Abstracts

Basic Science to Clinical Research: Segmentation of Ultrasound and Modelling in Clinical Informatics

by Ali K Hamou




Institution: University of Western Ontario
Department:
Year: 2017
Keywords: Segmentation; Active Contours; Alzheimers Disease; Sepsis; Markov modelling; Clustering; Bioinformatics; Biomedical; Health Information Technology; Signal Processing
Posted: 02/01/2018
Record ID: 2154978
Full text PDF: https://ir.lib.uwo.ca/etd/4494


Abstract

The world of basic science is a world of minutia; it boils down to improving even a fraction of a percent over the baseline standard. It is a domain of peer reviewed fractions of seconds and the world of squeezing every last ounce of efficiency from a processor, a storage medium, or an algorithm. The field of health data is based on extracting knowledge from segments of data that may improve some clinical process or practice guideline to improve the time and quality of care. Clinical informatics and knowledge translation provide this information in order to reveal insights to the world of improving patient treatments, regimens, and overall outcomes.In my world of minutia, or basic science, the movement of blood served an integral role. The novel detection of sound reverberations map out the landscape for my research. I have applied my algorithms to the various anatomical structures of the heart and artery system. This serves as a basis for segmentation, active contouring, and shape priors. The algorithms presented, leverage novel applications in segmentation by using anatomical features of the heart for shape priors and the integration of optical flow models to improve tracking. The presented techniques show improvements over traditional methods in the estimation of left ventricular size and function, along with plaque estimation in the carotid artery.In my clinical world of data understanding, I have endeavoured to decipher trends in Alzheimers disease, Sepsis of hospital patients, and the burden of Melanoma using mathematical modelling methods. The use of decision trees, Markov models, and various clustering techniques provide insights into data sets that are otherwise hidden. Finally, I demonstrate how efficient data capture from providers can achieve rapid results and actionable information on patient medical records. This culminated in generating studies on the burden of illness and their associated costs.A selection of published works from my research in the world of basic sciences to clinical informatics has been included in this thesis to detail my transition. This is my journey from one contented realm to a turbulent one.