From Investigation to Endoscope: Optical Biomarkers for Cancer Diagnostics

Abstract

In this dissertation, we conducted a multimodal imaging study on fresh human esophageal tissue samples obtained via endoscopic biopsies from 26 patients. Four distinct imaging modalities were performed: autofluorescence imaging, hyperspectral imaging, optical coherence tomography, and Mueller matrix polarimetry. The primary objectives were to address three key research questions: 1) Which individual modality best differentiates between healthy and cancerous tissue? 2) How can data from these modalities be integrated to maximize discrimination? and 3) What computational methods are suitable for analyzing the resulting high-dimensional multi-modal datasets? Our findings indicated polarized light imaging provided the highest discriminatory capability among the imaging modalities, yet when looking at flexible endoscopic translation we found a technology gap existed in polarimetry and flexible endoscopy. The next section of the dissertation focused on proposing and demonstrating a simple, compact, and low-cost architecture to enable snapshot spatial polarization measurements through a flexible imaging fiber. Specifically, we introduced a snapshot spatial polarization imaging approach using a pixelated polarizer integrated at the distal tip of a coherent imaging fiber bundle. The proposed optical architecture, combined with a robust computational workflow, represents a significant advancement towards addressing current technological challenges in flexible endoscopic polarimetry.