Cellular-resolution optical coherence tomography : low-cost, compact hand-held probe development and image analysis

• Main Author: Xie, Jun
• Other Authors: Liu Linbo
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/138154
• Institution: Nanyang Technological University

The superficial layers of the skin and mucosa form the physical barrier between the internal and external environment, where many important diseases originate, such as intraepithelial neoplasms/cancers and skin/subcutaneous diseases. The former contributed more than 80% of the cancer cases, which is among the top 3 leading causes of deaths for many years. The latter contributed 1.79% to the global burden of diseases measured in disability-adjusted life years. As the numbers are climbing up in recent years, it is of great importance to develop technologies for early detection and auxiliary diagnosis of these diseases. Intraepithelial neoplasia and most of the skin/subcutaneous diseases are similar in that they are both in close relationship to the anomalies in epithelium. In these circumstances, handheld optical coherence tomography (OCT) was chosen as our basic imaging method for monitoring and examining conditions of epithelium in skin and oral mucosa in vivo, and digestive tracts ex vivo. Although OCT’s micrometer-level resolution and millimeter-level penetration makes it possible to image details at the cellular level, there are several problems need to be solved before its wide application. First of all, access to the mucosal or skin areas with space constraints requires a compact probe made of miniaturized optics, for which off-axis optical aberration issues should be addressed to have enough field of view. Secondly, the cost of the high-resolution probe optics should be reduced to facilitate wide application of the tool. Lastly but also importantly, the processing and analysis of OCT images is time consuming and requires well trained observers with prior knowledge of OCT imaging mechanisms, which hinders timely diagnosis. To mitigate the issues caused by off-axis aberrations, we proposed the elliptical input beam in the sample arm, instead of the conventional circular beam. The major axis of the elliptical beam is along the B-mode scanning direction (X) so that the aberrations is reduced as the aperture size reduces along the minor axis. We developed an analytical framework of the elliptical-beam based flying spot scanning optics to characterize and analyze the Strehl ratio and resolution of the focus affected by individual aberrations based on scalar diffraction theory and Zernike polynomials. Using off-the-shelf low-cost optics, we further designed and developed an elliptical beam based sample arm with a major/minor axis ratio of ~2.5 so that the transverse resolutions along X-axis and Y-axis is 1.8 µm and 4.5 µm, respectively. The Y-axis resolution is close to one cell size and equivalent to the sectioning thickness of routine histology. Therefore, this transverse resolution will not compromise the image quality with respect to histological analysis. We demonstrated a 70.47% larger transverse field of view numerically and 69% experimentally with human skin in vivo with respect to the circular-beam input, given the worst transverse resolution of 2.9 μm and Strehl ratio of 80%. We also proposed novel high-resolution imaging markers and developed automatic algorithms which can efficiently detect and segment microstructures including the stratum corneum and dermis papillae, as well as the basement membrane and nuclei of the nonkeratinized squamous epithelia. I proposed to use the intensity contrast of the stratum basale in the dermal papillae patterns for detection of the pigment loss in vitiligo and other pigmented dermatoses. To facilitate the axial localization of the dermal-epithermal junction (DEJ), we used the stratum corneum as the depth reference. I have demonstrated the feasibility of the algorithm in a small number of vitiligo patients. I also propose to use images of the nuclei and basement membrane of the nonkeratinized epithelium to detect metaplasia, which is a pre-cancerous condition of high diagnostic value. To do so, I developed algorithms to extract the nuclei and basement membrane in high-resolution OCT images. The results of the two extracted feature structures are further combined using a weighted ensemble classifier for metaplasia detection. The feasibility of these algorithms has been demonstrated in 512 human tissue samples ex vivo. The on-going work on elliptical input beam-based probe is focused on development of a miniaturized low-cost probe for imaging oral mucosa at the cellular resolution. I have designed and fabricated the first prototype using three-dimensional printing and will iteratively test and improve the imaging performances in future. The future work on the automatic detection algorithms will be focused on reduction of the computational demand for real-time diagnosis, and validation studies in larger patient populations against clinical gold standards.