Algorithms for Fluorescence Lifetime Microscopy and Optical Coherence Tomography Data Analysis: Applications for Diagnosis of Atherosclerosis and Oral Cancer

With significant progress made in the design and instrumentation of optical imaging systems, it is now possible to perform high-resolution tissue imaging in near real-time. The prohibitively large amount of data obtained from such high-speed imaging systems precludes the possibility of manual data analysis by an expert. The paucity of algorithms for automated data analysis has been a major roadblock in both evaluating and harnessing the full potential of optical imaging modalities for diagnostic applications. This consideration forms the central theme of the research presented in this dissertation.

Specifically, we investigate the potential of automated analysis of data acquired from a multimodal imaging system that combines fluorescence lifetime imaging (FLIM) with optical coherence tomography (OCT), for the diagnosis of atherosclerosis and oral cancer. FLIM is a fluorescence imaging technique that is capable of providing information about auto fluorescent tissue biomolecules. OCT on the other hand, is a structural imaging modality that exploits the intrinsic reflectivity of tissue samples to provide high resolution 3-D tomographic images. Since FLIM and OCT provide complimentary information about tissue biochemistry and structure, respectively, we hypothesize that the combined information from the multimodal system would increase the sensitivity and specificity for the diagnosis of atherosclerosis and oral cancer.

The research presented in this dissertation can be divided into two main parts. The first part concerns the development and applications of algorithms for providing quantitative description of FLIM and OCT images. The quantitative FLIM and OCT features obtained in the first part of the research, are subsequently used to perform automated tissue diagnosis based on statistical classification models. The results of the research presented in this dissertation show the feasibility of using automated algorithms for FLIM and OCT data analysis for performing tissue diagnosis.