Magnetic navigation assisted PICC (peripherally inserted central catheter) visual positioning method

Abstract

The invention discloses a magnetic navigation assisted PICC (peripherally inserted central catheter) visual positioning method. The method comprises the following steps: acquiring signal field distribution data; acquiring real-time position information flow and direction information flow according to the signal field distribution data; based on the position information flow and the direction information flow, a data fusion processing technology is adopted, multi-source information is integrated, and dynamic space coordinate data of the PICC in the deep domain of the human body are obtained; according to the dynamic space coordinate data, a real-time acquisition rate optimization model is constructed, and the real-time acquisition rate optimization model is used for analyzing the delay condition of the information feedback speed and judging whether feedback delay exceeds a preset threshold value or not, and the signal field intensity is adjusted according to a judgment result; mapping the position and direction information flow of the PICC to a visual interface according to the adjusted signal field intensity and dynamic space coordinate data to obtain a preliminary dynamic visual image; and obtaining a dynamic visual image through the preliminary dynamic visual image in combination with a guide accuracy calibration mechanism.

Description

Technical Field

[0001] This invention belongs to the field of image visualization technology, and in particular relates to a magnetic navigation-assisted PICC catheter visualization and positioning method. Background Technology

[0002] In the field of medical technology, the precise positioning and guidance of medical devices within the human body is of paramount importance, especially in interventional treatments such as intravenous catheterization, directly impacting patient outcomes and safety. With advancements in medical technology, achieving precise device navigation within the complex human body environment has become an indispensable research direction. Developments in this area can not only improve surgical success rates but also reduce patient suffering and the risk of complications.

[0003] However, current mainstream guidance methods mostly rely on imaging equipment or ultrasound technology. These methods often have certain limitations in operation, such as insufficient image resolution, poor real-time performance, or high technical requirements for operators, making it difficult to accurately determine the position and orientation of instruments in complex environments, thus increasing operational risks.

[0004] Against this backdrop, the core challenges facing this field are becoming increasingly apparent. The primary issue lies in generating a stable and controllable external environmental signal within the deep tissue environment of the human body to guide the movement of instruments. This signal needs to possess sufficient penetration and precision. If this problem remains unresolved, it will further lead to an inability to accurately capture the dynamic changes of the instrument within the body, especially during subtle adjustments in position and orientation, which can easily result in deviations. Consequently, these deviations will directly affect the accuracy of the operation and may even lead to medical risks.

[0005] Therefore, how to construct a stable external signal field and combine it with a specific response mechanism on the device to obtain its position and orientation information in the human body in real time, and ultimately achieve dynamic visual guidance, has become a key problem that urgently needs to be solved. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention proposes a magnetic navigation-assisted visual positioning method for PICC catheters, which can acquire their position and orientation information within the human body in real time, ultimately achieving dynamic visual guidance.

[0007] To achieve the above objectives, the present invention provides a magnetic navigation-assisted visual positioning method for PICC catheters, comprising:

[0008] Acquire signal field distribution data;

[0009] Based on the signal field distribution data, obtain real-time location information stream and direction information stream;

[0010] Based on the location information flow and direction information flow, data fusion processing technology is used to integrate multi-source information and obtain dynamic spatial coordinate data of PICC catheter in the deep domain of the human body.

[0011] Based on dynamic spatial coordinate data, a real-time acquisition rate optimization model is constructed. The real-time acquisition rate optimization model is used to analyze the delay of information feedback speed, determine whether the feedback delay exceeds a preset threshold, and adjust the signal field strength according to the determination result.

[0012] Based on the adjusted signal field strength and dynamic spatial coordinate data, the PICC catheter position and orientation information flow is mapped to the visualization interface to obtain a preliminary dynamic visualization image;

[0013] By combining the initial dynamic visualization image with the guided accuracy calibration mechanism, a dynamic visualization image is obtained.

[0014] Optionally, acquiring signal field distribution data includes:

[0015] By using a pre-set external signal field generation module and employing multi-band signal superposition technology, a high-penetration signal field is generated to obtain initial signal field distribution data.

[0016] If the signal field strength of the initial signal field distribution data is lower than the preset threshold, the frequency combination of the multi-band signals is adjusted to regenerate the signal field and obtain the signal field distribution data.

[0017] Optionally, based on the signal field distribution data, obtaining real-time location information streams and direction information streams includes:

[0018] The signal field distribution data is input into a pre-established signal distribution model to obtain a signal characteristic distribution map.

[0019] Based on the signal characteristic distribution map and the constraints of the stable environment, a data stream related to position and direction information is obtained.

[0020] The data stream is preprocessed, and the support vector machine algorithm is used to classify the features of the preprocessed data stream to determine the key changing trends in the data stream.

[0021] Based on key changing trends, the real-time acquisition capabilities of micro-sensors are integrated to extract dynamically updated real-time position and orientation information streams from the signal field.

[0022] Optionally, preprocessing the data stream includes: removing outliers through filtering and determining the cleaned position and direction information.

[0023] Optionally, obtaining the dynamic spatial coordinate data of the PICC catheter in the deep domain of the human body includes:

[0024] The Kalman filter algorithm is used to fuse the position information stream and the direction information stream to determine the consistent coordinate values ​​of the multi-source data;

[0025] If the consistency coordinate values ​​do not match the preset threshold range, the multi-source data will be calibrated a second time to determine if there is a deviation and correct it, so as to obtain the calibrated spatial coordinate data.

[0026] Based on the calibrated spatial coordinate data and combined with the anatomical structure model of the deep human body, the dynamic spatial position of the PICC catheter in the deep domain is mapped.

[0027] By analyzing the changing trend of the dynamic spatial location, the next movement trajectory of the PICC catheter is predicted using time series analysis methods, and the predicted coordinate information is obtained.

[0028] The predicted coordinate information is compared with the real-time acquired location information. If there is a significant difference, the data fusion module is triggered to recalculate and obtain the dynamic spatial coordinate data.

[0029] Optionally, constructing the real-time acquisition rate optimization model based on dynamic spatial coordinate data includes:

[0030] Dynamic spatial coordinate data is acquired by collecting coordinate data streams in real time through a sensor network and smoothing them using a Kalman filter algorithm to obtain a smoothed coordinate dataset.

[0031] Based on the smoothed coordinate dataset, a real-time transmission efficiency optimization model is constructed.

[0032] Optionally, adjusting the signal field strength based on the judgment result includes:

[0033] Using the real-time acquisition rate optimization model, the information feedback delay is analyzed, and the average feedback delay is calculated using the sliding window method to obtain the feedback delay evaluation result.

[0034] If the feedback delay assessment result exceeds the preset threshold, the delay exceeds the limit status by comparing the current delay with historical delay data, and the delay exceeds the limit flag is obtained.

[0035] Based on the delay exceeding the limit flag, the signal field strength is adjusted, and the signal field parameters are optimized using a gradient descent algorithm to obtain the adjusted signal field strength value.

[0036] Optionally, based on the adjusted signal field strength and dynamic spatial coordinate data, the PICC catheter position and orientation information flow is mapped to the visualization interface to obtain a preliminary dynamic visualization image, including:

[0037] The adjusted signal field strength and dynamic spatial coordinate data are standardized to obtain a unified field strength coordinate dataset.

[0038] Based on a unified field intensity coordinate dataset, three-dimensional reconstruction technology is used to spatially model the position and orientation information of the PICC catheter, and to determine the position and orientation feature set of the PICC catheter in three-dimensional space.

[0039] Based on the location and orientation feature set, the feature data stream is converted into a visual data stream using an information flow mapping method, generating mapping data that is adapted to the visual interface.

[0040] Based on the mapping data, the preliminary dynamic visualization image is obtained.

[0041] Optionally, obtaining a dynamic visualization image through the initial dynamic visualization image, combined with a guided accuracy calibration mechanism, includes:

[0042] Pixel features are obtained from the preliminary dynamic visualization image, and the deviation distribution is extracted and determined using a precision calibration mechanism.

[0043] If the deviation distribution exceeds a preset threshold, then the deviation region is used to extract features through a convolutional neural network to obtain the deviation features;

[0044] Based on the aforementioned deviation characteristics, a local enhancement algorithm is used to adjust the intensity of the signal field and determine the enhanced signal field.

[0045] Spatial distribution data is obtained from the enhanced signal field, and the path is optimized using a dynamic programming algorithm to obtain the guiding path;

[0046] The guide path is mapped to an image to obtain the dynamic visualization image.

[0047] Compared with the prior art, the present invention has the following advantages and technical effects:

[0048] This invention solves the signal attenuation problem caused by the complexity of the deep human body environment by constructing a high-penetration multi-band signal field. It acquires real-time position and orientation information from a stable signal field and uses data fusion technology to obtain the dynamic spatial coordinates of the PICC catheter. By constructing a real-time acquisition rate optimization model, this invention can dynamically adjust the signal field strength according to feedback delay, improving data transmission efficiency. Finally, this invention applies 3D reconstruction technology to map the PICC catheter information to a visualization interface and combines it with an accuracy calibration mechanism to generate a continuous navigation guidance data stream, achieving precise positioning and real-time visualization guidance of medical devices in the deep human body, providing strong support for minimally invasive surgery. Attached Figure Description

[0049] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0050] Figure 1 This is a flowchart of a magnetic navigation-assisted PICC catheter visualization and positioning method according to an embodiment of the present invention. Detailed Implementation

[0051] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0052] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0053] This invention proposes a magnetic navigation-assisted visual positioning method for PICC catheters, such as... Figure 1 As shown, the specific steps include:

[0054] Acquire signal field distribution data;

[0055] Based on the signal field distribution data, obtain real-time location information stream and direction information stream;

[0056] Based on the location information flow and direction information flow, data fusion processing technology is used to integrate multi-source information and obtain dynamic spatial coordinate data of PICC catheter in the deep domain of the human body.

[0057] Based on dynamic spatial coordinate data, a real-time acquisition rate optimization model is constructed. The real-time acquisition rate optimization model is used to analyze the delay of information feedback speed, determine whether the feedback delay exceeds a preset threshold, and adjust the signal field strength according to the determination result.

[0058] Based on the adjusted signal field strength and dynamic spatial coordinate data, the PICC catheter position and orientation information flow is mapped to the visualization interface to obtain a preliminary dynamic visualization image;

[0059] By combining the initial dynamic visualization image with the guided accuracy calibration mechanism, a dynamic visualization image is obtained.

[0060] Furthermore, acquiring signal field distribution data includes:

[0061] By using a pre-set external signal field generation module and employing multi-band signal superposition technology, a high-penetration signal field is generated to obtain initial signal field distribution data.

[0062] If the signal field strength of the initial signal field distribution data is lower than the preset threshold, the frequency combination of the multi-band signals is adjusted to regenerate the signal field and obtain the signal field distribution data.

[0063] Specifically, a high-penetration signal field can be generated using a pre-set external signal field generation module and multi-band signal superposition technology. In principle, multi-band signals are superimposed with electromagnetic waves of different frequencies to form a composite signal field, enhancing the ability to penetrate complex media.

[0064] For example, in medical signal field applications, a combination of three frequencies—2.4 GHz, 5.8 GHz, and 10 GHz—can be selected, with an initial power of 10 mW, generating a coverage area of ​​10 m. 2 The signal field of the region is analyzed to obtain the initial signal field distribution data.

[0065] If the signal field strength is lower than the preset threshold of 0.5mW / cm 2 The frequency combination was then adjusted to 2.4GHz, 6GHz, and 12GHz to regenerate the signal field, resulting in an optimized signal field strength of 0.7mW / cm². 2 The coverage is more uniform.

[0066] Furthermore, based on the signal field distribution data, the real-time location information stream and direction information stream are obtained, including:

[0067] The signal field distribution data is input into a pre-established signal distribution model to obtain a signal characteristic distribution map.

[0068] Based on the signal characteristic distribution map and the constraints of the stable environment, a data stream related to position and direction information is obtained.

[0069] The data stream is preprocessed, and the support vector machine algorithm is used to classify the features of the preprocessed data stream to determine the key changing trends in the data stream.

[0070] Based on key changing trends, the real-time acquisition capabilities of micro-sensors are integrated to extract dynamically updated real-time position and orientation information streams from the signal field.

[0071] Furthermore, the preprocessing of the data stream includes: removing outliers through filtering and determining the cleaned position and direction information.

[0072] Specifically, if noise interference exists in the data stream, such as abnormal fluctuations of 0.2mV caused by muscle activity, a low-pass filter can be used for processing. The filter is set to a cutoff frequency of 50Hz. After removing high-frequency noise, a cleaned position information stream (e.g., coordinate sequence) and orientation information stream (e.g., angle sequence) are obtained. This process ensures the reliability of the data stream and provides high-quality input for subsequent analysis. For example, when using a support vector machine algorithm for feature classification of the cleaned data stream, the position and orientation data can be divided into "stable" and "unstable" categories. Assuming that at a depth of 3cm, the position coordinate change is less than 0.05cm and the angle change is less than 5°, it is classified as "stable," indicating that the signal field is suitable for continuous monitoring in this area. The classification results can reveal key trends in the signal field, such as the signal stabilizing at a specific depth or fluctuating due to changes in tissue density, thereby guiding the dynamic adjustment of the device.

[0073] In one possible implementation, the real-time acquisition capabilities of miniature sensors are integrated, allowing for dynamic updates of position and orientation information via a surface-embedded sensor array. For example, the sensor array acquires signals 10 times per second, recording real-time coordinate changes (e.g., from x = 2 cm to x = 2.1 cm) and angle changes (e.g., from 30° to 32°). This dynamic data is then integrated into a real-time information flow using a fusion algorithm, generating continuous directional guidance data, such as indicating that an instrument should move towards an area with higher signal strength.

[0074] Specifically, if the final directional guidance data deviates from a preset threshold range (e.g., an angle shift exceeding 10°), signal characteristic data can be reacquired through a feedback adjustment mechanism. For example, when an angle shift of 15° is detected, the sensor is triggered to reacquire the signal strength at a depth of 2 cm, and a new information flow direction is generated after correction. This mechanism ensures the accuracy of the device's response, especially in dynamic environments such as signal field changes during heartbeats. Through the above method, each step from signal extraction to dynamic updating is closely linked, jointly supporting the integrity of signal field analysis. The signal characteristic distribution map provides the foundation for localization, filtering ensures data quality, feature classification reveals changing trends, and real-time acquisition and feedback adjustment enhance dynamic adaptability. These technologies work together to provide reliable support for the accurate response of medical devices in complex human environments.

[0075] Furthermore, obtaining the dynamic spatial coordinate data of the PICC catheter in the deep domain of the human body includes:

[0076] The Kalman filter algorithm is used to fuse the position information stream and the direction information stream to determine the consistent coordinate values ​​of the multi-source data;

[0077] If the consistency coordinate values ​​do not match the preset threshold range, the multi-source data will be calibrated a second time to determine if there is a deviation and correct it, so as to obtain the calibrated spatial coordinate data.

[0078] Based on the calibrated spatial coordinate data and combined with the anatomical structure model of the deep human body, the dynamic spatial position of the PICC catheter in the deep domain is mapped.

[0079] By analyzing the changing trend of the dynamic spatial location, the next movement trajectory of the PICC catheter is predicted using time series analysis methods, and the predicted coordinate information is obtained.

[0080] The predicted coordinate information is compared with the real-time acquired location information. If there is a significant difference, the data fusion module is triggered to recalculate and obtain the dynamic spatial coordinate data.

[0081] Specifically, in scenarios involving real-time acquisition of location and orientation information, sensor devices can be deployed at specific locations within medical devices to capture subtle movements of the device deep within the body. Assuming the sensor acquires data 50 times per second, the raw data stream contains changes in spatial three-dimensional coordinates and orientation angles. This high-frequency acquisition ensures data continuity, providing sufficient raw information for subsequent processing. For example, when denoising and calibrating the raw data stream, a mean filtering method can be used, averaging the data from five consecutive time points to smooth out short-term noise interference. If the orientation angle data acquired in a given instance experiences an abnormal jump within a short period, suddenly changing from 30 degrees to 45 degrees, mean filtering can correct it to a value close to the true 32 degrees. This preprocessing method effectively reduces data fluctuations caused by environmental interference. For instance, the Kalman filter algorithm can be understood as a dynamic estimation tool used to fuse multi-source data. In a specific scenario, assuming the position information comes from two sensors, one providing x-axis and y-axis coordinates and the other providing z-axis coordinates, Kalman filtering can be used to combine the data from both to generate consistent coordinate values, such as ultimately determining the instrument's position as x = 2.5, y = 3.0, and z = 1.8. This method can balance the errors of different sensors and improve data reliability. For example, during secondary calibration, if the consistent coordinate values ​​deviate from a preset threshold range, such as the x-axis coordinate exceeding the expected range of 2.0 to 3.0 and reaching 3.5, it can be determined whether it is a systematic deviation by comparing with historical data trends, and then corrected to 2.8. This calibration method can further ensure the accuracy of the data and provide a reliable foundation for subsequent spatial mapping. For example, when performing dynamic spatial position mapping in conjunction with a deep anatomical structure model of the human body, the coordinate data of the instrument can be matched with a pre-constructed three-dimensional anatomical atlas. Assuming the instrument is currently located near a key area, mapping can visually display its spatial relationship with the surrounding structures. This method helps the operator understand the specific position of the instrument in a complex environment.

[0082] Furthermore, based on dynamic spatial coordinate data, the real-time acquisition rate optimization model is constructed as follows:

[0083] Dynamic spatial coordinate data is acquired by collecting coordinate data streams in real time through a sensor network and smoothing them using a Kalman filter algorithm to obtain a smoothed coordinate dataset.

[0084] Based on the smoothed coordinate dataset, a real-time transmission efficiency optimization model is constructed.

[0085] Furthermore, adjusting the signal field strength based on the judgment result includes:

[0086] Using the real-time acquisition rate optimization model, the information feedback delay is analyzed, and the average feedback delay is calculated using the sliding window method to obtain the feedback delay evaluation result.

[0087] If the feedback delay assessment result exceeds the preset threshold, the delay exceeds the limit status by comparing the current delay with historical delay data, and the delay exceeds the limit flag is obtained.

[0088] Based on the delay exceeding the limit flag, the signal field strength is adjusted, and the signal field parameters are optimized using a gradient descent algorithm to obtain the adjusted signal field strength value.

[0089] Specifically, real-time acquisition by sensor networks is a core component in obtaining dynamic spatial coordinate data. Sensor networks typically consist of multiple miniature sensors distributed across different locations within the target area to capture changes in the device's position within the deep layers of the human body. Assuming an application where the sensors acquire data 100 times per second, ensuring high-frequency updates to the data stream, this high-frequency acquisition provides sufficient raw information for subsequent processing, especially in scenarios with high real-time requirements in dynamic environments. In one possible implementation, smoothing using the Kalman filter algorithm can be understood as a data optimization technique to reduce noise interference in sensor acquisition. Assuming the acquired coordinate data contains jumps caused by signal interference, Kalman filtering can perform weighted predictions based on historical and current data to generate a more continuous, smooth coordinate dataset. This method is particularly suitable for device position tracking in complex environments, effectively improving data stability. For example, when constructing a real-time transmission efficiency optimization model, bottlenecks in the data transmission process can be analyzed based on the smooth coordinate dataset. Suppose that in a test, data transmission efficiency drops to 80% during peak hours. A linear regression algorithm predicts the efficiency trend over the next 10 minutes, suggesting a further drop to 75%. This prediction provides data support for subsequent optimization and helps in taking proactive measures. In one possible implementation, the sliding window method can be used to calculate the average feedback delay for information feedback latency analysis. Assuming a window size of 5 seconds, in a given operation, the delays at five consecutive time points are 0.2 seconds, 0.3 seconds, 0.4 seconds, 0.3 seconds, and 0.2 seconds, respectively, resulting in an average delay of 0.28 seconds. If a preset threshold of 0.25 seconds is set, a delay exceeding the limit flag is triggered. This method dynamically reflects the latency status, providing a basis for subsequent adjustments. For example, when adjusting signal field strength, the gradient descent algorithm can be used to optimize parameters. Assuming an initial signal field strength of 50 units, through multiple iterations, it can be optimized to 60 units, significantly improving data transmission stability. This optimization method dynamically adjusts parameters based on real-time feedback, ensuring transmission efficiency.

[0090] Furthermore, based on the adjusted signal field strength and dynamic spatial coordinate data, the PICC catheter position and orientation information is mapped onto the visualization interface to obtain preliminary dynamic visualization images, including:

[0091] The adjusted signal field strength and dynamic spatial coordinate data are standardized to obtain a unified field strength coordinate dataset.

[0092] Based on a unified field intensity coordinate dataset, three-dimensional reconstruction technology is used to spatially model the position and orientation information of the PICC catheter, and to determine the position and orientation feature set of the PICC catheter in three-dimensional space.

[0093] Based on the location and orientation feature set, the feature data stream is converted into a visual data stream using an information flow mapping method, generating mapping data that is adapted to the visual interface.

[0094] Based on the mapping data, the preliminary dynamic visualization image is obtained.

[0095] Specifically, signal field strength data and dynamic spatial coordinate data can be acquired in real time through a sensor network deployed in a specific area. For example, in an indoor navigation scenario, sensor nodes are distributed in multiple corners of the room. Signal field strength data is collected in milliwatts per square meter, while dynamic spatial coordinate data records the instrument's real-time position with centimeter-level precision. The data processing module standardizes the signal field strength and coordinate data, unifying the units to ensure consistency in subsequent analysis. For example, the signal field strength might be normalized to a range of 0 to 1, and the coordinate data might be converted to a relative coordinate system with the room center as the origin. This standardization facilitates subsequent modeling and improves data compatibility. In one possible implementation, a spatial model of the instrument is generated using 3D reconstruction technology based on a unified field strength and coordinate dataset. For example, assuming a medical surgery scenario, the instrument is a surgical robot arm, and sensors collect its position and orientation information in 3D space. Using stereomicroscopy combined with the field strength data, a precise 3D model of the arm is constructed, generating a feature set containing position coordinates and orientation angles (such as pitch and yaw). This feature set clearly describes the instrument's motion trajectory and attitude, providing a foundation for subsequent data stream conversion. Specifically, the information flow mapping method transforms a feature data stream into a visualization data stream. For example, in the surgical scenario described above, the positional and orientation feature set is converted into a pixel coordinate stream adapted to the display screen through a mapping algorithm, generating a preliminary two-dimensional or three-dimensional image data stream. During the mapping process, interpolation algorithms can be used to ensure a smooth data transition; for example, interpolating 100 frames of coordinate data per second to 200 frames per second improves display smoothness. This mapping method ensures the real-time nature and intuitiveness of the visualization interface.

[0096] Furthermore, by combining the initial dynamic visualization image with the guidance accuracy calibration mechanism, the dynamic visualization image is obtained by:

[0097] Pixel features are obtained from the preliminary dynamic visualization image, and the deviation distribution is extracted and determined using a precision calibration mechanism.

[0098] If the deviation distribution exceeds a preset threshold, then the deviation region is used to extract features through a convolutional neural network to obtain the deviation features;

[0099] Based on the aforementioned deviation characteristics, a local enhancement algorithm is used to adjust the intensity of the signal field and determine the enhanced signal field.

[0100] Spatial distribution data is obtained from the enhanced signal field, and the path is optimized using a dynamic programming algorithm to obtain the guiding path;

[0101] The guide path is mapped to an image to obtain the dynamic visualization image.

[0102] Specifically, when the dynamic visualization algorithm generates the initial image, it can construct a three-dimensional mesh model by real-time acquisition of signal field intensity data and combining it with spatial coordinate information. Assuming the signal field data originates from medical imaging equipment, with intensity values ​​ranging from 0 to 100 units and spatial coordinates in millimeters ranging from -50 to 50 millimeters, the algorithm first maps this data to a three-dimensional mesh to generate an initial image, presenting the preliminary position of the instrument in space. This method ensures that the image intuitively reflects the spatial distribution of the instrument, providing a foundation for subsequent analysis. In one possible implementation, real-time rendering technology can employ the stereo rendering method of stereomicroscopes. Through GPU acceleration, the signal field intensity data is converted into pixel values, generating an image with a resolution of 1080p. For example, if the signal intensity of a certain area is 80 units, it is mapped to high-brightness pixels during rendering, intuitively displaying the instrument's position. It should be noted that texture mapping can be introduced during rendering to enhance the image's three-dimensionality, facilitating quick identification of key areas by the operator. For pixel feature extraction, the initial image can be analyzed using edge detection algorithms. For example, if the pixel grayscale value change of the instrument edge in the image exceeds 20 units, it is determined to be a feature boundary. The precision calibration mechanism further analyzes these boundaries and calculates the deviation distribution. For example, if the deviation distribution shows that the grayscale value of a certain area deviates from the expected value by more than 10 units, it is marked as an abnormal area. This method can accurately locate deviations in the image, facilitating subsequent optimization. Specifically, convolutional neural networks can be used for deviation feature extraction. Assuming the input image is 256*256 pixels, the network extracts the texture and shape features of the deviation area through multiple convolution operations. For example, if the deviation feature of a certain area is displayed as irregular edges, the network can output the corresponding feature vector to guide signal field adjustment. This approach improves the automation of feature extraction. In one embodiment, a local enhancement algorithm can adjust areas with low signal field intensity. For example, if the signal intensity of a certain area increases from 50 units to 70 units, the image clarity is enhanced by amplifying the local signal features. This method ensures that the signal field data of key areas is more easily visualized. For example, when optimizing a path, a dynamic programming algorithm can calculate the optimal path for the movement of the device based on the spatial distribution data of the enhanced signal field. Assuming that the spatial distribution data shows that the signal intensity fluctuation of a certain path is less than 5 units, that path is preferentially selected. Gradient descent further smooths the path, for example, adjusting the path curvature from 0.8 to 0.5 to ensure a smooth trajectory. This method improves the stability of path planning. It's worth noting that when the final guided path is mapped to the image, path priorities can be distinguished using color coding. For example, priority paths are represented in green, and secondary paths in yellow. This visualization method intuitively presents the path selection results, facilitating quick decision-making by the operator.

[0103] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A magnetic navigation-assisted visual positioning method for PICC catheters, characterized in that, include: Acquire signal field distribution data; Based on the signal field distribution data, real-time position information stream and direction information stream are obtained; Based on the location information flow and direction information flow, data fusion processing technology is used to integrate multi-source information and obtain dynamic spatial coordinate data of PICC catheter in the deep domain of the human body. Based on the dynamic spatial coordinate data, a real-time acquisition rate optimization model is constructed. The real-time acquisition rate optimization model is used to analyze the delay of information feedback speed, determine whether the feedback delay exceeds a preset threshold, and adjust the signal field strength according to the determination result. Based on the adjusted signal field strength and dynamic spatial coordinate data, the PICC catheter position and orientation information flow is mapped to the visualization interface to obtain a preliminary dynamic visualization image; By combining the initial dynamic visualization image with the guided accuracy calibration mechanism, a dynamic visualization image is obtained.

2. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, Acquiring signal field distribution data includes: By using a pre-set external signal field generation module and employing multi-band signal superposition technology, a high-penetration signal field is generated to obtain initial signal field distribution data. If the signal field strength of the initial signal field distribution data is lower than the preset threshold, the frequency combination of the multi-band signals is adjusted to regenerate the signal field and obtain the signal field distribution data.

3. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, Based on the signal field distribution data, the real-time location information stream and direction information stream are obtained, including: The signal field distribution data is input into a pre-established signal distribution model to obtain a signal characteristic distribution map. Based on the signal characteristic distribution map and the constraints of the stable environment, a data stream related to position and direction information is obtained. The data stream is preprocessed, and the support vector machine algorithm is used to classify the features of the preprocessed data stream to determine the key changing trends in the data stream. Based on key changing trends, the real-time acquisition capabilities of micro-sensors are integrated to extract dynamically updated real-time position and orientation information streams from the signal field.

4. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 3, characterized in that, Preprocessing the data stream includes: removing outliers through filtering and determining the cleaned position and direction information.

5. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, Obtaining dynamic spatial coordinate data of the PICC catheter in the deep domain of the human body includes: The Kalman filter algorithm is used to fuse the position information stream and the direction information stream to determine the consistent coordinate values ​​of the multi-source data; If the consistency coordinate values ​​do not match the preset threshold range, the multi-source data will be calibrated a second time to determine if there is a deviation and correct it, so as to obtain the calibrated spatial coordinate data. Based on the calibrated spatial coordinate data and combined with the anatomical structure model of the deep human body, the dynamic spatial position of the PICC catheter in the deep domain is mapped. By analyzing the changing trend of the dynamic spatial location, the next movement trajectory of the PICC catheter is predicted using time series analysis methods, and the predicted coordinate information is obtained. The predicted coordinate information is compared with the real-time acquired location information. If there is a significant difference, the data fusion module is triggered to recalculate and obtain the dynamic spatial coordinate data.

6. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, Based on dynamic spatial coordinate data, the real-time acquisition rate optimization model is constructed as follows: Dynamic spatial coordinate data is acquired by collecting coordinate data streams in real time through a sensor network and smoothing them using a Kalman filter algorithm to obtain a smoothed coordinate dataset. Based on the smoothed coordinate dataset, a real-time transmission efficiency optimization model is constructed.

7. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, Adjusting the signal field strength based on the judgment result includes: Using the real-time acquisition rate optimization model, the information feedback delay is analyzed, and the average feedback delay is calculated using the sliding window method to obtain the feedback delay evaluation result. If the feedback delay assessment result exceeds the preset threshold, the delay exceeds the limit status by comparing the current delay with historical delay data, and the delay exceeds the limit flag is obtained. Based on the delay exceeding the limit flag, the signal field strength is adjusted, and the signal field parameters are optimized using a gradient descent algorithm to obtain the adjusted signal field strength value.

8. The magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, Based on the adjusted signal field strength and dynamic spatial coordinate data, the PICC catheter position and orientation information is mapped to the visualization interface to obtain preliminary dynamic visualization images, including: The adjusted signal field strength and dynamic spatial coordinate data are standardized to obtain a unified field strength coordinate dataset. Based on a unified field intensity coordinate dataset, three-dimensional reconstruction technology is used to spatially model the position and orientation information of the PICC catheter, and to determine the position and orientation feature set of the PICC catheter in three-dimensional space. Based on the location and orientation feature set, the feature data stream is converted into a visual data stream using an information flow mapping method, generating mapping data that is adapted to the visual interface. Based on the mapping data, the preliminary dynamic visualization image is obtained.

9. A magnetic navigation-assisted visual positioning method for PICC catheters according to claim 1, characterized in that, The process of obtaining dynamic visualization images, using the initial dynamic visualization image and a guided accuracy calibration mechanism, includes: Pixel features are obtained from the preliminary dynamic visualization image, and the deviation distribution is extracted and determined using a precision calibration mechanism. If the deviation distribution exceeds a preset threshold, then the deviation region is used to extract features through a convolutional neural network to obtain the deviation features; Based on the aforementioned deviation characteristics, a local enhancement algorithm is used to adjust the intensity of the signal field and determine the enhanced signal field. Spatial distribution data is obtained from the enhanced signal field, and the path is optimized using a dynamic programming algorithm to obtain the guiding path; The guide path is mapped to an image to obtain the dynamic visualization image.

Images