| MS Seminar

Name of the Speaker: Ms. Sneha Chand (EE19S090)
Guide: Prof. Mohanasankar Sivaprakasam
Online meeting link: https://meet.google.com/hid-hwpg-wgd
Date/Time: 27th December 2023 (Wednesday), 2:30 PM
Title: In-Vivo Non-contact Multispectral Imaging of Oral Tissues

Abstract

In oral imaging spectroscopy, the collection of tissue spectral data from resected samples may not accurately represent tissue signatures due to time-dependent changes, blood loss, protein degeneration, and preservation chemicals. In-vivo spectral imaging addresses these limitations but poses challenges such as device dimensions, tissue accessibility, and motion artifacts, impacting data quality and reliability. We present a novel spectral system that we developed, collecting a dataset of spectral images focused on oral diseases, addressing these challenges. Employing a state-of-the-art multispectral camera, we captured images at a resolution of 270x510 pixels in 16 spectral bands (460 nm to 600 nm). The dataset comprises 91 participants (15 healthy and 76 diseased), with multiple images per patient, totaling 243 spectral images. Various oral health conditions, including Oral Submucous Fibrosis (OSMF), Leukoplakia, and Oral Squamous Cell Carcinoma (OSCC), are represented, accompanied by detailed patient history records. We publicly release this oral health multispectral dataset with the potential to advance spectroscopy diagnosis. Our examination of over 200 spectral images investigates the potential of Multispectral Imaging (MSI) in distinguishing healthy and malignant areas within the oral cavity. Our study reveals several key findings: we successfully extract unique spectral signatures associated with different tissue types. Spectral Angle Mapper (SAM) yields promising results, with a correct identification rate of 86.87% for homogeneous OSCC lesions and 73.10% for non-homogeneous lesions. Additionally, we observe a noticeable increase of 5-6% in tissue oxygen saturation in malignant areas of OSCC compared to healthy tissue. Overall, these findings underscore the importance of a comprehensive spectral dataset and highlight the efficacy of various methods in investigating reflectance in oral tissues. These outcomes pave the way for further advancements in spectral imaging and its potential applications in healthcare.