Nieliniowe własności sygnałów elektromiografii powierzchniowej u pacjentów leczonych z powodu raka jelita grubego

Abstract

The subject of this dissertation is the analysis of surface electromyographic signals that characterize the activity of the external rectal sphincter during colorectal cancer treatment. The studied muscle activity can serve as a very valuable source of knowledge about possible neuromuscular dysfunction after surgical removal of the tumor. The methodology of data analysis contained a set of nonlinear methods, including standard multifractal analysis and the techniques based on the study of information carried by the examined time series. The selected techniques have been implemented taking into account a number of factors, i.e. the type of surgical treatment, the time elapsed since the surgery, the state of muscle tension and the signal registration point (the location of the electrodes at different depths of the rectal canal). Based on the Multifractal Detrended Fluctuation Analysis method, the multifractal nature of the signals was identified in the range of short time scales. The focal point of this research is the estimation of the multifractal spectrum. Two main spectrum parameters were analyzed - spectral width and maximum. The results showed statistically significant differences for selected treatment stages and for specific measuring depths. In addition, using the Anova Friedman statistics at the selected significance level, the presence of statistically significant differences between the parameters of the spectra calculated for the series measured before the operation and three times in the postoperative convalescence phase was demonstrated. The second part of the analysis considered the measures derived from information theory. Characterization of average values of information contained in the signal allowed for differentiation of individual stages of muscle tension and indicated a greater complexity of the myographic signal in the state of muscle relaxation. The classical characteristics of the basic measures of information theory in the form of Shannon entropy and spectral entropy did not show significant changes between individual treatment stages. Therefore, it was decided to apply more advanced techniques for extracting information from the signal. The Sample Entropy Analysis and its extension for multiple time scales led to the identify the specific treatment stages and the depths in the rectal canal for which the values of information contained in the signal are most diverse. In addition, it has been shown that in these specific conditions, the stage directly after surgery is distinguished by the highest degree of differentiation with respect to the type of operation, which has also a clinical justification.