters are dynamically optimized; the continuous learning mechanism enables the system to improve its ability to handle complex cases after multiple uses, and the puncture success rate for patients with venous sclerosis has increased by 65%.

[0040] This embodiment achieves full-process intelligence from blood vessel identification to safe puncture through closed-loop control of multimodal perception, dynamic decision-making and execution feedback; clinical verification shows that the system has a first-time puncture success rate of 98.2% for special groups such as obese patients and shock patients, the complication rate is reduced to 0.8%, and the average operation time is shortened by 40%.

[0041] Second embodiment,

[0042] S1 Multi-modality Image Synchronous Acquisition The patient is positioned in a Trendelenburg position, and the operator uses a sterile protective cover to enclose the multimodal acquisition module. A low-frequency convex array ultrasound probe contacts the neck skin at a 60° angle, adjusting the beam penetration depth to 8 cm to capture the anatomical relationship between the internal jugular vein and carotid artery. A long-wave infrared thermal imager scans at an angle of 30 cm from the body surface to detect differences in thermal radiation from deep vessels. A hyperspectral imager collects tissue spectral signatures in the 950nm near-infrared band, penetrating superficial tissue to identify deep veins.

[0043] S2 deep vascular feature extraction The data processing platform performs special algorithm optimization: Ultrasound imaging uses composite filtering technology to eliminate carotid artery pulsation artifacts and distinguishes venous lumens through dynamic threshold segmentation The thermodynamic model introduces a depth compensation algorithm, and the correction formula is: in for , ; The spectral data analysis used the hemoglobin differential absorption method, and the relative difference in venous oxygen saturation ΔSpO2 greater than 15% was calculated as a venous characteristic marker; S3 Hazardous Area Modeling Systematically build a multi-layered anatomical risk model: Define the dangerous structure warning zone: the 3mm radius of the carotid artery projection is set as a red restricted area; A pleural apex position prediction model was established to calculate respiratory displacement trajectory based on clavicle anatomical landmarks; Fuse CT image prior data to mark the vagus nerve probable hotspots; S4 safe puncture path planning Intelligent navigation system performs: Needle insertion point optimization algorithm: Based on the apex of the sternocleidomastoid muscle triangle, the optimal puncture angle α=35+5° is dynamically calculated; Real-time risk avoidance mechanism: The planned path automatically bypasses the carotid artery projection area, with a minimum safety distance set at 2.5mm; Pressure sensing warning: The puncture needle is equipped with a micro pressure sensor to monitor the breakthrough feeling in real time; S5 real-time dynamic navigation Operation process monitoring: The electromagnetic positioning system tracks the three-dimensional trajectory of the puncture needle with an accuracy of 0.3mm; Augmented reality glasses project multi-layered navigation information: Real-time location of the internal jugular vein; Planning the needle insertion path; Dangerous structure warnings; The tactile feedback system provides graded warnings: the tactile gloves trigger low-frequency vibrations when approaching the carotid artery, and generate reverse resistance when entering the danger zone. S6 Active prevention and control of complications Establish a multi-level defense system: Pneumothorax warning: monitors the distance between the needle tip and the pleural apex, and automatically locks the needle insertion mechanism when the respiratory displacement causes the distance to be less than 1mm; Response to misplaced artery puncture: Real-time comparison of spectral characteristics, identification of arterial blood characteristics, and immediate withdrawal of the needle; Nerve damage prevention: EMG monitoring patch detects vagal reflexes, and a sudden drop in heart rate >20 beats / minute triggers an audible and visual alarm; S7 emergency response support System integration emergency mechanism: When an artery is accidentally punctured, compression point navigation is automatically activated, and AR glasses project the optimal compression position; After high-risk pneumothorax procedures, an ultrasound scan plan is automatically generated to locate the pleural line; The knowledge base instantly pushes complication management guidelines and displays drug dosage calculation templates; This embodiment was verified in 42 critically ill patients, with a first-time puncture success rate of 96.3% and a critical complication rate of 0. The system achieved millimeter-level precision control of deep vascular puncture, shortening the median operation time to 3 minutes and 15 seconds, and maintaining a 94.1% success rate, especially in hypotensive patients.

[0044] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0045] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An ultrasonic venipuncture system integrating image recognition and data analysis, characterized by: The system includes a multimodal image acquisition module, an intelligent analysis and processing module, a real-time navigation execution module, and a complication warning module; The multimodal image acquisition terminal sets a blood vessel scanning area and formulates multimodal registration standard parameters, deploys ultrasonic sensor equipment in multiple locations in the target area, performs real-time spatiotemporal registration fusion on the images collected by the multi-source sensors, determines whether there is any modal data deviation in blood vessel positioning, and performs real-time registration correction if data deviation occurs; The intelligent analysis and processing terminal analyzes the vascular topology in real time and obtains the puncture path plan for the corresponding type of blood vessel based on the puncture safety standard parameters. It calculates the matching degree between the current vascular structure characteristics and the standard model to determine whether there is feature deviation in the vascular identification, and reconstructs the 3D model in real time if the vascular characteristics are offset. The real-time navigation execution terminal dynamically tracks the spatial position of the puncture needle, combines a tactile feedback device with an augmented reality display device, and detects the multi-dimensional deviation values ​​of the needle body's motion trajectory in real time, and generates navigation correction instructions in real time according to the deviation values ​​of different dimensions; The complication warning terminal performs graded assessment of puncture risk by real-time monitoring of vascular deformation parameters and hemodynamic indicators, detects vascular collapse rate and bleeding diffusion rate in real time, and triggers a graded blocking mechanism in real time according to different risk levels.

2. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The tactile feedback glove is provided with an 8×8 piezoelectric vibration array on the back of the hand, and the vibration pattern is associated with the deviation direction of the puncture needle; The AR puncture navigation glasses use waveguide optical display technology to superimpose the display in the field of view: Semi-transparent 3D blood vessel model; Real-time puncture needle trajectory; Safety needle insertion depth warning ring; Vascular elasticity thermogram.

3. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The system comprises the following steps: S1. Synchronously acquire ultrasonic image data, infrared thermal imaging data, and visible light image data of the target area through a multimodal image acquisition device; S2. Performing blood vessel boundary enhancement processing on the ultrasonic image data to generate blood vessel structure enhancement data, and simultaneously performing blood flow heat distribution analysis on the infrared thermal imaging data to generate blood vessel thermal map data; S3, performing spatiotemporal registration and fusion of the vascular structure enhancement data, vascular thermal map data, and visible light image data to construct a multimodal vascular three-dimensional topological model; S4. Extracting the central axis of the target blood vessel in the three-dimensional topological model based on a deep learning blood vessel segmentation algorithm, and generating puncture path planning data in combination with the calculation of the elastic modulus of the blood vessel wall; S5. Real-time collection of puncture needle spatial posture data, dynamic deviation analysis of the data and the puncture path planning data, and generation of puncture navigation correction instructions; S6. Executing the puncture navigation correction instruction through the tactile feedback device and the augmented reality display device, and outputting a puncture angle adjustment parameter and a depth warning signal; S7. During the puncture process, vascular deformation parameters and hemodynamic indicators are continuously monitored. When vascular collapse or blood extravasation is detected, the puncture termination instruction is triggered and a complication warning report is generated.

4. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: Said S1 comprises: S11. Use a 40MHz high-frequency linear ultrasound probe to obtain cross-sectional images of blood vessels within a depth of 5cm below the skin, with a sampling frequency of more than 30 frames per second; S12, collect 14μm band infrared radiation data of the target area through a macro thermal imager, with a spatial resolution of 0.05℃; S13. Use a polarization hyperspectral camera to capture 700-band visible light images and eliminate interference from skin surface reflections.

5. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The S2 includes: S21. Apply the improved U-Net network to segment blood vessels in ultrasound images. The network input layer integrates the prior information of blood vessel orientation preprocessed by Hough transform. S22. Based on the blood flow heat conduction equation Calculate the vascular thermal diffusion characteristics, where is the thermal diffusivity of the tissue, is the blood perfusion rate, is the heat source per unit volume; S23. Correlate the echo intensity of the vascular wall with thermodynamic parameters to construct a vascular elasticity characteristic matrix in It represents the deformation rate of the i-th blood vessel under the j-th pressure level.

6. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The S3 includes: S31. Establish a three-dimensional coordinate system with the puncture point as the origin, and achieve spatial alignment of multi-source images through a feature point matching algorithm, with an alignment error of less than 0.1 mm; S32. Use voxel fusion technology to integrate vascular structure data, thermodynamic data, and epidermal texture data into a unified coordinate system to generate a vascular model with multi-layer physical parameters. in is a multimodal fusion vascular model. is the blood vessel geometric feature set, is the thermodynamic characteristic set, is the optical feature set.

7. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The S4 includes: S41. Calculate the vascular punctureability score in is the blood vessel depth, is the pipe wall elasticity index, is the blood flow velocity, For power Weight coefficient; S42. Select the target blood vessel based on the scoring results and generate the optimal puncture path using B-spline curve fitting in is the control point, is the basis function.

8. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The S5 includes: S51, obtain the six-degree-of-freedom position data of the puncture needle through the needle body embedded MEMS sensor is the three-dimensional rectangular coordinate of the needle tip, The rotation angle of the needle body around the X, Y, and Z axes; S52. Dynamically adjust path deviation using the egret foraging optimization algorithm: S521, initialize the Bailong population position is the search radius; S522, exploration phase simulates egret stride search, position update formula in is the current position of the i-th candidate solution, 、 are the path attraction and group cooperation coefficients, is the current optimal solution position; S523, Development stage simulation precision pecking, position update formula in is the local search intensity factor; S524, output the posture correction value corresponding to the minimum deviation value .

9. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: The S7 includes: S71. Monitoring the rate of change of blood vessel diameter by ultrasound Doppler Determine blood vessels collapse; S72, based on the YOLOv5s architecture training blood seepage detection model, when the interstitial fluid dark area expansion rate is identified When triggering an early warning; S73. Establish a complication risk matrix in It represents the strength of association between the jth type of signs and the jth type of complications.

10. The ultrasonic venipuncture system integrating image recognition and data analysis according to claim 1, characterized in that: Also includes: S8. Build an adaptive learning system: S81, storing path planning data, real-time navigation data, and puncture result data for each puncture; S82. Use the Transformer network to analyze historical data and generate a correlation map between vascular characteristics and puncture success rate; S83. When the amount of new data reaches a threshold value N, the weight parameters of the path planning algorithm are automatically updated.

Images