High-precision automatic venous blood sampling system and method

Abstract

The invention relates to the technical field of medical devices, and specifically discloses a high-precision automatic venous blood sampling system and method, wherein the system comprises: an optimal puncture vessel decision module determines an optimal puncture vessel data set; a dynamic ultrasound data acquisition module guides an ultrasound probe on a mechanical arm to perform skin-fitting ultrasound imaging based on the optimal puncture vessel coordinates in a camera coordinate system to obtain dynamic ultrasound data of an optimal puncture area; an ultrasound vessel data acquisition module enhances the vessel signal in the dynamic ultrasound data and extracts ROI; an ultrasound vessel data analysis module extracts all potential vessel morphology distribution data sets from a time-series vessel ROI candidate set, and performs puncture preference and credibility analysis on all potential vessel morphology distribution data sets to determine a final needle insertion point and a final puncture angle; and a venous blood sampling puncture control module is used to control the puncture needle to perform puncture and blood sampling at a final puncture position at a final puncture angle; and is used to realize high-precision automatic control of venous blood sampling.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and in particular to a high-precision automatic venous blood sampling system and method. Background Art

[0002] At present, the high-precision automatic venous blood collection system and method is a means of integrating advanced sensing technology, image recognition technology and automatic control technology to achieve efficient, accurate and safe venous blood collection operations. The system and method are usually equipped with high-precision sensors to detect information such as the location, depth and diameter of the blood vessels, and use advanced image recognition algorithms to accurately analyze the collected blood vessel images to determine the best blood collection point. The blood collection process is completed by an automated mechanical device that can accurately control the insertion depth and angle of the blood collection needle to ensure smooth blood collection.

[0003] However, in actual applications, due to the large individual differences in human physiological structure, such as obesity level, variation in blood vessel direction, etc., as well as the complexity of vascular conditions under different disease states, such as vascular sclerosis, inflammatory response, etc., it is difficult to accurately identify the appropriate vein location. In the past, when selecting the puncture blood vessel area, it mainly relied on the experience and subjective judgment of medical staff, which was prone to inaccurate and unscientific selection defects. Or rely on rough estimates or fixed patterns to determine the needle insertion point and puncture angle, resulting in high puncture risks and poor results. Defects such as unscientific selection of puncture blood vessel areas and inaccurate determination of needle insertion points and puncture angles will lead to inaccurate blood collection results, greatly affecting the patient's medical experience.

[0004] Therefore, the present invention proposes a high-precision automatic venous blood sampling system and method. Summary of the invention

[0005] The present invention provides a high-precision automatic venous blood sampling system and method, which is used to determine the best puncture vessel data set in the infrared vascular image using a puncture decision model, thereby improving the scientificity and accuracy of puncture vessel selection. The ultrasound probe imaging is guided by the best puncture vessel coordinates in the camera coordinate system to obtain dynamic ultrasound data, providing more detailed and accurate subcutaneous vascular distribution information for subsequent analysis. The adaptive threshold segmentation method is used to construct a time-series vascular ROI candidate set, highlighting the vascular characteristics for subsequent processing. The potential vascular morphology distribution data set is analyzed to determine the final needle insertion point and puncture angle, ensuring the safety and effectiveness of the puncture. Precise control of puncture blood sampling is achieved, the success rate and quality of blood sampling are improved, and the patient's medical experience is also improved.

[0006] The present invention provides a high-precision automatic venous blood sampling system, comprising:

[0007] An optimal puncture vessel decision module is used to determine an optimal puncture vessel data set in an infrared vessel image based on a puncture decision model;

[0008] A dynamic ultrasound data acquisition module is used to determine the optimal puncture vessel coordinates in the camera coordinate system based on the optimal puncture vessel data set, and to guide the ultrasound probe on the robotic arm to perform skin-fitting ultrasound imaging based on the optimal puncture vessel coordinates in the camera coordinate system to obtain dynamic ultrasound data of the optimal puncture area;

[0009] Ultrasonic vascular data acquisition module, used to enhance vascular signals and extract ROI in dynamic ultrasound data based on an adaptive threshold segmentation method, and construct a time-series vascular ROI candidate set;

[0010] Ultrasonic vascular data analysis module, used to extract all potential vascular morphology distribution data sets from the time-series vascular ROI candidate set, and perform puncture preference and credibility analysis on all potential vascular morphology distribution data sets to determine the final needle insertion point and final puncture angle;

[0011] The venous blood sampling puncture control module is used to control the puncture needle to perform puncture and blood sampling at the final puncture angle and at the final puncture position to obtain the blood sampling control result.

[0012] Preferably, the optimal vascular puncture decision module includes:

[0013] The infrared blood vessel image extraction submodule is used to extract the infrared blood vessel image from the high-resolution infrared image of the target venous blood sampling site, and identify multiple candidate puncture blood vessel area data sets in the infrared blood vessel image;

[0014] The optimal puncture vessel decision submodule is used to screen out the optimal puncture vessel data set from all candidate puncture vessel region data sets based on the puncture decision model.

[0015] Preferably, the infrared blood vessel image extraction submodule includes:

[0016] A venous infrared imaging unit is used to perform infrared imaging of the target venous blood collection site to obtain a high-resolution infrared image of the target venous blood collection site;

[0017] A blood vessel image extraction unit, used to extract an infrared blood vessel image from a high-resolution infrared image of a target venous blood sampling site;

[0018] A blood vessel image enhancement unit, used for enhancing the infrared blood vessel image to obtain an enhanced blood vessel image;

[0019] The candidate puncture region extraction unit is used to segment the infrared blood vessel image using a morphology-based medical image segmentation method to obtain multiple candidate puncture blood vessel region data sets.

[0020] Preferably, the dynamic ultrasound data acquisition module includes:

[0021] The dual-camera posture conversion submodule is used to generate the three-dimensional coordinates of the best puncture vessel in the best puncture vessel data set through depth imaging and infrared-depth image alignment, and determine the best puncture vessel coordinates in the camera coordinate system based on the three-dimensional coordinates of the best puncture vessel;

[0022] The skin-fitting ultrasound imaging submodule is used to guide the ultrasound probe on the robotic arm to continuously capture multiple frames of effective fitting ultrasound images of the optimal puncture area based on the optimal puncture vessel coordinates in the camera coordinate system, and obtain dynamic ultrasound data of the optimal puncture area.

[0023] Preferably, the skin-fitting ultrasound imaging submodule comprises:

[0024] A signal-to-noise ratio evaluation unit, used to guide the ultrasound probe on the robot arm to continuously capture multiple frames of ultrasound images of the optimal puncture area based on the optimal puncture blood vessel coordinates in the camera coordinate system, and evaluate the signal-to-noise ratio of each frame of the ultrasound image;

[0025] An ultrasound image segmentation unit, used for identifying all valid lamination areas and all invalid lamination areas in each frame of ultrasound image based on a feature recognition algorithm;

[0026] An imaging quality scoring unit, used to input the signal-to-noise ratio, all effective lamination areas, and all invalid lamination areas of each frame of ultrasound image into an imaging quality scoring model to obtain an imaging quality scoring value for each frame of ultrasound image;

[0027] The dynamic ultrasound data acquisition unit is used to obtain dynamic ultrasound data of the optimal puncture area based on the imaging quality score values ​​of all frames of ultrasound images.

[0028] Preferably, the dynamic ultrasound data acquisition unit comprises:

[0029] The first judgment subunit is used to judge whether the imaging quality score value of each frame of ultrasound image reaches the expected score value. If so, the corresponding frame of ultrasound image is regarded as a valid fitting ultrasound image of the corresponding capture position. Otherwise, the posture adjustment path of the ultrasound probe to the corresponding capture position is determined based on the distribution positions of all valid fitting areas and all invalid fitting areas in the corresponding frame of ultrasound image.

[0030] The second judgment subunit is used to adjust the path of the ultrasound probe to the corresponding capture position, and recapture the ultrasound image at the corresponding capture position until the imaging quality score value of the latest captured ultrasound image reaches the expected score value, and then the latest captured ultrasound image is regarded as the effective fitting ultrasound image of the corresponding capture position;

[0031] The dynamic ultrasound data fitting subunit is used to sort the effective fitting ultrasound images of all capture positions according to the path sequence of all capture positions in the original capture path, so as to obtain the dynamic ultrasound data of the best puncture area.

[0032] Preferably, the first determination subunit includes:

[0033] The score comparison end is used to determine whether the imaging quality score of each frame of ultrasound image reaches the expected score value;

[0034] The first output terminal is used to treat the corresponding frame of ultrasound image as a valid fitting ultrasound image of the corresponding capture position when the imaging quality score value of each frame of ultrasound image reaches the expected score value;

[0035] The image analysis end is used to determine the physical center positions of all valid lamination areas and the physical center positions of all invalid lamination areas in the corresponding frame of ultrasound image when the imaging quality score value of each frame of ultrasound image does not reach the expected score value;

[0036] The adjustment vector determination end is used to determine a first two-dimensional adjustment vector pointing from the physical center position of all valid lamination areas in the corresponding frame ultrasound image to the physical center position of all invalid lamination areas, and at the same time, determine a second two-dimensional adjustment vector pointing from the physical center position of the corresponding ultrasound image to the physical center position of all valid lamination areas in the corresponding frame ultrasound image, and at the same time, determine a third two-dimensional adjustment vector pointing from the physical center position of all invalid lamination areas in the corresponding frame ultrasound image to the physical center position of the corresponding ultrasound image;

[0037] The adjustment path generating end is used to generate a posture adjustment path of the ultrasound probe to the corresponding capture position based on the first two-dimensional adjustment vector, the second two-dimensional adjustment vector, and the third two-dimensional adjustment vector of the corresponding frame ultrasound image.

[0038] Preferably, the ultrasound vascular data analysis module includes:

[0039] The vascular positioning and morphological fitting submodule is used to perform rough positioning and morphological fitting of vascular signals in the time-series vascular ROI candidate set based on the feature extraction method to obtain all potential vascular morphological distribution data sets;

[0040] The vascular credibility analysis submodule is used to perform puncture optimization and credibility analysis on all potential vascular morphology distribution data sets to obtain the final needle insertion point and the corresponding vascular morphology data of the local inspection area;

[0041] The final needle insertion strategy decision submodule is used to determine the final puncture angle based on the final vascular morphology data of the local inspection area corresponding to the final needle insertion point.

[0042] Preferably, the blood vessel credibility analysis submodule includes:

[0043] A blood vessel morphology distribution data acquisition unit is used to determine all candidate blood vessel puncture positions in the optimal puncture area, and determine all blood vessel morphology distribution data of the local inspection area of ​​each candidate blood vessel puncture position in all potential blood vessel morphology distribution data sets;

[0044] A puncture preference evaluation unit, configured to evaluate the puncture preference of each candidate blood vessel puncture position based on all blood vessel morphology distribution data in the local inspection area of ​​all candidate blood vessel puncture positions;

[0045] A cluster analysis calculation unit, used to treat all the blood vessel morphology distribution data in the local inspection area of ​​each candidate blood vessel puncture position as a cluster, and calculate the credibility of the blood vessel morphology distribution data of each cluster;

[0046] A final decision preference calculation unit, used to calculate the final decision preference of each candidate vascular puncture position based on the puncture preference of each candidate vascular puncture position and the credibility of the vascular morphology distribution data of the corresponding cluster;

[0047] The puncture position final decision unit is used to take the candidate vascular puncture position with the maximum puncture final decision preference among all candidate vascular puncture positions as the final needle insertion point, and determine the final vascular morphology distribution data of the corresponding local inspection area based on all vascular morphology distribution data of the local inspection area of ​​the candidate vascular puncture position with the maximum puncture final decision preference.

[0048] The present invention provides a high-precision automatic venous blood sampling method, which is applied to any of the above high-precision automatic venous blood sampling systems, comprising:

[0049] S1: Based on the puncture decision model, the optimal puncture vessel dataset is determined in the infrared vascular image;

[0050] S2: Determine the optimal puncture vessel coordinates in the camera coordinate system based on the optimal puncture vessel data set, and guide the ultrasound probe on the robotic arm to perform skin-fitting ultrasound imaging based on the optimal puncture vessel coordinates in the camera coordinate system to obtain dynamic ultrasound data of the optimal puncture area;

[0051] S3: Based on the adaptive threshold segmentation method, the vascular signals in the dynamic ultrasound data are enhanced and the ROI is extracted to construct a candidate set of time-series vascular ROIs;

[0052] S4: extract all potential vascular morphology distribution data sets from the time-series vascular ROI candidate set, perform puncture preference and credibility analysis on all potential vascular morphology distribution data sets, and determine the final needle insertion point and final puncture angle;

[0053] S5: Control the puncture needle to perform puncture and blood collection at the final puncture position at the final puncture angle to obtain the blood collection control result.

[0054] The beneficial effects of the present invention compared to the prior art are as follows: the optimal puncture vessel data set is determined in the infrared vascular image using the puncture decision model, thereby improving the scientificity and accuracy of puncture vessel selection. Dynamic ultrasound data is obtained by guiding ultrasound probe imaging through the optimal puncture vessel coordinates in the camera coordinate system, providing more detailed and accurate subcutaneous vascular distribution information for subsequent analysis. An adaptive threshold segmentation method is used to construct a time-series vascular ROI candidate set, highlighting vascular characteristics for easy subsequent processing. The potential vascular morphology distribution data set is analyzed to determine the final needle insertion point and puncture angle, ensuring the safety and effectiveness of the puncture. Accurate control of puncture and blood collection is achieved, the success rate and quality of blood collection are improved, and the patient's medical experience is also improved.

[0055] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0056] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0058] Figure 1 Schematic diagram of the internal functional modules of the high-precision automatic venous blood sampling system in an embodiment of the present invention;

[0059] Figure 2 This is a structural diagram of a high-precision automatic venous blood sampling device in an embodiment of the present invention;

[0060] Figure 3 This is a flow chart of a high-precision automatic venous blood sampling method in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0062] Example 1

[0063] The present invention provides a high-precision automatic venous blood sampling system. Figure 1 ,include:

[0064] An optimal puncture vessel decision module is used to determine an optimal puncture vessel data set in an infrared vessel image based on a puncture decision model;

[0065] A dynamic ultrasound data acquisition module is used to determine the optimal puncture vessel coordinates in the camera coordinate system based on the optimal puncture vessel data set, and to guide the ultrasound probe on the robotic arm to perform skin-fitting ultrasound imaging based on the optimal puncture vessel coordinates in the camera coordinate system to obtain dynamic ultrasound data of the optimal puncture area;

[0066] Ultrasonic vascular data acquisition module, used to enhance vascular signals and extract ROI in dynamic ultrasound data based on an adaptive threshold segmentation method, and construct a time-series vascular ROI candidate set;

[0067] Ultrasonic vascular data analysis module, used to extract all potential vascular morphology distribution data sets from the time-series vascular ROI candidate set, and perform puncture preference and credibility analysis on all potential vascular morphology distribution data sets to determine the final needle insertion point and final puncture angle;

[0068] The venous blood sampling puncture control module is used to control the puncture needle to perform puncture and blood sampling at the final puncture angle and at the final puncture position to obtain the blood sampling control result.

[0069] In this embodiment, reference Figure 2 The equipment using the high-precision automatic venous blood sampling system includes a base plate 4, an operating table 3 arranged on the base plate 4, a six-axis collaborative robot arm 2, a puncture device 6 (provided with a puncture needle) fixed at the end of the six-axis collaborative robot arm 2, an arm fixing device 5 is arranged at the bottom, and a human-machine interactive operation screen (dual-screen image and voice guidance device) 1 is arranged on the side of the six-axis collaborative robot arm 2.

[0070] In this embodiment, the infrared blood vessel image is extracted from the high-resolution infrared image of the target venous blood collection site obtained by the binocular infrared camera and multi-wavelength infrared light source provided on the puncture device 6. The binocular infrared camera simultaneously captures two infrared images with the left and right cameras at a position approximately 15 cm above the photographed limb site.

[0071] In this embodiment, the puncture decision model is a model constructed through data training and algorithms, and is used to determine the best puncture blood vessel data set based on relevant input information.

[0072] In this embodiment, the infrared blood vessel image is an image obtained by using infrared technology to show the distribution of blood vessels.

[0073] In this embodiment, the optimal puncture vessel data set is a data set of a series of blood vessels that are identified as the most suitable for puncture operation, and is a data set representing the characteristics of the optimal puncture vessel, such as vessel diameter, depth, visibility, and surrounding tissue structure.

[0074] In this embodiment, the optimal puncture area is a deep skin area consisting of the optimal puncture blood vessel determined based on the infrared blood vessel image and the surrounding subcutaneous area that needs to be imaged by ultrasound.

[0075] In this embodiment, the optimal puncture blood vessel coordinates in the camera coordinate system are used to describe the position coordinates of the optimal puncture blood vessel in the ultrasound imaging system.

[0076] In this embodiment, based on the adaptive threshold segmentation method, the vascular signal in the dynamic ultrasound data is enhanced and the ROI is extracted to construct a time-series vascular ROI candidate set, which is:

[0077] Adaptive threshold segmentation is used to enhance the signals of blood vessels in dynamic ultrasound data and extract regions of interest (ROIs), thereby forming a set of blood vessel ROIs that change over time as candidate analysis objects.

[0078] In this embodiment, the time-series vascular ROI candidate set is a candidate set of vascular regions of interest that change over time.

[0079] In this embodiment, the potential vascular morphology distribution data set includes a data set containing possible vascular morphology distribution conditions within the explorable range of ultrasound imaging. A large amount of image data about leg blood vessels is obtained through a series of medical imaging technologies, such as enhanced CT angiography, magnetic resonance angiography, etc. After processing and analyzing these data, the following possible vascular morphology distribution conditions are found:

[0080] Some of the data showed that deep in the calf muscle, there is a thick but tortuous main blood vessel, and its branch vessels are distributed at specific angles and patterns.

[0081] Another set of data showed that on the inner thigh, there is a group of thinner blood vessels that interweave with each other to form a complex network structure.

[0082] There are also data showing that near the knee, there are several blood vessels arranged in parallel, accompanied by some small collateral vessels.

[0083] These different vascular morphology distributions are sorted and recorded, and the resulting collection is called the "potential vascular morphology distribution dataset."

[0084] In this embodiment, the final needle insertion point is the specific position where the puncture needle finally penetrates, which is determined after analysis.

[0085] In this embodiment, the final puncture angle is the inclination angle of the puncture needle relative to the skin section or the world coordinate system when the puncture needle is finally inserted, which is determined through analysis.

[0086] The beneficial effects of the above technologies are as follows: the puncture decision model is used to determine the best puncture vessel data set in the infrared vascular image, which improves the scientificity and accuracy of puncture vessel selection. The best puncture vessel coordinates in the camera coordinate system are used to guide ultrasound probe imaging to obtain dynamic ultrasound data, providing more detailed and accurate subcutaneous vascular distribution information for subsequent analysis. The adaptive threshold segmentation method is used to construct a time-series vascular ROI candidate set, highlighting the vascular characteristics for subsequent processing. The potential vascular morphology distribution data set is analyzed to determine the final needle insertion point and puncture angle, ensuring the safety and effectiveness of the puncture. Accurate control of puncture and blood collection is achieved, the success rate and quality of blood collection are improved, and the patient's medical experience is also improved.

[0087] Example 2

[0088] Based on Example 1, the optimal puncture blood vessel decision module includes:

[0089] The infrared blood vessel image extraction submodule is used to extract the infrared blood vessel image from the high-resolution infrared image of the target venous blood sampling site, and identify multiple candidate puncture blood vessel area data sets in the infrared blood vessel image;

[0090] The optimal puncture vessel decision submodule is used to screen out the optimal puncture vessel data set from all candidate puncture vessel region data sets based on the puncture decision model.

[0091] In this embodiment, the target venous blood collection site is, for example, the right forearm of the patient to perform venous blood collection, and the right forearm is the target venous blood collection site.

[0092] In this embodiment, the high-resolution infrared image is an image of the patient's arm taken by a very clear and detailed infrared camera, which can clearly show the direction and distribution of blood vessels. For example, an infrared image with a resolution of 4K or even higher can clearly show the fine structure of blood vessels.

[0093] In this embodiment, the process of identifying multiple candidate puncture blood vessel region data sets in the infrared blood vessel image includes:

[0094] The image is captured by using the difference in absorption between blood vessels and surrounding tissues under the infrared spectrum to achieve non-invasive visualization of subcutaneous blood vessels. Then, image processing techniques such as contrast enhancement, filtering denoising and morphological operations are used to enhance the vascular signals in the infrared vascular image to improve the visibility and recognition accuracy of the blood vessels. Next, the morphological-based medical image segmentation technology is used to automatically identify and accurately segment the vascular area to obtain a candidate vascular dataset suitable for puncture.

[0095] In this embodiment, the candidate puncture blood vessel area data set is: assuming that five blood vessel areas on the patient's arm that may be suitable for puncture are identified in the infrared image, each area has relevant data description, such as the thickness, position, depth, etc. of the blood vessel, the collection of these five groups of data is the candidate puncture blood vessel area data set.

[0096] In this embodiment, based on the puncture decision model, the best puncture vessel data set is selected from all candidate puncture vessel area data sets. The puncture decision model may comprehensively consider factors such as the thickness and depth of the blood vessels, whether they are close to nerves, the age and condition of the patient, etc. For example, among the five identified candidate areas, according to the calculation and judgment of the model, it is considered that the blood vessels in one area are thicker, the location is superficial, far away from the nerves, and suitable for the current patient's condition, so the data of this area is selected, and this group of selected data is the best puncture vessel data set.

[0097] Accordingly, the puncture decision model is obtained by collecting a large amount of data on patients’ blood vessels, individual characteristics, and puncture results, preprocessing and labeling them, selecting algorithm training, evaluating and optimizing them with test data, and iterating them multiple times until they perform well on the test data.

[0098] The beneficial effects of the above technology are as follows: the infrared vascular image extraction submodule can accurately extract infrared vascular images from the high-resolution infrared images of the target venous blood collection site, and identify multiple candidate puncture vascular area data sets, providing a rich data basis for subsequent screening. The optimal puncture vascular decision submodule uses the puncture decision model to screen out the best puncture vascular area data set from many candidate area data sets, thereby improving the scientificity and accuracy of puncture area selection. This step-by-step processing method first widely extracts candidate areas, and then accurately screens the best areas, effectively improving work efficiency and decision-making reliability. It helps to reduce blindness and uncertainty in the puncture process, reduce puncture risks, and improve the success rate of puncture. It provides a more optimized solution for subsequent blood collection and puncture operations, ensuring the smooth progress of the blood collection process and the quality of blood collection.

[0099] Example 3

[0100] Based on Example 2, the infrared blood vessel image extraction submodule includes:

[0101] A venous infrared imaging unit is used to perform infrared imaging of the target venous blood collection site to obtain a high-resolution infrared image of the target venous blood collection site;

[0102] A blood vessel image extraction unit, used to extract an infrared blood vessel image from a high-resolution infrared image of a target venous blood sampling site;

[0103] A blood vessel image enhancement unit, used for enhancing the infrared blood vessel image to obtain an enhanced blood vessel image;

[0104] The candidate puncture region extraction unit is used to segment the infrared blood vessel image using a morphology-based medical image segmentation method to obtain multiple candidate puncture blood vessel region data sets.

[0105] In this embodiment, the infrared vascular image is enhanced to obtain an enhanced vascular image, which means that some image processing techniques, such as adjusting contrast, brightness, sharpening, etc., are used to make the characteristics of the blood vessels in the original infrared vascular image more obvious, clear, and detailed, so as to facilitate subsequent analysis and processing.

[0106] For example, the originally unclear blood vessel contours become sharper and more prominent after enhancement, and the contrast between the blood vessels and surrounding tissues is higher.

[0107] In this embodiment, the infrared blood vessel image is segmented using a morphology-based medical image segmentation method to obtain multiple candidate puncture blood vessel region data sets, which are specifically explained as follows:

[0108] Morphological methods are usually based on the shape and structural features of objects in the image. Through this method, the infrared vascular image is divided into different regions, each of which is considered to be a possible candidate punctured blood vessel region, and the data of these regions are organized into a data set.

[0109] For example, the continuous parts of the image with similar characteristics (such as blood vessel thickness, direction, etc.) are divided into a region, and finally multiple such regions and related data are obtained.

[0110] The beneficial effects of the above technologies are as follows: the venous infrared imaging unit can obtain high-resolution infrared images of the target venous blood collection site, providing clear and accurate raw data for subsequent vascular analysis. The vascular image extraction unit can accurately extract infrared vascular images from high-resolution infrared images, highlighting key vascular information and reducing interference from irrelevant factors. The vascular image enhancement unit enhances the infrared vascular image, further highlighting the characteristics of the blood vessels, making the vascular structure clearer and more recognizable, which is helpful for subsequent analysis and processing. The candidate puncture area extraction unit uses a morphological-based medical image segmentation method to obtain multiple candidate puncture vascular area data sets, providing a wealth of options for screening the best puncture area. The overall solution gradually optimizes and prepares data through a series of processing steps, from image acquisition to vascular extraction, enhancement, and then to candidate area determination, thereby improving the accuracy and reliability of subsequent puncture decisions, reducing the difficulty and risk of puncture operations, and helping to improve the efficiency and quality of venous blood collection.

[0111] Example 4

[0112] Based on Example 1, the dynamic ultrasound data acquisition module includes:

[0113] The dual-camera posture conversion submodule is used to generate the three-dimensional coordinates of the best puncture vessel in the best puncture vessel data set through depth imaging and infrared-depth image alignment, and determine the best puncture vessel coordinates in the camera coordinate system based on the three-dimensional coordinates of the best puncture vessel;

[0114] The skin-fitting ultrasound imaging submodule is used to guide the ultrasound probe on the robotic arm to continuously capture multiple frames of effective fitting ultrasound images of the optimal puncture area based on the optimal puncture vessel coordinates in the camera coordinate system, and obtain dynamic ultrasound data of the optimal puncture area.

[0115] In this embodiment, determining the optimal puncture vessel coordinates in the camera coordinate system based on the optimal puncture vessel data set means obtaining the position coordinates of the optimal puncture area in the camera imaging system based on a series of data information about the optimal puncture vessel area through a specific calculation and conversion method.

[0116] For example, if the best blood vessel puncture area is at a specific position on the human body, the X, Y, and Z coordinate values ​​of this position relative to the camera of the ultrasound imaging system are calculated through relevant algorithms and equipment parameters.

[0117] In this embodiment, the effectively fitted ultrasound image refers to an ultrasound image in which the ultrasound probe is in close contact with the skin and can clearly and accurately reflect the internal structure and information of the optimal puncture area.

[0118] For example, under the guidance of the robotic arm, the ultrasound probe fits against the skin with appropriate pressure. The clear ultrasound images obtained are free of blurring, missing images, etc. due to poor contact or other interferences, which can provide valuable information for subsequent analysis.

[0119] The beneficial effects of the above technology are as follows: the dual-camera pose conversion submodule can accurately determine the camera coordinates based on the optimal puncture vessel data set, provide accurate position guidance for subsequent ultrasound imaging, and improve the accuracy of positioning. The skin-fitting ultrasound imaging submodule guides the ultrasound probe to continuously capture multiple frames of effective fitting ultrasound images based on accurate camera coordinates, thereby obtaining dynamic ultrasound data, which can more comprehensively and dynamically reflect the situation in the optimal puncture area. This precise positioning and continuous image capture help to obtain richer and more detailed puncture area information, providing strong support for subsequent analysis and decision-making. It improves the safety and success rate of the puncture operation and reduces puncture errors caused by inaccurate positioning or incomplete information. Overall, the collaborative work of these two submodules makes the preparation before puncture more scientific and efficient, helps to optimize the medical process and improve the quality of medical services.

[0120] Example 5

[0121] Based on Example 4, the skin-fitting ultrasound imaging submodule includes:

[0122] A signal-to-noise ratio evaluation unit, used to guide the ultrasound probe on the robot arm to continuously capture multiple frames of ultrasound images of the optimal puncture area based on the optimal puncture blood vessel coordinates in the camera coordinate system, and evaluate the signal-to-noise ratio of each frame of the ultrasound image;

[0123] An ultrasound image segmentation unit, used for identifying all valid lamination areas and all invalid lamination areas in each frame of ultrasound image based on a feature recognition algorithm;

[0124] An imaging quality scoring unit, used to input the signal-to-noise ratio, all effective lamination areas, and all invalid lamination areas of each frame of ultrasound image into an imaging quality scoring model to obtain an imaging quality scoring value for each frame of ultrasound image;

[0125] The dynamic ultrasound data acquisition unit is used to obtain dynamic ultrasound data of the optimal puncture area based on the imaging quality score values ​​of all frames of ultrasound images.

[0126] In this embodiment, guiding the ultrasonic probe on the robot arm to continuously capture multiple frames of ultrasonic images of the optimal puncture area based on the coordinates of the optimal puncture blood vessel in the camera coordinate system means that the ultrasonic probe on the robot arm is directed to continuously collect data of the corresponding optimal puncture area based on the position information of the previously determined optimal puncture blood vessel in the camera coordinate system, thereby obtaining multiple ultrasonic images. For example, the ultrasonic probe is accurately aligned with the optimal puncture area according to the coordinates, and 10 frames of ultrasonic images are obtained by continuously collecting 10 times.

[0127] In this embodiment, the signal-to-noise ratio of each frame of ultrasound image is evaluated. The signal-to-noise ratio is an indicator of image quality. The clarity and reliability of the image are determined by calculating and analyzing the ratio of useful signal to noise in each frame of ultrasound image. For example, if the intensity of useful blood vessel information in a frame of image is 100 and the noise intensity is 20, then the signal-to-noise ratio is 100 / 20=5.

[0128] In this embodiment, the ineffective adhesion area refers to the area where the ultrasound probe does not contact the skin well, and the ultrasound image obtained in this area may be unclear or inaccurate. For example, the ultrasound probe is not completely in contact with the skin, resulting in blurry or missing images in this area.

[0129] In this embodiment, all valid lamination areas and all invalid lamination areas in each frame of ultrasound image are identified based on a feature recognition algorithm. A specific algorithm is used to distinguish areas where the ultrasound probe is in good contact with the skin and thus the image is valid, and areas where the contact is poor and the image is invalid, based on certain features of the image (such as clarity, brightness uniformity, etc.). For example, a portion of the image with uniform brightness and clear blood vessel contours is identified as a valid lamination area, while a portion with uneven brightness, shadows or severe noise is identified as an invalid lamination area.

[0130] In this embodiment, the imaging quality scoring model is a mathematical model obtained through data training for evaluating ultrasound image quality, which comprehensively considers multiple factors (such as signal-to-noise ratio, effective fitting area size, etc.) to give an image quality score.

[0131] The training process of the image quality scoring model usually includes the following steps:

[0132] Collect training samples: A large number of ultrasound images are needed as training data. These images should cover a variety of different situations, including different puncture areas, different patients, different probe fitting states, etc. At the same time, each image is labeled with the corresponding imaging quality label, such as excellent, good, medium, poor, etc.

[0133] Feature extraction: Extract features related to imaging quality from the collected ultrasound images, such as signal-to-noise ratio, the area of ​​effective and ineffective fitting areas, image clarity, contrast, etc.

[0134] Choose a model architecture: You can choose common machine learning models such as decision trees, support vector machines, or deep learning models such as convolutional neural networks.

[0135] Training model: Use the extracted features and annotated image quality labels to train the model. During the training process, the model learns how to predict the image quality label based on the input features.

[0136] Model evaluation: Use an independent test set to evaluate the trained model and calculate evaluation metrics such as accuracy, recall, F1 value, etc. to determine the performance of the model.

[0137] Adjustment and optimization: If the model performance is not ideal, you need to adjust the model parameters, increase the amount of training data, improve the feature extraction method, or try other model architectures, and then retrain and evaluate until the model achieves satisfactory performance.

[0138] In summary, the training of imaging quality scoring models requires a large number of diverse ultrasound image samples with accurate annotations, as well as a suitable model architecture and optimization methods to ensure that the model can accurately evaluate the imaging quality of ultrasound images.

[0139] In this embodiment, the imaging quality score is a specific value obtained by calculating various parameters and features of the ultrasound image through the imaging quality score model, representing the quality level of the image. For example, a score of 80 indicates good image quality, and a score of 30 indicates poor image quality.

[0140] The beneficial effects of the above technologies are as follows: the signal-to-noise ratio evaluation unit can quantify the quality characteristics of the image by evaluating the signal-to-noise ratio of each frame of ultrasound image, providing important data indicators for subsequent analysis. The ultrasound image segmentation unit identifies the effective fitting area and the invalid fitting area based on the feature recognition algorithm, which helps to more accurately understand the contact between the ultrasound probe and the skin, thereby improving the accuracy of subsequent analysis. The imaging quality scoring unit inputs multiple key factors into the imaging quality scoring model to obtain the imaging quality scoring value of each frame of ultrasound image, making the evaluation of image quality more comprehensive and objective. The dynamic ultrasound data acquisition unit obtains dynamic ultrasound data based on the imaging quality scoring values ​​of all frames of ultrasound images, and can screen out high-quality image data, providing a more reliable basis for subsequent puncture decisions. The overall solution improves the quality and reliability of dynamic ultrasound data through comprehensive evaluation and screening of ultrasound images, helps to improve the accuracy and safety of puncture operations, and optimizes ultrasound imaging quality control during medical procedures.

[0141] Example 6

[0142] Based on Example 5, the dynamic ultrasound data acquisition unit includes:

[0143] The first judgment subunit is used to judge whether the imaging quality score value of each frame of ultrasound image reaches the expected score value. If so, the corresponding frame of ultrasound image is regarded as a valid fitting ultrasound image of the corresponding capture position. Otherwise, the posture adjustment path of the ultrasound probe to the corresponding capture position is determined based on the distribution positions of all valid fitting areas and all invalid fitting areas in the corresponding frame of ultrasound image.

[0144] The second judgment subunit is used to adjust the path of the ultrasound probe to the corresponding capture position, and recapture the ultrasound image at the corresponding capture position until the imaging quality score value of the latest captured ultrasound image reaches the expected score value, and then the latest captured ultrasound image is regarded as the effective fitting ultrasound image of the corresponding capture position;

[0145] The dynamic ultrasound data fitting subunit is used to sort the effective fitting ultrasound images of all capture positions according to the path sequence of all capture positions in the original capture path, so as to obtain the dynamic ultrasound data of the best puncture area.

[0146] In this embodiment, the score expectation value is a pre-set standard score value for measuring whether the imaging quality of the ultrasound image is qualified. For example, it is set to 80 points. If the imaging quality score of a frame of ultrasound image reaches or exceeds 80 points, it is considered that its quality meets the requirements.

[0147] In this embodiment, the posture adjustment path of the ultrasound probe to the corresponding capture position refers to the specific route and method of the probe moving and adjusting its posture when moving to the corresponding capture position in order to allow the ultrasound probe to obtain an image with better imaging quality at a specific capture position. For example, the probe needs to be moved 2 cm upward and rotated 15 degrees to the left.

[0148] In this embodiment, based on the posture adjustment path of the ultrasound probe to the corresponding capture position, re-capturing the ultrasound image at the corresponding capture position means moving to the original capture position according to the previously determined posture adjustment path and operating the ultrasound probe again to capture the image in a new posture. For example, after adjusting the probe position and posture according to the planned path, re-acquire the ultrasound image at the same position, in the hope of obtaining an image whose imaging quality score reaches the expected score value.

[0149] The beneficial effects of the above technology are as follows: the first judgment subunit can quickly screen out ultrasound images whose imaging quality score values ​​meet the expectations, ensure that the images used have a certain quality assurance, and improve the reliability of subsequent analysis. For ultrasound images whose imaging quality does not meet the standards, the second judgment subunit determines the posture adjustment path and recaptures it until the score expectations are met, effectively improving the quality of the effective fitting ultrasound images finally obtained. The dynamic ultrasound data fitting subunit sorts the effective fitting ultrasound images according to the path sequence of the capture position, and the obtained dynamic ultrasound data has good timing and accuracy, providing more comprehensive and accurate information for puncture decisions. The entire scheme ensures the high quality and effectiveness of dynamic ultrasound data by continuously optimizing and screening ultrasound images, which helps to improve the accuracy and safety of puncture operations. The scheme can automatically adjust and optimize the posture of the ultrasound probe, reduce the need for manual intervention, improve work efficiency, and reduce errors caused by human factors.

[0150] Example 7

[0151] On the basis of Example 6, the first determination subunit includes:

[0152] The score comparison end is used to determine whether the imaging quality score of each frame of ultrasound image reaches the expected score value;

[0153] The first output terminal is used to treat the corresponding frame of ultrasound image as a valid fitting ultrasound image of the corresponding capture position when the imaging quality score value of each frame of ultrasound image reaches the expected score value;

[0154] The image analysis end is used to determine the physical center positions of all valid lamination areas and the physical center positions of all invalid lamination areas in the corresponding frame of ultrasound image when the imaging quality score value of each frame of ultrasound image does not reach the expected score value;

[0155] The adjustment vector determination end is used to determine a first two-dimensional adjustment vector pointing from the physical center position of all valid lamination areas in the corresponding frame ultrasound image to the physical center position of all invalid lamination areas, and at the same time, determine a second two-dimensional adjustment vector pointing from the physical center position of the corresponding ultrasound image to the physical center position of all valid lamination areas in the corresponding frame ultrasound image, and at the same time, determine a third two-dimensional adjustment vector pointing from the physical center position of all invalid lamination areas in the corresponding frame ultrasound image to the physical center position of the corresponding ultrasound image;

[0156] The adjustment path generating end is used to generate a posture adjustment path of the ultrasound probe to the corresponding capture position based on the first two-dimensional adjustment vector, the second two-dimensional adjustment vector, and the third two-dimensional adjustment vector of the corresponding frame ultrasound image.

[0157] In this embodiment, based on the first two-dimensional adjustment vector, the second two-dimensional adjustment vector, and the third two-dimensional adjustment vector of the corresponding frame ultrasound image, a posture adjustment path of the ultrasound probe to the corresponding capture position is generated. This means that the direction and size information represented by the three two-dimensional adjustment vectors are comprehensively considered to plan the specific route that the ultrasound probe needs to move and adjust its posture in order to obtain better imaging quality at the corresponding capture position.

[0158] For example, the first two-dimensional adjustment vector, the second two-dimensional adjustment vector, and the third two-dimensional adjustment vector of the corresponding frame ultrasound image are input into a preset posture adjustment path generation model to obtain a posture adjustment path of the ultrasound probe to the corresponding capture position;

[0159] The posture adjustment path generation model is a mathematical model obtained through data training, which can calculate and output the specific posture adjustment path of the ultrasound probe at the corresponding capture position according to the first two-dimensional adjustment vector, the second two-dimensional adjustment vector, the third two-dimensional adjustment vector and other information of the input corresponding frame ultrasound image;

[0160] For example, it may be a neural network model based on deep learning, or a model built on a traditional machine learning algorithm. During the training phase, the model learns the relationship between a large number of different combinations of two-dimensional adjustment vectors and the corresponding reasonable posture adjustment paths. In actual applications, when a new two-dimensional adjustment vector is input, the model can predict a suitable posture adjustment path to help the ultrasound probe obtain better imaging quality.

[0161] The beneficial effects of the above technology are as follows: the score comparison end can quickly and accurately determine whether the imaging quality of the ultrasound image meets expectations, providing a clear basis for subsequent processing and improving processing efficiency. The first output end can directly output the ultrasound image that meets the expected score value as an effectively fitted ultrasound image, ensuring the effective use of high-quality images. When the image quality does not meet the standard, the image analysis end can deeply analyze the physical center position of the effective and invalid fitting areas in the image, providing detailed data support for subsequent adjustments.

[0162] The adjustment vector determination end calculates various adjustment vectors accurately, comprehensively considers the distribution of effective and invalid areas of the image, and lays the foundation for generating accurate posture adjustment paths. The adjustment path generation end generates posture adjustment paths based on detailed vector analysis, so that the ultrasound probe can be adjusted in a targeted manner, thereby increasing the possibility of capturing high-quality ultrasound images in the future, helping to obtain more accurate and reliable dynamic ultrasound data, providing stronger support for medical operations, and improving the safety and accuracy of punctures.

[0163] Example 8

[0164] Based on Example 1, the ultrasonic blood vessel data analysis module includes:

[0165] The vascular positioning and morphological fitting submodule is used to perform rough positioning and morphological fitting of vascular signals in the time-series vascular ROI candidate set based on the feature extraction method to obtain all potential vascular morphological distribution data sets;

[0166] The vascular credibility analysis submodule is used to perform puncture optimization and credibility analysis on all potential vascular morphology distribution data sets to obtain the final needle insertion point and the corresponding vascular morphology data of the local inspection area;

[0167] The final needle insertion strategy decision submodule is used to determine the final puncture angle based on the final vascular morphology data of the local inspection area corresponding to the final needle insertion point.

[0168] In this embodiment, the vascular signal is roughly located and morphologically fitted in the time-series vascular ROI candidate set based on the feature extraction method to obtain all potential vascular morphological distribution data sets, which means that the position of the vascular signal is roughly determined in the previously determined time-series vascular ROI candidate set, and the shape of the blood vessel is simulated, so as to obtain a data set containing all possible vascular morphological distribution conditions. For example, by analyzing the grayscale, texture and other features in the image, the possible location of the blood vessel is found, and the approximate shape of the blood vessel is outlined.

[0169] In this embodiment, the local investigation area corresponding to the final needle insertion point refers to an area of ​​a specific range surrounding the final needle insertion position. For example, a circular area with a radius of 1 cm centered on the final needle insertion point. This area is a specific area that is considered to be the best choice among all possible puncture areas. After comprehensively considering factors such as the thickness, depth, position stability, and surrounding tissue conditions of the blood vessels, a specific area is determined. For example, about 5 cm from the elbow on the inside of the arm, the blood vessels are relatively thick, the position is fixed, and there are fewer surrounding nerves and tissues. This specific area is the "local investigation area corresponding to the final needle insertion point."

[0170] For example, in cases where arterial puncture is required, if testing finds that the blood flow rate in a certain artery is stable, the blood vessel wall is thick, and the location is easy to operate, then this location will be identified as the "local inspection area corresponding to the final needle insertion point."

[0171] In short, the "local investigation area corresponding to the final needle insertion point" is the specific area that is considered to be most conducive to the puncture operation, minimize risks and increase the success rate after comprehensive evaluation of various factors.

[0172] In this embodiment, the blood vessel morphology data of the local inspection area is specific data information about the shape, thickness, direction, etc. of the blood vessels in this local area.

[0173] In this embodiment, the final blood vessel morphology data of the local investigation area is further analyzed and processed to obtain the most accurate and complete blood vessel morphology-related data about the local investigation area.

[0174] In this embodiment, the final puncture angle is determined based on the final vascular morphology data of the local inspection area corresponding to the final needle insertion point. The angle of entry of the puncture needle is calculated and determined based on the data on the final vascular morphology of this local area. For example, according to the data on the direction and depth of the blood vessel, it is determined that the puncture needle is most suitable to be inserted at an angle of 30 degrees.

[0175] The beneficial effects of the above technologies are as follows: the vascular positioning and morphological fitting submodule performs rough positioning and morphological fitting of vascular signals through feature extraction methods, which can accurately obtain the potential vascular morphological distribution data set, providing comprehensive and accurate basic data for subsequent analysis. The vascular credibility analysis submodule performs puncture preference and credibility analysis on the potential vascular morphological distribution data set, which can effectively screen out the optimal needle insertion point and related vascular morphological data, improving the scientificity and reliability of puncture decisions. The final needle insertion strategy decision submodule determines the final puncture angle based on the final vascular morphological data of the local inspection area, making the selection of puncture angle more accurate and reasonable, reducing the risk of puncture and improving the success rate of puncture. Overall, these three submodules work together to realize a complete decision-making process from vascular positioning, morphological analysis to needle insertion strategy, improving the accuracy and safety of venous blood sampling puncture, and helping to optimize medical operation processes and improve the quality of medical services.

[0176] Example 9

[0177] Based on Example 8, the blood vessel credibility analysis submodule includes:

[0178] A blood vessel morphology distribution data acquisition unit is used to determine all candidate blood vessel puncture positions in the optimal puncture area, and determine all blood vessel morphology distribution data of the local inspection area of ​​each candidate blood vessel puncture position in all potential blood vessel morphology distribution data sets;

[0179] A puncture preference evaluation unit, configured to evaluate the puncture preference of each candidate blood vessel puncture position based on all blood vessel morphology distribution data in the local inspection area of ​​all candidate blood vessel puncture positions;

[0180] A cluster analysis calculation unit, used to treat all the blood vessel morphology distribution data in the local inspection area of ​​each candidate blood vessel puncture position as a cluster, and calculate the credibility of the blood vessel morphology distribution data of each cluster;

[0181] A final decision preference calculation unit, used to calculate the final decision preference of each candidate vascular puncture position based on the puncture preference of each candidate vascular puncture position and the credibility of the vascular morphology distribution data of the corresponding cluster;

[0182] The puncture position final decision unit is used to take the candidate vascular puncture position with the maximum puncture final decision preference among all candidate vascular puncture positions as the final needle insertion point, and determine the final vascular morphology distribution data of the corresponding local inspection area based on all vascular morphology distribution data of the local inspection area of ​​the candidate vascular puncture position with the maximum puncture final decision preference.

[0183] In this embodiment, determining all candidate blood vessel puncture positions in the optimal puncture region refers to finding all specific positions that may be suitable for puncture in the puncture region identified as the optimal one.

[0184] In this embodiment, the candidate blood vessel puncture location is a specific point in the optimal puncture area that is considered to have the possibility of puncture. Assume that the optimal puncture area is a specific area on the patient's arm. In this area, possible candidate blood vessel puncture locations include point A, point B and point C, which are all the candidate blood vessel puncture locations determined. Point A, point B and point C are candidate blood vessel puncture locations. For example, point A is located at a straight and thicker position of the blood vessel, point B is near the branch of the blood vessel, and point C is at a shallower part of the blood vessel.

[0185] In this embodiment, all the vascular morphology distribution data of the local investigation area of ​​each candidate vascular puncture position are determined from all the potential vascular morphology distribution data sets, that is, all the data related to the vascular morphology within a specific range (i.e., the local investigation area) around each candidate puncture position is obtained from the data set containing various possible vascular morphologies. Assume that the potential vascular morphology distribution data set contains a large amount of detailed information about the vascular morphology. For the candidate vascular puncture position point A, the area with a radius of 1 cm with point A as the center is its local investigation area. All the morphological distribution data about the thickness, direction, curvature, etc. of the blood vessels in this area are obtained from the data set.

[0186] In this embodiment, all the vascular morphology distribution data in the local inspection area are all the data on various morphological features such as shape, size, and direction of the blood vessels in the local area around the candidate puncture position. Taking point A as an example, the thickness of the blood vessels in the local inspection area is 3 mm, and the direction is slightly tilted upward from left to right, with a small bend. These specific descriptions are all the vascular morphology distribution data in the local inspection area.

[0187] In this embodiment, based on all the blood vessel morphology distribution data of the local inspection area of ​​all candidate blood vessel puncture positions, the puncture preference of each candidate blood vessel puncture position is evaluated as follows:

[0188] Based on the vascular morphology data of the local areas of all candidate locations, the degree of advantage of each candidate puncture location in terms of puncture is evaluated. For example, the vascular morphology distribution data of the local inspection areas of points A, B, and C are integrated. If the blood vessels at point A are thick and straight and the surrounding tissues are simple, its puncture preference is evaluated as 90 points; point B is at a branch of the blood vessels, the risk is slightly higher, and the puncture preference is 70 points; point C has shallow but thin blood vessels, and the puncture preference is 80 points. The higher the value, the more preferred the location is for puncture.

[0189] In this embodiment, the puncture preference of the candidate blood vessel puncture position indicates the degree to which each candidate blood vessel puncture position is preferentially selected for puncture after comprehensive consideration of various factors.

[0190] In this embodiment, the credibility of the vascular morphology distribution data of each cluster is calculated, that is, the aggregation degree of the vascular morphology distribution data of each cluster, and also the reliability of the vascular morphology distribution data in the cluster; it means evaluating the consistency and stability of the data in each cluster;

[0191] Taking the vascular morphology distribution data as an example, suppose that the vascular morphology distribution data in a cluster includes the diameter, curvature, branching, etc. of the blood vessels. If most of these data are similar, for example, the diameter of the blood vessels fluctuates within a very small range, the curvature is similar, and the branching is relatively consistent, then it means that the data in this cluster is highly aggregated;

[0192] This means that the data are not randomly scattered, but have certain regularities and similarities. This high degree of clustering reflects the reliability of the data because they show more consistent and predictable patterns;

[0193] On the contrary, if the vascular morphology data in a cluster are very different, some blood vessels are very thick, some are very thin, the curvatures are different, and the branching is chaotic, then the clustering degree is low and the credibility of the data is low, because they lack obvious rules and consistency and may contain more uncertainty and errors;

[0194] For example, in one cluster, the blood vessel diameters of the 10 samples are 2.1mm, 2.2mm, 2.0mm, 2.3mm, 1.9mm, 2.1mm, 2.2mm, 2.0mm, 2.2mm, and 2.1mm, respectively. The data is relatively concentrated, indicating that the measurement of the blood vessel diameter in this cluster is more reliable and has high credibility. In another cluster, the blood vessel diameters are 1.5mm, 3.0mm, 2.5mm, 1.8mm, 2.8mm, etc., with large differences and low credibility.

[0195] In short, clusters with high credibility indicate that the vascular morphology distribution data therein are more representative and reliable, and can provide more valuable information for subsequent analysis and decision-making.

[0196] In this embodiment, based on the puncture preference of each candidate vascular puncture position and the credibility of the vascular morphology distribution data of the corresponding cluster, the final puncture decision preference of each candidate vascular puncture position is calculated, that is,

[0197] The puncture priority of each candidate position and the credibility of the cluster data to which it belongs are comprehensively considered to calculate the preference of each candidate position for final decision-making. For example, at point A, the puncture preference of 90 points multiplied by the credibility of 85% gives a final decision preference of 76.5 points; at point B, 70 points multiplied by 75% gives 52.5 points; and at point C, 80 points multiplied by 90% gives 72 points.

[0198] In this embodiment, based on all the vascular morphology distribution data of the local inspection area of ​​the candidate vascular puncture position with the maximum puncture final decision preference, the final vascular morphology distribution data of the corresponding local inspection area is determined as follows: based on the vascular morphology data of the local area of ​​the candidate puncture position with the highest final decision preference, the most accurate and final vascular morphology distribution data of the local area is determined. According to the above steps, point A has the highest score of 76.5, so based on all the vascular morphology distribution data of the local inspection area of ​​point A, further accurate analysis and confirmation are performed to finally determine the most accurate and complete vascular morphology distribution data of this area, such as more accurate vascular depth, peripheral nerve distribution, etc.

[0199] The beneficial effects of the above technology are as follows: the vascular morphology distribution data acquisition unit can comprehensively determine the candidate vascular puncture position and the corresponding vascular morphology distribution data, providing a rich and detailed information basis for subsequent evaluation and analysis. The puncture preference evaluation unit provides a preliminary criterion for selecting the best puncture position by evaluating the puncture preference of each candidate vascular puncture position. The cluster analysis calculation unit calculates the credibility of the vascular morphology distribution data of each cluster cluster, increases the evaluation of the reliability of the vascular morphology data from a statistical perspective, and makes the decision more scientific. The final decision preference calculation unit comprehensively considers the puncture preference and credibility, and calculates the final puncture decision preference of each candidate vascular puncture position, making the decision more comprehensive and accurate. The puncture position final decision unit can accurately determine the final needle insertion point and the corresponding final vascular morphology distribution data, providing the most optimized solution for the actual puncture operation, and maximizing the success rate and safety of the puncture. The overall solution achieves accurate decision-making on the puncture position through a series of fine calculations and analyses, which helps to improve the quality and effect of medical operations and reduce the pain and risks of patients.

[0200] Embodiment 10:

[0201] The present invention provides a high-precision automatic venous blood sampling method, which is applied to the high-precision automatic venous blood sampling system described in any one of embodiments 1 to 9, referring to Figure 3 ,include:

[0202] S1: Based on the puncture decision model, the optimal puncture vessel dataset is determined in the infrared vascular image;

[0203] S2: Determine the optimal puncture vessel coordinates in the camera coordinate system based on the optimal puncture vessel data set, and guide the ultrasound probe on the robotic arm to perform skin-fitting ultrasound imaging based on the optimal puncture vessel coordinates in the camera coordinate system to obtain dynamic ultrasound data of the optimal puncture area;

[0204] S3: Based on the adaptive threshold segmentation method, the vascular signals in the dynamic ultrasound data are enhanced and the ROI is extracted to construct a candidate set of time-series vascular ROIs;

[0205] S4: extract all potential vascular morphology distribution data sets from the time-series vascular ROI candidate set, perform puncture preference and credibility analysis on all potential vascular morphology distribution data sets, and determine the final needle insertion point and final puncture angle;

[0206] S5: Control the puncture needle to perform puncture and blood collection at the final puncture position at the final puncture angle to obtain the blood collection control result.

[0207] The beneficial effects of the above technologies are as follows: the puncture decision model is used to determine the best puncture vessel data set in the infrared vascular image, which improves the scientificity and accuracy of puncture vessel selection. The best puncture vessel coordinates in the camera coordinate system are used to guide ultrasound probe imaging to obtain dynamic ultrasound data, providing more detailed and accurate subcutaneous vascular distribution information for subsequent analysis. The adaptive threshold segmentation method is used to construct a time-series vascular ROI candidate set, highlighting the vascular characteristics for subsequent processing. The potential vascular morphology distribution data set is analyzed to determine the final needle insertion point and puncture angle, ensuring the safety and effectiveness of the puncture. Accurate control of puncture and blood collection is achieved, the success rate and quality of blood collection are improved, and the patient's medical experience is also improved.

[0208] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. High-precision automatic venous blood sampling system, characterized in that: include: An optimal puncture vessel decision module is used to determine an optimal puncture vessel data set in an infrared vessel image based on a puncture decision model; A dynamic ultrasound data acquisition module is used to determine the optimal puncture vessel coordinates in the camera coordinate system based on the optimal puncture vessel data set, and to guide the ultrasound probe on the robotic arm to perform skin-fitting ultrasound imaging based on the optimal puncture vessel coordinates in the camera coordinate system to obtain dynamic ultrasound data of the optimal puncture area; Ultrasonic vascular data acquisition module, used to enhance vascular signals and extract ROI in dynamic ultrasound data based on an adaptive threshold segmentation method, and construct a time-series vascular ROI candidate set; Ultrasonic vascular data analysis module, used to extract all potential vascular morphology distribution data sets from the time-series vascular ROI candidate set, and perform puncture preference and credibility analysis on all potential vascular morphology distribution data sets to determine the final needle insertion point and final puncture angle; The venous blood sampling puncture control module is used to control the puncture needle to perform puncture and blood sampling at the final puncture angle and at the final puncture position to obtain the blood sampling control result; The dynamic ultrasound data acquisition module includes: The skin-fitting ultrasound imaging submodule is used to guide the ultrasound probe on the robot arm to continuously capture multiple frames of effective fitting ultrasound images of the optimal puncture area based on the optimal puncture vessel coordinates in the camera coordinate system, and obtain dynamic ultrasound data of the optimal puncture area; Among them, the skin-fitting ultrasound imaging submodule includes: A signal-to-noise ratio evaluation unit, used to guide the ultrasound probe on the robot arm to continuously capture multiple frames of ultrasound images of the optimal puncture area based on the optimal puncture blood vessel coordinates in the camera coordinate system, and evaluate the signal-to-noise ratio of each frame of the ultrasound image; An ultrasound image segmentation unit, used for identifying all valid lamination areas and all invalid lamination areas in each frame of ultrasound image based on a feature recognition algorithm; An imaging quality scoring unit, used to input the signal-to-noise ratio, all effective lamination areas, and all invalid lamination areas of each frame of ultrasound image into an imaging quality scoring model to obtain an imaging quality scoring value for each frame of ultrasound image; A dynamic ultrasound data acquisition unit, used for acquiring dynamic ultrasound data of the optimal puncture area based on the imaging quality score values ​​of all frames of ultrasound images; Wherein, the dynamic ultrasound data acquisition unit includes: The first judgment subunit is used to judge whether the imaging quality score value of each frame of ultrasound image reaches the expected score value. If so, the corresponding frame of ultrasound image is regarded as a valid fitting ultrasound image of the corresponding capture position. Otherwise, the posture adjustment path of the ultrasound probe to the corresponding capture position is determined based on the distribution positions of all valid fitting areas and all invalid fitting areas in the corresponding frame of ultrasound image. The second judgment subunit is used to adjust the path of the ultrasound probe to the corresponding capture position, and recapture the ultrasound image at the corresponding capture position until the imaging quality score value of the latest captured ultrasound image reaches the expected score value, and then the latest captured ultrasound image is regarded as the effective fitting ultrasound image of the corresponding capture position; A dynamic ultrasound data fitting subunit is used to sort the effective fitting ultrasound images of all capture positions according to the path sequence of all capture positions in the original capture path, so as to obtain dynamic ultrasound data of the optimal puncture area; Among them, the first judgment subunit includes: The score comparison end is used to determine whether the imaging quality score of each frame of ultrasound image reaches the expected score value; The first output terminal is used to treat the corresponding frame of ultrasound image as a valid fitting ultrasound image of the corresponding capture position when the imaging quality score value of each frame of ultrasound image reaches the expected score value; The image analysis end is used to determine the physical center positions of all valid lamination areas and the physical center positions of all invalid lamination areas in the corresponding frame of ultrasound image when the imaging quality score value of each frame of ultrasound image does not reach the expected score value; The adjustment vector determination end is used to determine a first two-dimensional adjustment vector pointing from the physical center position of all valid lamination areas in the corresponding frame ultrasound image to the physical center position of all invalid lamination areas, and at the same time, determine a second two-dimensional adjustment vector pointing from the physical center position of the corresponding ultrasound image to the physical center position of all valid lamination areas in the corresponding frame ultrasound image, and at the same time, determine a third two-dimensional adjustment vector pointing from the physical center position of all invalid lamination areas in the corresponding frame ultrasound image to the physical center position of the corresponding ultrasound image; The adjustment path generating end is used to generate a posture adjustment path of the ultrasound probe to the corresponding capture position based on the first two-dimensional adjustment vector, the second two-dimensional adjustment vector, and the third two-dimensional adjustment vector of the corresponding frame ultrasound image.

2. The high-precision automatic venous blood sampling system according to claim 1, characterized in that: The optimal vascular puncture decision module includes: The infrared blood vessel image extraction submodule is used to extract the infrared blood vessel image from the high-resolution infrared image of the target venous blood sampling site, and identify multiple candidate puncture blood vessel area data sets in the infrared blood vessel image; The optimal puncture vessel decision submodule is used to screen out the optimal puncture vessel data set from all candidate puncture vessel region data sets based on the puncture decision model.

3. The high-precision automatic venous blood sampling system according to claim 2, characterized in that: Infrared vascular image extraction submodule, including: A venous infrared imaging unit is used to perform infrared imaging of the target venous blood collection site to obtain a high-resolution infrared image of the target venous blood collection site; A blood vessel image extraction unit, used to extract an infrared blood vessel image from a high-resolution infrared image of a target venous blood sampling site; A blood vessel image enhancement unit, used for enhancing the infrared blood vessel image to obtain an enhanced blood vessel image; The candidate puncture region extraction unit is used to segment the infrared blood vessel image using a morphology-based medical image segmentation method to obtain multiple candidate puncture blood vessel region data sets.

4. The high-precision automatic venous blood sampling system according to claim 1, characterized in that: The dynamic ultrasound data acquisition module also includes: The dual-camera posture conversion submodule is used to generate the three-dimensional coordinates of the best puncture vessel in the best puncture vessel data set through depth imaging and infrared-depth image alignment, and determine the best puncture vessel coordinates in the camera coordinate system based on the three-dimensional coordinates of the best puncture vessel.

5. The high-precision automatic venous blood sampling system according to claim 1, characterized in that: Ultrasound vascular data analysis module, including: The vascular positioning and morphological fitting submodule is used to perform rough positioning and morphological fitting of vascular signals in the time-series vascular ROI candidate set based on the feature extraction method to obtain all potential vascular morphological distribution data sets; The vascular credibility analysis submodule is used to perform puncture optimization and credibility analysis on all potential vascular morphology distribution data sets to obtain the final needle insertion point and the corresponding vascular morphology data of the local inspection area; The final needle insertion strategy decision submodule is used to determine the final puncture angle based on the final vascular morphology data of the local inspection area corresponding to the final needle insertion point.

6. The high-precision automatic venous blood sampling system according to claim 5, characterized in that: The vascular credibility analysis submodule includes: A blood vessel morphology distribution data acquisition unit is used to determine all candidate blood vessel puncture positions in the optimal puncture area, and determine all blood vessel morphology distribution data of the local inspection area of ​​each candidate blood vessel puncture position in all potential blood vessel morphology distribution data sets; A puncture preference evaluation unit, configured to evaluate the puncture preference of each candidate blood vessel puncture position based on all blood vessel morphology distribution data in the local inspection area of ​​all candidate blood vessel puncture positions; A cluster analysis calculation unit, used to treat all the blood vessel morphology distribution data in the local inspection area of ​​each candidate blood vessel puncture position as a cluster, and calculate the credibility of the blood vessel morphology distribution data of each cluster; A final decision preference calculation unit, used to calculate the final decision preference of each candidate vascular puncture position based on the puncture preference of each candidate vascular puncture position and the credibility of the vascular morphology distribution data of the corresponding cluster; The puncture position final decision unit is used to take the candidate vascular puncture position with the maximum puncture final decision preference among all candidate vascular puncture positions as the final needle insertion point, and determine the final vascular morphology distribution data of the corresponding local inspection area based on all vascular morphology distribution data of the local inspection area of ​​the candidate vascular puncture position with the maximum puncture final decision preference.

Images