Avci, RecepCheng, LeoPaskaranandavadivel, NiraDrake, Chad Ephraim2024-02-132024-02-132023https://hdl.handle.net/2292/67463Bioelectric slow waves are key to the coordination of gastric motility. Dysrhythmic slow wave activity has been linked with functional motility disorders, and the non-invasive identification of dysrhythmias is therefore an attraction option for assessing gastric pathology. Magnetogastrography (MGG) is the non-invasive measurement of the biomagnetic fields generated by slow waves. The identification of spatially complex dysrhythmias using MGG is challenging because current analysis methods do not reference subject-specific anatomy and cannot separate multiple events when they are active simultaneously. Anatomically-based gastric source imaging enables non-invasive slow wave measurements to be mapped to the 3D stomach, but the development of these methods has been limited by a lack of groundtruth data. Therefore, the objectives of this thesis were to develop a method for invasively mapping 3D slow wave propagation over the stomach, and to use this method to develop and validate MGG-based gastric source imaging. The first anatomically-based MGG source imaging method was developed. The method involved the distribution of electric dipoles with fixed positions and orientations over the surface of subject-specific stomach models, and the estimation of dipole weights required to best reproduce MGG signals. Source imaging was first evaluated using MGG simulated for the bradygastric ectopic activity (1.9 cpm) and bradygastric anterograde activity (1.2 cpm) recorded from the serosa of the anaesthetised pigs. Additionally, a normogastric anterograde propagation pattern (3.2 cpm) was generated from the bradygastric case by reducing the interval between events. Source imaging correctly identified all activity patterns, and simultaneously active slow wave events were correctly localised, demonstrating that the method can isolate multiple events. The impact of uncorrelated measurement noise was evaluated for signal-to-noise ratios in the range of 20 dB to -10 dB, and the method was substantially robust up to a noise level of 5 dB. At 5 dB, all evaluated propagation patterns remained identifiable, mean position error was at most 6.4±0.1% of the longitudinal organ axis (13±0.2 mm), and the lowest mean correlation with the ground-truth activity was 0.42±<0.01. Additionally, the method was substantially robust against a realistic level of geometric error (AHD of 5 mm ±20%), as all propagation patterns remained identifiable, mean position error was at most 6.6±0.9% of the longitudinal organ axis (13.4±1.9 mm), and the lowest mean correlation with the ground-truth activity was 0.38±0.07. The source imaging method was then validated using the experimental MGG recordings. Anterograde propagation (1.2 cpm) and the retrograde component of the ectopic activity (1.9 cpm) were correctly identified, although the estimates were offset from the ground-truth activity by 11.9±7.0% and 16.0±8.8% of the longitudinal organ axes (25.7±15.1 mm and 32.6±17.9 mm), respectively. The characterisation of the ectopic activity pattern, in particular, is significant because the direction of propagation was not identifiable from traditional MGG propagation maps. Source imaging accuracy was lower for experimental MGG compared to simulated MGG, and this was likely due to challenges relating to experimental methods, MGG signal processing, and forward modelling. However, the ground-truth activity provided by gastric electroanatomical mapping should aid the development of forward models and signal processing techniques. In this thesis, the use of MGG source imaging to estimate the distribution of the slow wave activity over the 3D stomach was validated against ground-truth serosal measurements. Additionally, the simulation studies provided evidence that the method is robust against realistic levels of measurement noise and geometric error. These results demonstrate that MGG source imaging is a compelling option for identifying dysrhythmias, which is a capability that would aid the diagnosis of functional motility disorders and guide targeted treatments (e.g., ablation and stimulation).Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated.https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htmhttps://creativecommons.org/licenses/by-nc-sa/3.0/nz/Thesis2024-02-06Copyright: The authorhttp://purl.org/eprint/accessRights/OpenAccess