Pulse Signal Acquisition And Analysis For Liver Cirrhosis And The Design Of Multi-point Pulse Signal Collector

Pulse diagnosis is the core part of TCM diagnosis.The pulse signal contains various physiological arid pathological information of the human body.The main clinical manifestations of liver cirrhosis are liver function damage and portal hypertension,which belong to the category of "swelling" and "Zhengjia" in TCM pulse diagnosis,with certain TCM theories.In this thesis,the pulse data of patients with liver cirrhosis and healthy people are collected by the pulse data collector,and the wavelet threshold denoising method is used to preprocess the data.The time-frequency domain characteristics and nonlinear characteristics of pulse signals are extracted,and an improved forest model classification algorithm is designed to realize the early warning of cirrhosis.Aiming at the problems encountered in the process of data acquisition and the shortcomings of pulse collection instrument at present,a portable three-part pulse collection instrument is designed in combination with the practical engineering application project to realize the synchronous acquisition of three parts of radial artery:"cun,guan and chi".The main work of this thesis is divided into the following three parts:The first part is the collection and processing of pulse signal.The first is clinical data collection,which mainly formulates a standardized collection process,including the training of the collectors,the standards for the collection and discharge of the collected people,the collection location and the collection posture,etc.;the second is the preprocessing of the pulse signal,including de-noising and de-baseline.etc.;finally,the preprocessed signal is extracted in the time-frequency domain and nonlinear features,and the 2000-dimensional data is compressed to 28-dimensional,which effectively reduces the complexity of later signal analysis.The second part is the design of the liver cirrhosis disease classification algorithm.Build a classification model for the extracted feature vectors,and improve the deep forest model with small data volume dependence.In the multi-granularity scanning part of the model,data subsets are divided according to categories,and different forest models are trained according to the subsets and assigned different weights parameters to reduce the impact of unbalanced data samples;K-nearest neighbors and logistic regression algorithms are added to the cascade forest module,and the feature loss of each layer output is considered to improve the accuracy of the classification model.After experimental analysis,the designed classification model has achieved a good classification effect on the liver cirrhosis data set,with an accuracy rate of 98.57%.It proves that the pulse signal of liver cirrhosis patients is separable from the normal pulse signal from the side,and it can help doctors to diagnose the health status of patients.The third part is the design of three portable pulse collectors.In view of the problems such as long acquisition time,poor positioning of the pulse collection part and the limitations of the single-point pulse collection instrument during the data collection process,a portable three-part pulse acquisition instrument is designed in this thesis.The acquisition instrument is powered by lithium batteries and has strong portability.At the same time,it can realize the real-time acquisition,local real-time display and data storage of the pulse conditions of the cun,guan and chi.It can also transmit data with the host computer through the USB interface to realize in-depth analysis and processing of pulse conditions data.