Multivariate signal processing and theoretical modeling for the study of gastrointestinal bioelectromagnetism

Abstract

The detection and characterization of gastrointestinal (GI) electrical activity has important clinical applications, including the early diagnosis of GI diseases in humans. In this dissertation, two important multivariate signal processing techniques (principal and independent component analyses, PCA and ICA) are employed to demonstrate the removal of both biological and non-biological artifacts from multi-channel magnetogastrographic recordings acquired using a superconducting quantum interference device (SQUID) magnetometer. The two methods are then employed to compute spatiotemporal parameters that can be used to differentiate between the healthy and diseased state of the GI tract. In addition, nonlinear dynamical analysis is employed to study the evolution of GI magnetically-recorded signals in the normal and abnormal states.