Noise Reduction And Analysis Of Non-stationary Gastrointestinal Physiological Signals

Slow transit constipation(STC)is a common disease,its main symptoms are difficult or incomplete excretion of stool,often accompanied by abdominal pain,abdominal distension,nausea and other uncomfortable symptoms.There are problems of nonlinearity and nonstationarity of physiological signals in traditional gastrointestinal diagnosis methods,and different gastrointestinal diseases have different physiological characteristics and abnormal changes,which leads to low recognition rate.In this paper,a variety of signal noise reduction methods are compared,and the gastrointestinal physiological signal feature extraction method is proposed,more effective classification and recognition of diseases,improve the level of diagnosis of STC.This paper aims to use the colonic pressure data under normal physiological conditions to eliminate the noise caused by respiratory movement,gastrointestinal peristalsis and muscle movement in gastrointestinal signals,extract the effective features of colonic motility,propose an effective classification model of slow transit constipation,and analyze the difference of colonic motility parameters between healthy control group and STC patients.The main research contents of this paper are as follows:1.In this paper,microelectronic capsule equipment was used to collect human gastrointestinal pressure signals,and noise reduction was carried out in combination with gastrointestinal system dynamics knowledge.Four methods including EEMD,Gauss,SSA and wavelet transform were compared,and the noise reduction effect of six wavelet bases on gastrointestinal pressure signals was comprehensively compared.The signal-to-noise ratio was increased to 14.5763 d B,and the root-mean-square error was reduced to 0.3767,which proved the effectiveness of the noise reduction method in disease diagnosis.2.In this study,pressure curve and informatics analysis were used to extract gastrointestinal dynamic characteristics,including mean value,contraction relaxation coefficient,standard deviation,area under the curve,energy and complexity,and it was found that multiple parameters of colonic movement were increased in patients with STC compared with healthy control group,and average energy was higher(HC: 228.21,STC: 859.46),but the mean complexity was low(HC: 0.85,STC: 0.67).These results indicate that this method can effectively obtain the dynamic characteristics of gastrointestinal tract and has important application prospects in the diagnosis of STC.3.We discussed the feasibility of KNN algorithm for classifying gastrointestinal diseases,and compared it with Naive Bayesian algorithm and SVM algorithm in classification accuracy.The results show that KNN algorithm performs better in classification accuracy,accuracy,specificity and sensitivity are 86.00%,86.20% and 86.00%,respectively.In this paper,a method for noise reduction and analysis of non-stationary gastrointestinal signals is proposed.Wavelet transform noise reduction and pressure curve feature extraction are adopted to classify diseases combined with KNN algorithm,which improves the diagnostic accuracy of slow transit constipation and provides theoretical and practical basis for the diagnosis and treatment of STC.