Positive Time-Frequency Distribution Analysis of the Human Colonic Electrical Activity

Abstract

The electrical activity recorded from the human colon could play an important role in analyzing the pattern of contractions under different physiologic or experimental states. In general, the frequency of the electrical activity is extremely irregular and time-varying. Its analysis requires a technique that considers variations in both time and frequency domains. The research undertaken was to analyze time-frequency variations of the human colonic electrical activity, to implement positive time- frequency distribution techniques in a computer system and to analyze theoretical signals using this technique to characterize a kernel function. Our results show that the uncertainty coefficient together with the marginal conditions and the average of the conditional PTFD in time and also in frequency can be applied to determine which kernel function and c-value were appropriate for calculating the PTFD of a sinusoidal signal. The selected kernel function constituted a comparative template of signals with similar characteristics. We found that those results were useful in analyzing the time- frequency variations of the electrical activity recorded in the human colon. The comparison of relative contributions of frequency bands showed that the band with the highest values during the pre- and interprandial period was 30-40 cpm suggesting an important role in the generation of bursts of these signals. The numerical results suggested that a meal can induce changes in the relative importance of frequencies below 10 cpm and a significant change in the 30-40 cpm band. In addition, a computer program of the Chakravarti method was implemented to calculate the Fourier transform of nonperiodical signals. This program was part of the computer program system developed to compute the PTFD of theoretical and experimental signals.