Techniques for quantification and interpretation of gastric slow wave activity

Abstract

Rhythmic bio-electrical events known as slow waves coordinate the muscular contractions of the stomach and are essential for physical propulsion and digestion of food. Dysrhythmic slow wave activity is associated with major gastrointestinal (GI) motility disorders, such as gastroparesis, and convey a substantial socioeconomic health burden. Unlike the eld of electrocardiology, where cardiac bioelectrical signals are routinely used for diagnostic purposes, the bio-electrical activity in the GI tract is not well understood and is not widely used in clinical practice. In recent years, the advent of extracellular multi-electrode high-resolution (HR) mapping in the GI eld has signi cantly improved experimental and clinical understanding of serosal gastric slow wave activity in normal and dysrhythmic states. One of the current limitations of HR mapping is the inordinate amount of data, making manual analysis a tedious and time-consuming task. More importantly, the use of manual analysis is prone to observer bias and error and could misconstrue experimental observations. Another traditional technique, that has been used since the discovery of bio-electrical events in the GI tract in the 1920s, is the use of extracellular body surface electrogastrograms (EGG). EGG signals are inherently noisy, leading to misleading quantitative estimates of the signal and are currently unable to reliably discriminate against di erent gastric motility disorders. The thesis begins by providing a broad overview of the bio-electrical activity present in the stomach and the various cooperating mechanisms that in uence gastric motility. Then, the focus is on developing a standardised set of ltering techniques and methods for detecting and visualising slow wave events using HR mapping in an e cient and accurate manner. It was found that inclusion of the dominant frequency of the slow wave (3{5 cpm) and major harmonics up to 2 Hz are vital for morphological and time based analysis. Reliable quantitative methods were developed for estimating the velocity and amplitude of the gastric slow wave propagation. The velocity method (FDSM) and amplitude method (derivative based) were compared to currently used methods (FD for velocity and `maximum-minimum' for amplitude) with realistic synthetic propagation patterns and signals. The newly developed FDSM method resulted in half the error of the FD method (Speed error: 12% vs 6%, angle error: 7 vs 3 ), while the new amplitude estimation method resulted in a third of the error of the current method (mean error: 44 V vs 16 V). These methods allowed for an improved understanding of the gastric slow wave propagation in normal and dysrhythmic states by prescribing a de ned quanti cation for various types of propagation. One of the major ndings was that dysrhythmic slow wave propagations were associated with rapid, high amplitude circumferential wavefronts. The biophysical basis of the slow wave activity was explored with an investigation into normal and abnormal gastric slow wave propagation. In normal slow wave propagation, the activation-recovery interval of the slow wave interval was higher compared to dysrhythmic slow wave propagation (mean: 4.3 s vs 3.3 s), while the activation-activation interval was shorter in normal propagation compared to dysrhythmic slow wave propagation (mean: 16.4 s vs 32.1 s). Thus, with a reduced slow wave interval during dysrhythmic slow wave propagation, it was hypothesised that the potential for spike activity and muscular contraction was reduced, potentially causing gastric motility disorders. With the framework of processing and quantifying HR mapping rmly established, an automated classi cation and identi cation tool was developed, capable of detecting and localising the pacemaker, colliding wavefronts and conduction block sites. The automated classi cation and classi cation was on average 96% accurate compared to manual classi cation and identi cation. While the manual method took an expert in the eld 20{40 minutes to classify and identify the slow wave propagation patterns, the automated methods performed the task almost instantaneously. The inclusion of automated analysis, which can be implemented in real-time, permits HR mapping to be employed as a routine clinical utility in the GI eld. Automated methods for analysing cutaneous EGG signals were developed that can discard sections of the signals that are corrupted by noise, thereby allowing reliable quantitative estimates of frequency and amplitude. This work was motivated due to the fact that in current clinical practice, the frequency and amplitude of the EGG traces are frequently manually estimated in an expedited manner, which introduces observer bias and error. The rst step in the automated approach was to develop ltering techniques to suppress noise, after which, running estimates of frequency and amplitude of the EGG signal were calculated. Then the frequency and amplitude characteristics of the EGG signal were assessed as to whether the EGG signal was corrupted by noise to discard quantitative estimates. The manual versus automated frequency (mean: 3.3 cpm vs 3.4 cpm) and amplitude (mean: 0.143 mV vs 0.144 mV) estimates were in concordance with each other. The automation of estimating the frequency and amplitude of the EGG could validate the usefulness of EGG in routine clinical practice through large scale clinical trials performed in normal patients, and those with gastric motility disorders. The work presented in this thesis presents a path forward in the GI eld to utilise the bio-electrical slow wave signals in a quantitative manner for patient care to prescribe a diagnosis, prognosis and direct treatment strategies accordingly.