Cellular-resolution endoscopic optical coherence tomography and image analytics

Coronary artery disease (CAD) and gastrointestinal (GI) cancers are among the top killers worldwide and in Singapore. Acute myocardial infarction (AMI, = heart attack), the clinical manifestation of CAD, is one of the leading causes of global morbidity and mortality. Due to the high prevalence and high risk of AMI, identification of vulnerable plaque associated with AMI is critical for early detection and subsequently interventional treatment. Esophageal, gastric and colorectal cancers are among the commonly diagnosed GI cancers worldwide and early diagnosis of these cancers is vital for the effective therapeutics. Intraepithelial neoplasia (IEN) may develop rapidly into invasive cancers without noticeable symptoms. Therefore, early detection of IEN when they are confined in epithelium is the key to achieving a good prognosis. However, we are limited in methods to obtain histology images of coronary arteries because it is hardly practical to take biopsy from coronary arteries. Even though biopsies can be routinely taken from GI tissues, sampling error may also lead to diagnostic errors. Therefore, a nondestructive imaging tool that can provide cellular-resolution images is critical for improving diagnostic accuracy. We have developed a desktop micro-optical coherence tomography (µOCT) imaging system and a novel endoscopic µOCT fiber-optic probe towards cellular-resolution imaging in coronary arteries and GI tracts. The desktop µOCT imaging system is an improved version of previously reported µOCT system in that it has a fiber-optic flexible handheld probe, which enables in vivo imaging of living animals. The first main technical contribution of this thesis is the development of the endoscopic µOCT fiber-optic probe. The µOCT fiber-optic probe achieves an axial resolution of 2.48 µm in air and transverse resolution of 4.8 µm, which are 4 times and 4.17 times better than the current endoscopic OCT devices, respectively. In particular, we used a novel beamsplitter design at the distal end of the fiber-optic probe so that we can realize the common-path design, annular focusing, and an all-glass optical path. The common-path design eliminates the dispersion difference between the sample and reference arms so that optimal axial resolution of 2.48 µm in air can be achieved, as well as eliminates the polarization mismatch between the two arms during the probe rotation for circumferential scanning. The annular focusing enables a 1.3 times extended depth-of-focus (DOF) which mitigates the problem of DOF limitation. The all-glass optical path makes it easier to fabricate the probe. Besides, we also employed a rigid outer sheath surrounding the probe so that areas of interest were properly maintained around the relatively small focal region to alleviate the issue of limited DOF. In order to investigate the capability of µOCT for visualizing cellular structures, we firstly imaged rat colon in vivo using the desktop µOCT imaging system. Imaging results show that the detailed microstructures, such as the crypt lumens and the goblet cells, could be clearly identified which was supported by corresponding histology images. Secondly, we conducted ex vivo imaging of fresh intact swine colon, swine coronary arteries, and human atherosclerotic coronary arteries by the endoscopic µOCT fiber probe. In normal swine colon we were able to visualize cellular-level microstructures such as goblet cells; we also clearly visualized smooth muscle cells and foam cells in atherosclerotic plaques. These results demonstrate that this endoscopic µOCT fiber-optic probe is capable of visualizing cellular-level morphological features of both GI tracts and coronary arteries. The second main technical contribution of this thesis is the design and simulation of the second fiber probe, which may improve the DOF by 2 times. The rationale for this effort is that the DOF of the above-mentioned first probe is still not enough for in vivo use. This second probe design follows the principle of multiple aperture synthesis (MAS) and digital refocusing. It uses a novel calcite-based polarization beamsplitter to create two apertures at the pupil plane of the objective lens and to form three apertures in the detection path. The phases of interferometric signals collected through these three sub-apertures can be digitally manipulated so that the three beams can be "refocused" at an out-of-focus point. I have completed the optical design and simulation of the optical performances, and fabricated the second fiber probe. In addition to the above-mentioned two technical contributions, I have conducted mechanical simulations to model the plaque stability in coronary atherosclerosis. As firstly pointed out, rupture of vulnerable plaque is critically associated with cardiovascular thrombosis and even AMI, whereas its detailed mechanisms are not fully understood. Recent studies have found abundant cholesterol crystals in ruptured plaques, and it has been proposed that the rapid expansion of cholesterol crystals in a limited space during crystallization may contribute to plaque rupture. However, the potential effect of cholesterol crystals on plaque rupture remains elusive due to the lack of the geometry of cholesterol crystals for analysis. In previous studies, µOCT can clearly visualize cholesterol crystals within arterial tissues ex vivo, and opens the possibility for evaluating the relationship between cholesterol crystallization and plaque rupture. Based on the measured geometric information of cholesterol crystals in human atherosclerotic aorta tissues, we developed a two-dimensional finite element method model of atherosclerotic plaques containing expanding cholesterol crystals and investigated the effect of the magnitude and distribution of crystallization on the peak circumferential stress.Doctor of Philosophy (EEE