Multipixel Si PN Junction Photodetector Sensor Arrays for the Depth Resolved Measurement of Tissue using Diffuse Reflectance Spectroscopy

The optical characteristics of tissue can give insight into the health status of that tissue. Tissue health is indicated by tissue properties such as chromophore concentrations (including hemoglobin, iron, glucose and cytochrome oxidase) as well as by the type of cells (cancerous vs normal) present in the tissue. Diffuse reflectance spectroscopy (DRS) is a non-invasive optical interrogation technique for the quantification of tissue properties and the classification of tissue. DRS systems are generally composed of two parts, the sensing system and the data processing used for tissue/sample analysis. Systems and analysis techniques can both range widely depending on the application. Moreover, the classification accuracy of both can suffer in low signal-to-noise ratio (SNR) settings.

This thesis combines a high density multipixel Si photodetector (PD) probe, Monte Carlo (MC) simulations, and partial least squares regression (PLSR) analysis to enable the extraction of optical properties and the classification of samples in low SNR settings. The development and characterization of a DRS multipixel Si PD sensor array is presented. A multipixel Si PD sensor array is used herein to measure homogenous liquid tissue-mimicking phantoms. This data was then used to develop a PLSR algorithm to directly predict, for the first time, the optical coefficients of a tissue-mimicking phantom in a leave-one-out approach. The Si PD array was also used to conduct DRS measurements of 2-layer and 3-layer solid phantoms that were modeled using MC simulations to analyze the depth sensitivity of the multipixel Si PD array. This depth sensitivity analysis contributes to the DRS field by examining the difference in DRS signals between homogeneous and heterogenous samples using a multipixel Si PD probe, demonstrating, for the first time, that low contrast buried layers surrounding by homogenous media can be detected.