An adaptive resistance-adjusted hydrophilic coated guidewire device

Abstract

The present application relates to the technical field of medical devices, and discloses a hydrophilic coating guide wire device with self-adaptive resistance adjustment, comprising a guide wire body, a rotating and twisting device, a clamping device, a collecting device and a control device. The guide wire body is coated with a TPU outer skin and a hydrophilic coating, the outer skin is in a spiral zebra pattern and is provided with a mark point; the rotating and twisting device comprises a threaded section, a petal structure and a locking mechanism; the clamping device is provided with a micro motor, and the output shaft of the micro motor is connected with the rotating and twisting device. The collecting device comprises a resistance sensor and an image monitoring device; the control device is electrically connected with the micro motor and the collecting device, is used for collecting image and resistance data, determines a resistance threshold value through edge detection and a deep regression neural network, and adjusts the power of the micro motor in real time based on a random forest model. The present application automatically adjusts the power of the micro motor, and reduces the risk of damage to the blood vessel or the lumen wall.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical devices, in particular to a hydrophilic coating guide wire device with adaptive resistance adjustment. BACKGROUND

[0002] As a key instrument in interventional medical operations, guide wires are widely used in cardiovascular, neurovascular, urinary system and digestive tract and other various lumen interventional treatment and diagnosis processes. In order to improve the passability and control performance of the guide wire in the narrow and curved lumen, the existing guide wire is mostly provided with a hydrophilic coating structure, so that the surface is lubricated after being in contact with the body fluid, and the friction resistance in the pushing process can be effectively reduced.

[0003] In actual operation, the guide wire not only needs to be pushed linearly, but also often needs to be rotated to adjust the direction of the front end and pass through the complex bifurcation or blocked part. However, during the rotation, there is frictional contact between the guide wire and the inner wall of the blood vessel or other lumen, and such contact will generate a certain rotational resistance. In the current clinic, the judgment of the rotational resistance mainly depends on the feeling of the operator, and the rotational force and speed are adjusted by the human perception of the resistance change. This method has two significant shortcomings: the subjectivity of resistance perception, the perception of resistance size by different operators is quite different, and the perception accuracy is prone to decline due to fatigue in long-time operation; potential risk of lumen wall damage, when the operator fails to accurately judge the sudden increase of resistance or local jamming, continuing to apply rotational force may cause excessive friction or collision between the front end or outer wall of the guide wire and the lumen wall, thereby causing damage to the lumen wall. In addition, although the hydrophilic coating can reduce the friction to a certain extent, it cannot provide feedback or adjust control in time when the resistance suddenly changes, which makes it difficult for the traditional rotation operation relying on human perception to balance safety.

[0004] Therefore, it is necessary to design a hydrophilic coating guide wire device with adaptive resistance adjustment to solve the problems in the prior art. SUMMARY

[0005] In view of this, the present application provides a hydrophilic coating guide wire device with adaptive resistance adjustment, which aims to solve the problem that the size of the resistance during the rotation of the guide wire cannot be accurately perceived, and the front end or outer wall of the guide wire is prone to excessive friction or collision with the lumen wall, causing damage to the lumen wall.

[0006] The present application provides a hydrophilic coating guide wire device with adaptive resistance adjustment, which comprises:

[0007] The guide wire body is wrapped with a TPU outer skin, and a hydrophilic coating is coated on the surface of the TPU outer skin, and the color of the TPU outer skin is a spiral zebra pattern; the TPU outer skin is also provided with a mark point;

[0008] The twist device comprises a twist main body and a locking device; the twist main body comprises a threaded segment, a petal structure and a matching hole; the locking device comprises a locking threaded segment and a tapered fastening segment;

[0009] The clamping device is internally provided with a micro motor, and the output shaft of the micro motor is connected with the twist device through the matching hole;

[0010] The collection device comprises a resistance sensor and an image monitoring device;

[0011] The control device is arranged on the twist device and is electrically connected with the micro motor and the collection device; the control device comprises a collection unit, an analysis unit and a processing unit; the collection unit is configured to collect image data in the image monitoring device in real time, and collect resistance values of the guide wire body during rotation detected by the resistance sensor in real time based on an exponentially weighted moving average method; and the image data and the resistance values are respectively preprocessed; the analysis unit is configured to analyze the use length data of the guide wire body in the preprocessed image data based on edge detection, and determine a resistance threshold value based on a deep regression neural network model; compare the resistance threshold value with the resistance value to determine whether to adjust the operating power of the micro motor; the processing unit is configured to adjust the operating power of the micro motor based on a random forest model when it is determined to adjust the operating power of the micro motor, and run the micro motor at the adjusted operating power.

[0012] Further, when the collection unit collects the resistance values of the guide wire body during rotation detected by the resistance sensor in real time based on the exponentially weighted moving average method, it comprises:

[0013] The collection unit collects the resistance values of the resistance sensor at each moment, and generates continuous data from the resistance values; the resistance values are smoothed, and the latest resistance values are given high weights and the past resistance values are given low weights.

[0014] Further, when the collection unit pre-processes the image data and the resistance values respectively, it comprises:

[0015] The image data preprocessing comprises denoising to eliminate random noise in the collection process, standardizing the image to eliminate background differences, and region cropping or scaling to focus on the area where the guide wire is located, reduce irrelevant information interference, and generate clear image data;

[0016] The resistance value preprocessing comprises abnormal value filtering to eliminate sudden abnormal values caused by sensor jitter or short-term interference, and normalization processing to map the resistance data to a unified numerical range and eliminate dimension differences.

[0017] Further, the analysis unit analyzes the pre-processed image data based on edge detection, including:

[0018] The analysis unit extracts contour information of the guide wire body in the image data, generates a center line of the guide wire body based on the contour information, performs pixel-level tracking and accumulation calculation along the center line, and obtains first initial use length data of the guide wire body.

[0019] The analysis unit identifies the number of remaining mark points in the image data, compares the mark point spacing with the initial number, and calculates second initial use length data of the guide wire body.

[0020] Calculate the length difference between the first initial use length data and the second initial use length data, and compare the length difference with a length difference threshold value. The analysis unit determines whether the guide wire body use length recognition data is correct.

[0021] When the length difference is greater than the length difference threshold value, the analysis unit determines that the guide wire body use length recognition is incorrect, and issues an alarm.

[0022] When the length difference is less than or equal to the length difference threshold value, the analysis unit determines that the guide wire body use length recognition is correct.

[0023] Further, before the analysis unit determines the resistance threshold value based on the use length data using a deep regression neural network model, including:

[0024] The analysis unit constructs a historical feature vector from a plurality of historical initial use length data sets, each of which includes historical first initial use length data and historical second initial use length data. The analysis unit constructs a historical information set from the historical feature vector and the resistance threshold value, and each historical feature vector corresponds to a resistance threshold value.

[0025] The historical information set is sampled according to a predetermined proportion to obtain a training subset and a test subset.

[0026] Obtain a pre-established deep regression neural network model, and iteratively train the deep regression neural network model according to the training subset. Evaluate the iteratively trained deep regression neural network model according to the test subset, and determine whether to stop iterative training according to the evaluation value.

[0027] Further, when the analysis unit determines the resistance threshold value based on the use length data using a deep regression neural network model, including:

[0028] The deep regression neural network model comprises an input layer, 8 hidden layers and an output layer; the input layer is used to receive an initial feature vector constituted by an initial use length data set; the hidden layers extract the complex nonlinear relationship between the use length and the resistance change through the weighted sum of neurons and the nonlinear activation function mapping, and convert the input features into high-order abstract features layer by layer; and the output layer maps the high-order abstract features into the resistance threshold under the current guide wire use length data.

[0029] Further, when the hidden layers extract the complex nonlinear relationship between the use length and the resistance change through the weighted sum of neurons and the nonlinear activation function mapping, the following is included:

[0030] Each neuron of the hidden layers multiplies each input feature by a corresponding weight coefficient, and sums the weighted results to obtain the weighted sum of the neuron, and the neuron maps the weighted sum through a nonlinear activation function to convert the linearly combined input into a nonlinear output, so that the nonlinear activation function captures the nonlinear change rule between the use length and the resistance; the output of the nonlinear activation function mapping is transmitted to the next layer until the high-order abstract features are formed by the last hidden layer.

[0031] Further, when the analysis unit compares the resistance threshold with the resistance value to determine whether to adjust the operating power of the micro motor, the following is included:

[0032] When the resistance value is greater than or equal to the resistance threshold, the analysis unit determines to adjust the operating power of the micro motor.

[0033] When the resistance value is less than the resistance threshold, the analysis unit determines not to adjust the operating power of the micro motor.

[0034] Further, when it is determined to adjust the operating power of the micro motor, the analysis unit adjusts the operating power of the micro motor based on a random forest model, and the following is included:

[0035] The random forest model is established by using cross-validation combined with grid search to find the best parameters of the random forest model;

[0036] The random forest model is fitted using the training set; the test set is brought into the random forest model, and the accuracy of the operating power is calculated;

[0037] When the accuracy reaches a preset accuracy threshold, the operating power of the micro motor at the current moment is obtained according to the resistance value.

[0038] Further, when the random forest model is established by using cross-validation combined with grid search to find the best parameters of the random forest model, the following is included:

[0039] dividing the training set into k mutually exclusive subsets; defining a hyperparameter search space for the random forest model, the hyperparameters including a range of number of trees, a range of maximum depth, a range of minimum sample split threshold, and a range of maximum feature number proportion; performing k-fold cross-validation for each parameter combination in the hyperparameter search space: in turn taking each subset as the validation set and the remaining k-1 subsets as the training set, training the random forest model and calculating the accuracy of the running power prediction on the validation set; calculating the average accuracy of each parameter combination in k-fold cross-validation; selecting the parameter combination with the highest average accuracy as the best hyperparameters; and establishing the random forest model based on the best hyperparameters.

[0040] Compared with the prior art, the beneficial effects of the present application are that the rotation resistance and the length of use data of the guide wire are collected in real time by the resistance sensor and the image monitoring device, and the micro motor running power is intelligently adjusted by combining the deep regression neural network model and the random forest model, so that the guide wire rotation operation is more accurate. The sudden increase or jamming of the resistance can be identified in time, and the micro motor power is automatically adjusted, thereby reducing the risk of blood vessel or lumen wall damage caused by excessive force, and ensuring the safety of the patient. Through intelligent adjustment, automatic control is realized, the burden of the operator is reduced, and the operation stability and consistency are improved. The guide wire is designed with TPU outer skin + hydrophilic coating, which maintains low friction while combining adaptive resistance adjustment, so that the guide wire is more smooth in advancing and rotating in a complex, narrow or curved lumen. The joint analysis of resistance value and image data combined with the machine learning model makes the micro motor power adjustment more reasonable, realizes real-time and fine control of the rotation resistance, and improves the guide wire control performance. BRIEF DESCRIPTION OF DRAWINGS

[0041] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The detailed description is merely meant to teach a person of ordinary skill a way to make and use the application. It is not meant to put limitations on the scope of the application. Moreover, the same reference numerals in the accompanying drawings refer to the same or like components throughout the several drawings. In the drawings:

[0042] Figure 1 Structure matching schematic diagram of the adaptive resistance adjustment hydrophilic coating guide wire device provided by the embodiments of the present application;

[0043] Figure 2 Schematic diagram of the adaptive resistance adjustment hydrophilic coating guide wire device provided by the embodiments of the present application.

[0044] Wherein: 1, guide wire body; 11, mark point; 2, twist device; 21, twist main body; 211, threaded section; 212, petal structure; 213, matching hole; 22, locking device; 221, locking threaded section; 222, tapered fastening section; 223, threaded connection port; 3, clamping device; 31, micro motor; 32, output shaft; 4, acquisition device; 5, control device. DETAILED DESCRIPTION

[0045] Exemplary embodiments of the present disclosure will be described in detail with reference to the drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood, and the scope of the present disclosure can be accurately conveyed to those skilled in the art. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in conjunction with the embodiments.

[0046] In some embodiments of the present application, referring to Figures 1-2 As shown, a hydrophilic coating guide wire device with adaptive resistance adjustment is proposed, comprising:

[0047] The guide wire body 1 is wrapped with a TPU outer skin, and a hydrophilic coating is applied to the surface of the TPU outer skin. The color of the TPU outer skin is a spiral zebra pattern. The TPU outer skin is also provided with a mark point 11.

[0048] The twist device 2 includes a twist main body 21 and a locking device 22. The twist main body 21 includes a threaded section 211, a petal structure 212, and a matching hole 213. The locking device 22 includes a locking threaded section 221 and a tapered fastening section 222. The threaded section 211 is connected with the locking threaded section 221 for fixed connection of the twist main body 21 and the locking device 22. The tapered fastening section 222 is used to fasten the petal structure 212. The twist device 2 is hollow inside for placing the guide wire body 1. The locking device 22 is also provided with a threaded connection port 223 connected with the catheter;

[0049] The clamping device 3 is provided with a micro motor 31 inside. The output shaft 32 of the micro motor 31 is connected with the twist device 2 through the matching hole 213.

[0050] The acquisition device 4 includes a resistance sensor and an image monitoring device. The acquisition device 4 is arranged at the outlet end of the guide wire body 1 on the right side of the locking device 22. The resistance sensor is used to monitor the resistance value in real time when the guide wire body 1 rotates. The image monitoring device is used to monitor the image data of the guide wire body 1.

[0051] The control device 5 is arranged on the rotating device 2 and is electrically connected with the micro motor 31 and the acquisition device 4. The control device 5 is used for acquiring the resistance value of the resistance sensor and the image data of the image monitoring device, and adjusting the operation power of the micro motor 31.

[0052] The control device 5 comprises:

[0053] The acquisition unit is configured to acquire the image data in the image monitoring device in real time, and acquire the resistance value of the guide wire body in rotation detected by the resistance sensor in real time based on an exponential weighted moving average method; and perform data preprocessing on the image data and the resistance value respectively;

[0054] The analysis unit is configured to analyze the use length data of the guide wire body 1 in the preprocessed image data based on edge detection, determine the resistance threshold based on a deep regression neural network model, compare the resistance threshold with the resistance value, and determine whether to adjust the operation power of the micro motor 31;

[0055] The processing unit is configured to, when it is determined to adjust the operation power of the micro motor 31, adjust the operation power of the micro motor 31 based on a random forest model, and operate the micro motor 31 at the adjusted operation power.

[0056] Specifically, the adaptive resistance-adjusted hydrophilic coating guide wire device comprises a guide wire body 1, a rotation and twist device 2, a clamping device 3, a collection device 4 and a control device 5, which work cooperatively to realize real-time monitoring and adaptive adjustment of the rotation resistance of the guide wire body 1. The guide wire body 1 is composed of a wire core, a TPU outer skin and a hydrophilic coating. The TPU outer skin has flexibility and elasticity, which can adapt to the curved path of the blood vessel or lumen. The hydrophilic coating coated on the surface reduces friction after contacting with body fluid, and improves the smoothness of propulsion and rotation. The outer skin of the guide wire body 1 is in a spiral zebra pattern, and a plurality of marker points 11 are arranged thereon, which are used to identify the position, rotation state and use length of the guide wire under image monitoring, and provide accurate analysis basis for the control device 5. The rotation and twist device 2 has a hollow structure for placing the guide wire body 1. The rotation and twist main body 21 is fixedly connected with the locking device 22 through the cooperation connection of the threaded segment 211 and the locking threaded segment 221. The petal structure 212 is fastened through the tapered fastening segment 222 to fix the guide wire body 1 and ensure the radial stability of the guide wire body 1 during rotation. The locking device 22 is also provided with a catheter threaded connection port 223, which can realize the stable interface between the guide wire body 1 and the catheter, and improve the stability and safety of the guide wire body during the propulsion and rotation process. The clamping device is internally provided with a micro motor, the output shaft of which is connected with the rotation and twist device through the cooperation hole of the rotation and twist main body to realize the rotation driving of the guide wire. The power of the micro motor can be accurately adjusted to ensure that the guide wire can smoothly pass through the narrow and curved path during rotation in the complex cavity, and can also avoid excessive force to cause damage to the lumen. The collection device 4 comprises a resistance sensor and an image monitoring device, which are installed at the outlet end of the guide wire body 1. The resistance sensor is used to monitor the resistance change of the guide wire body 1 during rotation, and the image monitoring device is used to collect the marker points and contour information of the outer skin of the guide wire body 1 to determine the use length and rotation state of the guide wire body 1. The control device 5 is arranged on the rotation and twist device 2 and is electrically connected with the micro motor 31 and the collection device 4, which is used to collect the data of the resistance sensor and the image monitoring device in real time, pre-process the collected data, and ensure the accuracy and reliability of the analysis results. The marker points and contour information of the guide wire body 1 are identified through image recognition, the use length of the guide wire body 1 is calculated, and the current rotation resistance threshold value is generated combined with a deep regression neural network model. The control device 5 compares the real-time resistance value with the threshold value, and intelligently adjusts the power of the micro motor 31 based on the random forest model to realize adaptive rotation control, thereby reducing the risk of damage to the blood vessel or lumen wall.

[0057] The control device 5 comprises a collection unit, an analysis unit and a processing unit, which cooperates to realize adaptive adjustment of the rotation resistance of the guide wire body 1. The collection unit obtains the image of the outlet end of the guide wire body 1 in real time through the image monitoring device, and collects resistance sensor data based on the exponentially weighted moving average method, denoises, crops and standardizes the image, and removes outliers and normalizes the resistance data, to provide reliable data for subsequent analysis. The analysis unit uses edge detection to extract the guide wire contour and center line, calculates the use length of the guide wire body 1, and verifies the accuracy in combination with the marker point 11 information, and then inputs the use length data into a deep regression neural network model to predict the current rotation resistance threshold, compares the real-time resistance with the threshold, and judges whether the power of the micro motor 31 needs to be adjusted. When it is determined that adjustment is needed, the processing unit calculates the optimal operating power of the micro motor 31 through a random forest model, and applies the adjusted power to the micro motor 31, to realize adaptive control of the rotation of the guide wire body 1.

[0058] In operation of the guide wire adaptive resistance adjustment device, after the guide wire body 1 is advanced to the target position, the micro motor 31 in the clamping device 3 drives the rotation device 2 to rotate the guide wire body 1, to adjust the direction of the front end of the guide wire body 1. The resistance sensor and the image monitoring device simultaneously collect the rotation resistance of the guide wire body 1 and the image information of the marker point 11, and the control device 5 pre-processes the collected data. The use length of the guide wire body 1 is identified through image recognition, and the rotation resistance threshold is calculated in combination with a deep regression neural network, the resistance value is compared in real time, and it is judged whether the power of the micro motor 31 needs to be adjusted. When the use length of the guide wire body 1 is detected to be abnormal, an alarm is issued. The adjustment of the power of the micro motor 31 is based on a random forest model, to realize adaptive control of the rotation of the guide wire body 1. When the resistance of the guide wire body 1 is detected to suddenly increase, the operating power of the micro motor 31 is automatically adjusted.

[0059] It can be understood that the rotation resistance of the guide wire body 1 and the marker point 11 information are collected in real time through the resistance sensor and the image monitoring device, the control device 5 intelligently adjusts the power of the micro motor 31 according to data analysis, to realize accurate control of the rotation of the guide wire body 1, and to reduce the subjective judgment error of the operator. When the guide wire body 1 encounters sudden resistance increase or jamming, the micro motor power can be adjusted in real time, to reduce the excessive friction and potential damage risk of the guide wire body 1 to the blood vessel or lumen wall. The guide wire body 1 is covered with a TPU outer skin and a hydrophilic coating, in combination with a spiral zebra pattern and a marker point 11 design, to smoothly advance and rotate in a complex, narrow or curved lumen, to improve the operation efficiency. Through the collection unit, the resistance value and image data are obtained in real time, the analysis unit predicts the resistance threshold in combination with a deep regression neural network, and the processing unit adjusts the power of the micro motor 31 based on a random forest model, to realize accurate control of the rotation of the guide wire body 1, and to reduce the human judgment error. The power of the micro motor 31 can be adjusted when the resistance suddenly increases or the guide wire is jammed, to reduce the excessive friction and potential damage risk of the guide wire body 1 to the blood vessel or lumen wall.

[0060] In some embodiments of the present application, the acquisition unit acquires the resistance value detected by the resistance sensor in real time based on the exponential weighted moving average method, including:

[0061] The acquisition unit acquires the resistance value of the resistance sensor at each moment, and generates continuous data from the resistance value; the resistance value is smoothed, and the latest resistance value is given a high weight, and the past resistance value is given a low weight.

[0062] Specifically, the acquisition unit first records the resistance value collected by the resistance sensor at each moment continuously to form continuous time series data. The resistance values are smoothed to reduce the influence of sensor jitter or short-term interference on control judgment. The resistance value at the latest moment is given a higher weight, and the past resistance value is given a lower weight. The weighted average value is calculated by a pre-set exponential weighting formula, so that the data is more sensitive to the latest resistance change, while keeping the overall curve smooth.

[0063] For example, during the rotation of the guidewire body 1, the resistance sensor continuously collects the resistance value of the guidewire body 1. In order to ensure that the data is sensitive to the latest change and smooth overall, the acquisition unit uses the exponential weighted moving average method for processing. When the resistance values collected continuously for 5 seconds are 10, 12, 15, 20 and 18 N·cm, the smoothing coefficient α=0.6 is selected, and the smoothed values can be calculated as 10, 11.2, 13.48, 17.39 and 17.76 N·cm, respectively.

[0064] In some embodiments of the present application, the acquisition unit respectively pre-processes the image data and the resistance value, including:

[0065] The image data preprocessing includes denoising to eliminate random noise in the acquisition process; standardizing the image to eliminate background differences; region cropping or scaling to focus on the area where the guidewire is located, reduce irrelevant information interference, and generate clear image data;

[0066] The resistance value preprocessing includes filtering out abnormal values to eliminate sudden abnormal values caused by sensor jitter or short-term interference; normalizing to map the resistance data to a unified numerical range and eliminate dimension differences.

[0067] Specifically, the acquisition unit systematically pre-processes the image data and resistance values obtained during the rotation of the guide wire body 1 to ensure the accuracy and reliability of subsequent analysis. For image data, first, denoising processing is performed, usually using Gaussian filtering, median filtering or bilateral filtering, etc. to remove random noise caused by camera sensor noise, light fluctuations or environmental interference, thereby improving image clarity. Then, standardization processing is performed, including uniform brightness, contrast and color range, to eliminate the influence of different shooting conditions or background differences on the analysis results, so that the images remain consistent in brightness and color. Subsequently, the region where the guide wire body 1 is located is cropped or scaled, focusing on the key region where the guide wire body 1 and its marker points are located, while reducing the irrelevant background area, reducing interference information and improving the accuracy of edge detection and center line extraction. After these processes, the generated image is clear, focused and suitable for subsequent guide wire body 1 contour extraction, length calculation and marker point identification. For resistance value data, the acquisition unit first filters out abnormal values to eliminate sudden abnormal data caused by sensor jitter, short-term interference or mechanical vibration to avoid the influence of these abnormal values on the judgment. Subsequently, normalization processing is performed to map the resistance values to a unified numerical range (such as 0~1), eliminating differences in different dimensions or units, making subsequent analysis and neural network model input more stable and reliable.

[0068] It can be understood that the clear and focused guide wire image and smooth and normalized resistance data enable the analysis unit to more accurately extract the guide wire contour, calculate the length of use and judge the resistance threshold, thereby improving the accuracy of the guide wire body 1 rotation control.

[0069] In some embodiments of the present application, when the analysis unit analyzes the guide wire body length of use data in the pre-processed image data based on edge detection, it includes:

[0070] The analysis unit extracts the contour information of the guide wire body 1 in the image data, generates the center line of the guide wire body 1 based on the contour information, and performs pixel-level tracking and cumulative calculation along the center line to obtain the first initial length of use data of the guide wire body 1;

[0071] The analysis unit identifies the number of remaining marker points 11 in the image data, and calculates the second initial length of use data of the guide wire body 1 by comparing the distance between the marker points 11 and the initial number;

[0072] The length difference between the first initial length of use data and the second initial length of use data is calculated, and the length difference is compared with the length difference threshold value, and the analysis unit determines whether the guide wire body 1 length of use recognition data is correct;

[0073] When the length difference is greater than the length difference threshold value, the analysis unit determines that the guide wire body 1 length of use recognition is incorrect, and issues an alarm;

[0074] When the length difference value is less than or equal to the length difference threshold value, the analysis unit determines that the recognized use length of the guide wire body 1 is correct.

[0075] Specifically, the analysis unit performs edge detection on the pre-processed image data to extract the contour information of the guide wire body 1. By continuously tracking the contour at the pixel level, a center line of the guide wire body 1 is generated, and the length of the guide wire is calculated by accumulating along the center line, thereby obtaining first initial use length data. This method can accurately reflect the actual spread of the guide wire body 1 in the image and exclude the influence of background and interference information. Secondly, the analysis unit identifies the remaining marker points in the image. By counting the number of marker points and combining the distance between the marker points and the total number of initial marker points of the guide wire, the second initial use length data of the guide wire body 1 is calculated. This method can be used as a verification method for the first initial use length to increase the reliability of the identification. Subsequently, the analysis unit compares the first initial use length data and the second initial use length data, calculates the length difference value therebetween, and compares the length difference value with a preset length difference threshold value: if the length difference value is greater than the threshold value, it is determined that the use length recognition is incorrect, and the analysis unit will issue a warning to prompt the operator that there may be an identification error or guide wire abnormality; if the length difference value is less than or equal to the threshold value, it is determined that the use length recognition is correct, and subsequent resistance analysis and micro motor 31 control can continue.

[0076] It can be understood that by combining the contour center line accumulation calculation and the marker point number estimation, a double verification mechanism is formed, which improves the accuracy of the guide wire use length recognition. When the length difference value exceeds the preset threshold value, a warning can be issued in time to prompt that there may be an identification error or guide wire abnormality, thereby avoiding blood vessel or cavity damage caused by incorrect judgment.

[0077] In some embodiments of the present application, before the analysis unit determines the resistance threshold value based on the deep regression neural network model using the use length data, the analysis unit includes:

[0078] The analysis unit constructs a historical feature vector from a plurality of historical initial use length data sets, each of which includes historical first initial use length data and historical second initial use length data; and constructs a historical information set from the historical feature vector and the resistance threshold value, each historical feature vector corresponding to a resistance threshold value;

[0079] The historical information set is sampled according to a preset proportion to obtain a training subset and a test subset;

[0080] A pre-established deep regression neural network model is obtained, and the deep regression neural network model is iteratively trained according to the training subset. The iteratively trained deep regression neural network model is evaluated according to the test subset, and whether to stop the iterative training is determined according to the evaluation value.

[0081] Specifically, the analysis unit collects a plurality of historical guidewire usage length data sets, each data set including historical first initial usage length data (obtained based on guidewire center line cumulative calculation) and historical second initial usage length data (obtained based on the number of marker points). Each set of historical length data is combined into a historical feature vector, and is paired with the actual corresponding guidewire rotation resistance threshold to form a complete historical information set, so that the neural network learns the non-linear relationship between the usage length and the resistance threshold. The historical information set is divided according to a preset proportion (for example, 70% training set, 30% test set) to generate a training subset and a test subset. The training subset is used for iterative learning of the neural network model, and the test subset is used to evaluate the training effect and the model generalization ability, to ensure the prediction accuracy of the model under new data. A pre-established deep regression neural network model is obtained, and the training subset is input into the model for iterative training. During training, the network calculates the error between the predicted resistance threshold and the actual resistance threshold through forward propagation, and updates the weight parameters through back propagation, so that the model is continuously optimized. After each round of training, the test subset is input into the model for evaluation, and the error index (such as mean square error or coefficient of determination R²) is calculated to determine whether the model prediction performance meets the preset standard. If the evaluation value does not meet the requirement, iterative training continues; if the standard is met, training is stopped, and the final neural network model that can be used for real-time resistance threshold prediction is obtained.

[0082] In some embodiments of the present application, when the analysis unit determines the resistance threshold based on the usage length data using the deep regression neural network model, it includes:

[0083] The deep regression neural network model includes an input layer, 8 hidden layers, and an output layer; the input layer is used to receive an initial feature vector composed of an initial usage length data set; the hidden layers extract the complex non-linear relationship between the usage length and the resistance change through the weighted sum of neurons and non-linear activation function mapping, and convert the input features into high-order abstract features layer by layer; the output layer maps the high-order abstract features to the resistance threshold under the current guidewire usage length data.

[0084] In some embodiments of the present application, when the hidden layer extracts the complex non-linear relationship between the usage length and the resistance change through the weighted sum of neurons and non-linear activation function mapping, it includes:

[0085] Each neuron of the hidden layer multiplies each input feature by a corresponding weight coefficient, and sums the weighted results to obtain the weighted sum of the neuron, which is then mapped by a non-linear activation function to convert the linear combination of inputs into non-linear outputs, so that the non-linear activation function captures the non-linear change law between the usage length and the resistance; the output of the non-linear activation function mapping is passed to the next layer until the last hidden layer forms high-order abstract features.

[0086] Specifically, the deep regression neural network model is composed of an input layer, 8 hidden layers, and an output layer. The structure is designed to capture the complex nonlinear relationship between the guide wire usage length and the rotation resistance and map it to the resistance threshold that can be used for micro motor power regulation. The input layer receives an initial feature vector composed of an initial usage length dataset, which includes a first initial usage length (calculated based on guide wire centerline pixel-level accumulation) and a second initial usage length (obtained based on the number of marker points), which reflect the actual usage state of the guide wire during the rotation operation and provide basic input for resistance threshold prediction. Each hidden layer is composed of a number of neurons, each of which processes the input features by weighting and processing: weighted sum, each neuron multiplies each input feature by the corresponding weight coefficient and sums all the weighted results, while adding a bias term to form the weighted sum of the neuron, realizing the linear combination of input features. Nonlinear mapping, the weighted sum is mapped through a nonlinear activation function (such as ReLU, Sigmoid, Tanh, etc.), which converts the linear combination of inputs into a nonlinear output, thereby capturing the complex nonlinear relationship between the guide wire usage length and the rotation resistance. Layer-by-layer transmission, the output of the current layer of neurons is used as the input of the next layer of hidden layers, and after 8 layers of layer-by-layer mapping and high-order feature abstraction, the deep regression neural network model can extract the multidimensional influence of usage length change on rotation resistance and realize deep feature representation. The high-order abstract features extracted by the hidden layer are sent to the output layer, which maps them to the rotation resistance threshold under the current guide wire usage length. This threshold represents the maximum resistance allowed for safe rotation of the micro motor 31 under the current guide wire body 1 state, and is an important basis for controlling the power regulation of the micro motor 31. Through the weighted sum and nonlinear mapping of multiple hidden layers, as well as high-order feature abstraction, the deep regression neural network model can accurately capture the nonlinear relationship between the guide wire body 1 usage length and the rotation resistance, realize real-time, dynamic and adaptive resistance threshold prediction. During operation, when the guide wire length or operating conditions change, the micro motor 31 output power can be adjusted in real time to ensure smooth rotation of the guide wire body 1.

[0087] It can be understood that the complex nonlinear relationship between the guide wire body 1 usage length and the rotation resistance can be accurately captured, and the dynamic adjustment of the resistance threshold is realized through data-driven, ensuring that the guide wire body 1 rotation process can be stably performed under different operating conditions.

[0088] In some embodiments of the present application, when the analysis unit compares the resistance threshold with the resistance value and determines whether to adjust the operating power of the micro motor, it includes:

[0089] When the resistance value is greater than or equal to the resistance threshold, the analysis unit determines to adjust the operating power of the micro motor;

[0090] When the resistance value is less than the resistance threshold, the analysis unit determines not to adjust the operating power of the micro motor.

[0091] Specifically, after receiving the resistance value collected by the resistance sensor in real time, the analysis unit first compares the resistance value with the resistance threshold value obtained based on the neural network model or empirical calculation. When the comparison result shows that the resistance value is greater than or equal to the resistance threshold value, it indicates that the friction or resistance borne by the guide wire body 1 during rotation or advancement has approached the upper limit of safety, and if not adjusted in time, it may cause the guide wire movement to be blocked, the control accuracy to be reduced, or even cause the blood vessel or cavity wall to be damaged. Therefore, the analysis unit determines that the operating power of the micro motor 31 needs to be adjusted to reduce the rotation or advancement speed, thereby reducing the friction between the guide wire and the pipe wall or tissue, and ensuring the smoothness and safety of the operation process. On the contrary, when the comparison result shows that the resistance value is less than the resistance threshold value, it indicates that the operating state of the guide wire body 1 is within a reasonable range and will not affect the operation safety and the service life of the guide wire. Therefore, the analysis unit determines that the operating power of the micro motor 31 does not need to be adjusted to avoid the problems of increased energy consumption and delayed response caused by frequent adjustment, thereby ensuring the stability and reliability of the device.

[0092] In some embodiments of the present application, when it is determined to adjust the operating power of the micro motor, the analysis unit adjusts the operating power of the micro motor based on the random forest model, including:

[0093] Using cross-validation combined with grid search to find the best parameters of the random forest model, and establishing the random forest model;

[0094] Using the training set to fit the random forest model; bringing the test set into the random forest model and calculating the accuracy of the operating power;

[0095] When the accuracy reaches the preset accuracy threshold, the operating power of the current micro motor is obtained according to the resistance value.

[0096] Specifically, the method of cross-validation combined with grid search is used to optimize the key parameters of the random forest model, such as the number of trees, the maximum depth of the tree, the minimum sample size of the leaf node, the feature selection method, etc., to find the optimal parameter combination, so as to establish a random forest model suitable for the target task; this optimization process can effectively avoid overfitting and improve the generalization ability of the model in practical application. Secondly, the preprocessed training dataset is used to fit the random forest model. During the training process, the model learns the mapping relationship between the input resistance value and the corresponding micro motor operating power through the ensemble learning of multiple decision trees, effectively capturing the complex nonlinear characteristics. Then, the independent test set is input into the trained random forest model, and the accuracy between the output result and the true label is calculated. Through this verification process, the reliability and stability of the model in predicting the operating power of the micro motor are determined. When the accuracy reaches or exceeds the preset accuracy threshold, the analysis unit will take the real-time collected resistance value as input, call the trained random forest model, and output the micro motor operating power corresponding to the resistance value.

[0097] It can be understood that the random forest model can effectively model the nonlinear relationship and has strong anti-overfitting ability; during the training process, the training set is used to fit the model, and then the test set is brought in for accuracy calculation, ensuring high matching degree of power prediction and actual resistance. By setting a preset accuracy threshold, when the prediction accuracy of the model reaches or exceeds the threshold, the operating power of the micro motor is output based on the resistance value, avoiding errors caused by unstable or low-precision models. By accurately adjusting the power of the micro motor 31, the normal operation of the guide wire body 1 can be ensured, power waste is reduced, energy consumption is reduced, and damage caused by excessive driving is avoided.

[0098] Further, the best parameters of the random forest model are found using cross-validation combined with grid search, and the random forest model is established, including:

[0099] The training set is divided into k mutually exclusive subsets; define the search space of the hyperparameters of the random forest model, including the range of the number of trees, the range of the maximum depth, the range of the minimum sample splitting threshold, and the range of the maximum feature number ratio; perform k-fold cross-validation on each parameter combination in the hyperparameter search space: in turn, each subset is used as the validation set, and the remaining k-1 subsets are used as the training set, the random forest model is trained, and the accuracy of the operating power prediction on the validation set is calculated; calculate the average accuracy of each parameter combination in k-fold cross-validation; select the parameter combination with the highest average accuracy as the best hyperparameters; establish the random forest model based on the best hyperparameters.

[0100] Specifically, the training set data is randomly divided into k mutually exclusive subsets, ensuring that each subset is as balanced as possible in terms of sample size and distribution characteristics to improve the generalization ability of the model. The hyperparameter search space of the random forest model is set, which is used to limit the range of main adjustment parameters of model performance, including: the number of decision trees (such as 50 to 500), the maximum depth of the tree (such as 5 to 50 layers), the minimum sample splitting threshold (such as 2 to 20), and the maximum feature number ratio (such as 0.2 to 1.0). Perform k-fold cross-validation on each combination of hyperparameters in the search space. In each cross-validation, one of the subsets is selected as the validation set, and the remaining k-1 subsets are used as the training set to train the random forest model. After training, the validation set is input into the model, and the accuracy of the model in predicting the running power of the micro motor on the validation set is calculated. Repeat the above process k times so that each subset is used as a validation set once, and finally calculate the average accuracy of the hyperparameter combination in k cross-validations. After completing the cross-validation of all hyperparameter combinations, compare the average accuracy of all combinations, and select the parameter combination with the highest average accuracy as the optimal hyperparameters. Based on the optimal hyperparameters, the random forest model is reconstructed and trained to form the final power adjustment prediction model.

[0101] It can be understood that by evaluating the impact of parameter combinations on model performance, overfitting or underfitting problems are avoided, thereby ensuring the accuracy and stability of the prediction results of the micro motor running power adjustment.

[0102] In summary, by real-time acquisition of the rotating resistance and length of use of the guide wire body 1 through the resistance sensor and image monitoring device, and combining the deep regression neural network model and the random forest model to intelligently adjust the running power of the micro motor, the rotating operation of the guide wire body 1 is more precise. It can timely identify sudden increases in resistance or sticking situations and automatically adjust the power of the micro motor 31, thereby reducing the risk of damage to blood vessels or lumen walls caused by excessive force and ensuring patient safety. Through intelligent adjustment, automation control is achieved, reducing the burden on the operator and improving operation stability and consistency. The guide wire is designed with a TPU outer skin + hydrophilic coating, which maintains low friction while combining adaptive resistance adjustment, making the guide wire more smooth in complex, narrow, or curved cavities. The joint analysis of resistance values and image data combined with machine learning models makes the power adjustment of the micro motor 31 more reasonable, achieving real-time and fine control of rotating resistance and improving the control performance of the guide wire body 1.

[0103] Those skilled in the art will appreciate that embodiments of the application can be supplied as a method, a system or a computer program product. Accordingly, the application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the application can be embodied in the form of a computer program product on one or more computer readable storage media (including, without limitation, magnetic disks; optical disks; magneto-optical disks; ROMs; flash memory; etc.) having computer usable program code embodied therein.

[0104] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks

[0105] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks

[0106] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks

[0107] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing and illustrating, not limiting, the technical solutions of the present application, and although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and any modifications or equivalent replacements made without departing from the spirit and scope of the present application shall be included in the protection scope of the claims of the present application.

Claims

1. A hydrophilic coated guide wire device with adaptive resistance adjustment, characterized in that, include: The guidewire body has a TPU outer sheath covering the core surface, and a hydrophilic coating is applied to the surface of the TPU outer sheath. The TPU outer sheath has a spiral zebra stripe pattern and is also provided with marking points. A rotary device includes a rotary body and a locking device; the rotary body includes a threaded section, a petal structure, and a mating hole; the locking device includes a locking threaded section and a tapered fastening section. A clamping device, wherein a micro motor is provided inside the clamping device, and the output shaft of the micro motor is connected to the rotary device through the mating hole; The data acquisition device includes a resistance sensor and an image monitoring device; A control device is mounted on a rotary device and electrically connected to the micro motor and the data acquisition device. The control device includes a data acquisition unit, an analysis unit, and a processing unit; The acquisition unit is configured to acquire image data from the image monitoring device in real time, and to acquire the resistance value detected by the resistance sensor when the guide wire body rotates in real time based on the exponentially weighted moving average method; and to perform data preprocessing on the image data and the resistance value respectively. The analysis unit is configured to analyze the guidewire body usage length data in the preprocessed image data based on edge detection, and determine the resistance threshold based on the usage length data using a deep regression neural network model; compare the resistance threshold with the resistance value to determine whether to adjust the operating power of the micro motor; The processing unit is configured to adjust the operating power of the micromotor based on a random forest model when it is determined that the operating power of the micromotor should be adjusted, and to run the micromotor with the adjusted operating power.

2. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 1, characterized in that, When the acquisition unit acquires the resistance value detected by the resistance sensor during the rotation of the guidewire body in real time based on the exponentially weighted moving average method, it includes: The acquisition unit acquires the resistance value of the resistance sensor at each moment and generates continuous data from the resistance value; the resistance value is then smoothed.

3. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 2, characterized in that, When the acquisition unit performs data preprocessing on the image data and the resistance value, it includes: The image data preprocessing includes denoising; image standardization; and region cropping or scaling. The resistance value preprocessing includes outlier filtering to remove sudden outliers caused by sensor jitter or short-term interference; and normalization to map the resistance data to a uniform numerical range.

4. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 3, characterized in that, When the analysis unit analyzes the guidewire body length data in the preprocessed image data based on edge detection, it includes: The analysis unit extracts the contour information of the guidewire body from the image data, generates the centerline of the guidewire body based on the contour information, performs pixel-level tracking and cumulative calculation along the centerline, and obtains the first initial usage length data of the guidewire body. The analysis unit identifies the number of remaining marker points in the image data and calculates the second initial usage length data of the guide wire body by comparing the spacing between the marker points with the initial number. The analysis unit calculates the length difference between the first initial usage length data and the second initial usage length data, and compares the length difference with a length difference threshold to determine whether the guidewire body usage length identification data is correct. When the length difference is greater than the length difference threshold, the analysis unit determines that the guidewire body is incorrectly identified in terms of length and issues an alarm. When the length difference is less than or equal to the length difference threshold, the analysis unit determines that the guidewire body is correctly identified in terms of length.

5. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 4, characterized in that, Before the analysis unit determines the resistance threshold using the length data based on a deep regression neural network model, it includes: The analysis unit constructs historical feature vectors from several historical initial usage length datasets, each of which includes historical first initial usage length data and historical second initial usage length data; and constructs a historical information set by combining the historical feature vectors with resistance thresholds, with each historical feature vector corresponding to a resistance threshold. The historical information set is sampled according to a preset ratio to obtain a training subset and a test subset; A pre-defined deep regression neural network model is obtained, and the deep regression neural network model is iteratively trained according to the training subset. The iteratively trained deep regression neural network model is evaluated according to the test subset, and it is determined whether to stop iterative training based on the evaluation value.

6. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 5, characterized in that, When the analysis unit determines the resistance threshold using the length data based on a deep regression neural network model, it includes: The deep regression neural network model includes an input layer, eight hidden layers, and an output layer. The input layer receives an initial feature vector composed of an initial usage length dataset. The hidden layers extract the complex nonlinear relationship between usage length and resistance change through weighted aggregation of neurons and mapping using nonlinear activation functions, transforming the input features layer by layer into higher-order abstract features. The output layer maps the higher-order abstract features to the resistance threshold under the current guidewire usage length data.

7. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 6, characterized in that, The hidden layer extracts complex nonlinear relationships between length and resistance changes by using weighted and nonlinear activation function mapping of neurons, including: Each neuron in the hidden layer multiplies each input feature by a corresponding weight coefficient, and the weighted results are summed to obtain the neuron's weighted sum. The neuron then maps the weighted sum through a nonlinear activation function, converting the linear combination of inputs into a nonlinear output, so that the nonlinear activation function captures the nonlinear variation between length and resistance. The output of the nonlinear activation function is passed to the next layer until the last hidden layer forms the higher-order abstract feature.

8. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 7, characterized in that, When the analysis unit compares the resistance threshold with the resistance value to determine whether to adjust the operating power of the micromotor, it includes: When the resistance value is greater than or equal to the resistance threshold, the analysis unit determines to adjust the operating power of the micro motor; When the resistance value is less than the resistance threshold, the analysis unit determines that the operating power of the micro motor should not be adjusted.

9. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 8, characterized in that, When it is determined that the operating power of the micromotor should be adjusted, the analysis unit adjusts the operating power of the micromotor based on the random forest model, including: The optimal parameters for the random forest model are found by combining cross-validation with grid search, and the random forest model is built. Fit the random forest model using the training set; feed the test set into the random forest model and calculate the accuracy of the running power; When the accuracy reaches a preset accuracy threshold, the current operating power of the micro motor is obtained based on the resistance value.

10. The adaptive resistance adjustment hydrophilic coating guide wire device according to claim 9, characterized in that, The optimal parameters for a random forest model are found using a combination of cross-validation and grid search. Building the random forest model includes: The training set is divided into k mutually exclusive subsets; the hyperparameter search space of the random forest model is defined, and the hyperparameters include the range of the number of trees, the range of the maximum depth, the range of the minimum sample splitting threshold, and the range of the maximum feature ratio; k-fold cross-validation is performed on each parameter combination in the hyperparameter search space: each subset is used as the validation set, and the remaining k-1 subsets are used as the training set to train the random forest model and calculate the accuracy of power prediction on the validation set; the average accuracy of each parameter combination in k-fold cross-validation is calculated; the parameter combination with the highest average accuracy is selected as the optimal hyperparameter; the random forest model is built based on the optimal hyperparameter.

Images