CN-120919493-B - Intelligent PICC catheter and positioning system for tumor chemotherapy

CN120919493BCN 120919493 BCN120919493 BCN 120919493BCN-120919493-B

Abstract

The invention discloses an intelligent PICC catheter and a positioning system for tumor chemotherapy, and belongs to the technical field of biomedical engineering. The method solves the problem of inaccurate puncture caused by certain subjectivity in selection of puncture points in the prior art, can screen out the characteristic with the most influence on the predicted puncture result by carrying out importance assessment on the blood vessel image characteristic and the blood flow parameter characteristic, thereby reducing the complexity of a model, improving the running efficiency and generalization capability of the model, carrying out weighted average on input data according to the characteristic importance, enabling a random forest model to pay more attention to important characteristics, improving the accuracy and the reliability of model prediction, and enabling a positioning system to be more accurate and efficient when recommending the optimal puncture points by cooperative work of a plurality of modules, thereby effectively improving the puncture success rate, reducing the complication risk, providing powerful decision support for medical staff and improving the treatment experience and the safety of patients.

Inventors

ZUO GUOQIN
XU AN
FAN ZHENPING
Pang Xiangmin
KOU SHUMIN
SUN SHASHA

Assignees

中国人民解放军总医院第二医学中心

Dates

Publication Date: 20260505
Application Date: 20250818

Claims (6)

1. A intelligent PICC pipe positioning system for tumor chemotherapy realizes through the intelligent PICC pipe that is used for tumor chemotherapy, its characterized in that, the intelligent PICC pipe positioning system for tumor chemotherapy includes: The characteristic extraction module is configured to collect historical data, extract blood vessel image characteristics and blood flow parameter characteristics in the historical data, wherein the blood vessel image characteristics comprise the shape, the inner diameter and the depth of a blood vessel, extract the central line and the outline of the blood vessel and calculate the length and the curvature of the blood vessel; extracting the average value, the maximum value and the minimum value of the blood flow velocity, and calculating the change rate and the stability index of the blood flow velocity; The intelligent analysis module is configured to train and test historical data, blood vessel image characteristics and blood flow parameter characteristics of the historical data through a random forest model, and output coordinates and recommended reasons of optimal puncture points, wherein the recommended reasons include blood vessel inner diameter, blood flow speed and blood vessel depth as key factors; the importance evaluation module is configured to output the average non-purity reduction value of each feature through the random forest model after the random forest model is trained, and normalize the average non-purity reduction value of each feature to obtain an importance result of each feature; The data weighting module is configured to perform weighted average on the input data of each feature according to the importance of the feature to obtain weighted input data, and the weight corresponding to each feature is further compensated by the following method, which comprises the following steps: Acquiring an average non-purity reduction value corresponding to each feature according to the average non-purity reduction value corresponding to each feature; the method comprises the steps of extracting the original weight corresponding to each feature, wherein the original weight ratio corresponding to each feature is consistent with the importance result corresponding to each feature; Comparing the standard deviation of the weight value of the original weight corresponding to each feature with a preset standard deviation threshold, and when the standard deviation of the weight value of the original weight corresponding to the feature exceeds the preset standard deviation threshold, compensating and adjusting the original weight corresponding to the feature and replacing the original weight with the compensated and adjusted weight value, wherein the compensating and adjusting the original weight corresponding to the feature comprises the following steps: Extracting clinical correlation coefficient between feature i and feature j from the medical knowledge graph, wherein the clinical correlation coefficient has a value range of ; Carrying out ratio processing on the standard deviation of the weight value corresponding to each feature and a preset standard deviation threshold value to obtain a standard deviation ratio parameter Wherein, the method comprises the steps of, Representing standard deviation ratio parameters corresponding to the ith feature; Representing a preset standard deviation threshold; representing the standard deviation of the weight value corresponding to the ith feature; Performing compensation adjustment by combining the standard deviation ratio parameters corresponding to each feature with the original weights of the clinical correlation coefficient pairs between the feature i and the feature j; The intelligent PICC catheter for tumor chemotherapy comprises a catheter main body (1) and an external component (2), and is characterized in that one end of the catheter main body (1) is connected with the external component (2) to form an integral structure of the intelligent PICC catheter, a miniature magnetic head (3) and a high-frequency linear probe (4) are embedded in a superposition manner on the tip of the catheter main body (1), and a miniature electrode (5) is arranged in the catheter main body (1); The miniature magnetic head (3) is made of high-performance magnetic materials and is configured to generate magnetic field signals; the microelectrode (5) is connected with the tip of the catheter main body (1) and is configured to collect an electric signal generated when the tip of the catheter main body (1) contacts the heart chamber in real time; the high-frequency linear probe (4) is configured to acquire cross section and longitudinal section images of blood vessels in real time, automatically identify the depth, the inner diameter and the blood flow velocity of the blood vessels through a Doppler mode, predict the identification result based on a random forest model, and recommend an optimal puncture point.
2. The intelligent PICC catheter for tumor chemotherapy according to claim 1, wherein the miniature magnetic head (3) connects the magnetic head with a signal transmission interface at the tail end of the catheter through an internal lead wire and transmits magnetic field signals to an external sensor, the external sensor consists of a plurality of high-sensitivity magnetic field sensors distributed at specific positions on the surface of a patient body and is configured to sense magnetic field signals generated by the miniature magnetic head (3) in real time, convert the magnetic field signals into electric signals and transmit the electric signals to a host computer for processing and analysis based on a wireless mode; The microelectrode (5) is connected with an external electrocardio monitoring device through an internal lead, the electrocardio monitoring device consists of an electrocardio amplifier, a filter and a signal processor, and is configured to process an electric signal generated when the tip of the catheter main body (1) contacts a heart cavity in real time, form an electrocardiogram signal after being amplified, filtered and processed, and transmit the electrocardiogram signal to a host machine for analysis and judgment based on a wireless mode.
3. The intelligent PICC catheter positioning system for tumor chemotherapy as defined in claim 1, further comprising a data labeling module and a feature selection module; The data marking module is configured to mark the puncture result and the complications in the historical data, wherein the puncture result marking comprises marking the puncture result as success or failure, and recording the coordinates of the puncture point for successful puncture; the feature selection module is configured to select features with higher importance from the blood vessel image features and the blood flow parameter features according to the importance of the features, to serve as input features of the random forest model, and to select marked puncture results and puncture point coordinates as target variables of the random forest model.
4. The intelligent PICC catheter localization system for tumor chemotherapy of claim 1, wherein the system further comprises a visualization interface and a data processing module: the visual interface is configured to display the blood vessel image and the recommended puncture point in real time based on the monitoring screen and provide visual guidance of the puncture path for the PICC catheter; the data processing module is configured to delete repeated records in the historical data, fill up missing blood flow parameters or image data, process abnormal data by using a Z-score statistical method, and divide the processed historical data into a training set and a testing set.
5. The intelligent PICC catheter localization system for tumor chemotherapy of claim 1, wherein the intelligent analysis module comprises: The model training module is configured to set parameters of the random forest model, train the random forest model by using training set data, evaluate the performance of the random forest model by using test set data, calculate the accuracy, recall rate and F1 score index of the random forest model on the test set, analyze the confusion matrix of the random forest model, and evaluate the prediction capability of the random forest model on different data.
6. The intelligent PICC catheter positioning system for tumor chemotherapy as defined in claim 1, further comprising an actual application module and a user feedback module; The practical application module is configured to apply the trained random forest model to practical chemotherapy, acquire blood vessel images and blood flow parameters in real time, output optimal puncture point coordinates and recommended reasons by using the random forest model, display recommended puncture points and recommended reasons based on a visual interface, and provide visual guidance for PICC catheter positioning; And the user feedback module is configured to adjust parameters of the random forest model according to feedback and real-time data of medical staff and optimize performance of the random forest model.

Description

Intelligent PICC catheter and positioning system for tumor chemotherapy Technical Field The invention relates to the technical field of biomedical engineering, in particular to an intelligent PICC catheter and a positioning system for tumor chemotherapy. Background Central venous catheter (PICC) placement via peripheral veins is a common infusion modality during tumor chemotherapy. PICC pipe can provide long-term, stable venous access for the patient, reduces the misery of repeated puncture, reduces the infection risk simultaneously. However, the conventional PICC catheterization technique has some limitations, mainly in the following aspects: The traditional method mainly depends on the experience and manual operation of medical staff to select the puncture point, and lacks accurate image and blood flow parameter support, so that the selection of the puncture point has certain subjectivity; Because of inaccurate puncture point selection, the puncture success rate may be affected, especially for patients with poor vascular conditions, and in addition, complications such as hematoma, nerve injury and the like may occur in the puncture process, thereby increasing pain and treatment risk of the patients. Therefore, the existing needs are not met, for which we propose an intelligent PICC catheter and localization system for tumor chemotherapy. Disclosure of Invention The invention aims to provide an intelligent PICC catheter and a positioning system for tumor chemotherapy, which are used for carrying out importance assessment through blood vessel image features and blood flow parameter features, screening out features with the most influence on a predicted puncture result, reducing model complexity, improving the running efficiency and generalization capability of a model, carrying out weighted average on input data according to feature importance, enabling a random forest model to pay more attention to important features, improving the accuracy and reliability of model prediction, enabling the positioning system to be more accurate and efficient when recommending an optimal puncture point through cooperative work of a plurality of modules, effectively improving the puncture success rate, reducing the risk of complications, providing powerful decision support for medical staff, improving the treatment experience and safety of patients, and solving the problems in the background art. In order to achieve the above purpose, the present invention provides the following technical solutions: A intelligent PICC pipe positioning system for tumor chemotherapy realizes through the intelligent PICC pipe that is used for tumor chemotherapy, its characterized in that, the intelligent PICC pipe positioning system for tumor chemotherapy includes: The characteristic extraction module is configured to collect historical data, extract blood vessel image characteristics and blood flow parameter characteristics in the historical data, wherein the blood vessel image characteristics comprise the shape, the inner diameter and the depth of a blood vessel, extract the central line and the outline of the blood vessel and calculate the length and the curvature of the blood vessel; extracting the average value, the maximum value and the minimum value of the blood flow velocity, and calculating the change rate and the stability index of the blood flow velocity; The intelligent analysis module is configured to train and test historical data, blood vessel image characteristics and blood flow parameter characteristics of the historical data through a random forest model, and output coordinates and recommended reasons of optimal puncture points, wherein the recommended reasons include blood vessel inner diameter, blood flow speed and blood vessel depth as key factors; the importance evaluation module is configured to output the average non-purity reduction value of each feature through the random forest model after the random forest model is trained, and normalize the average non-purity reduction value of each feature to obtain an importance result of each feature; The data weighting module is configured to perform weighted average on the input data of each feature according to the importance of the feature to obtain weighted input data, and the weight corresponding to each feature is further compensated by the following method, which comprises the following steps: Acquiring an average non-purity reduction value corresponding to each feature according to the average non-purity reduction value corresponding to each feature; the method comprises the steps of extracting the original weight corresponding to each feature, wherein the original weight ratio corresponding to each feature is consistent with the importance result corresponding to each feature; Comparing the standard deviation of the weight value of the original weight corresponding to each feature with a preset standard deviation threshold, and when the standard deviation of the weight