Abstract: Enhanced visualization is critical for improving early disease detection in gastrointestinal (GI) endoscopy, as conventional white-light imaging often fails to adequately highlight subtle tissue and vascular changes. This work presents a progressive investigation into spectral approaches for improving endoscopic visualization, spanning prototype development, in vivo validation, and practical clinical translation. Initially, an in-house multispectral illumination system was designed and developed to study wavelength-dependent light–tissue interactions, demonstrating the ability of different wavelengths to selectively enhance superficial and deep vascular structures. The feasibility of this approach was further validated through animal studies using a porcine model, confirming the potential of spectral imaging for improved visualization in realistic endoscopic conditions. However, practical constraints associated with deploying narrow-band illumination hardware in routine clinical settings motivated the development of a software-based alternative. Consequently, an approximate spectral estimation method was proposed to extract spectral information directly from standard RGB white-light endoscopic images, enabling the generation of pseudo-colored enhanced images without specialized illumination hardware. This approach was integrated into an indigenously developed endoscopy system and clinically validated under ethical approval. Collectively, this work demonstrates a complete translational pipeline—from experimental multispectral imaging to clinically feasible image enhancement—aimed at enabling practical, scalable improvement in endoscopic visualization.

Event Details
Title: Translating Spectral Imaging Techniques into Practical Endoscopic Visualization.
Date: January 22, 2026 at 3:00 PM
Venue: Online (https://meet.google.com/roe-romh-scz)
Speaker: Mr. AMALAN S (EE20D408)
Guide: Dr. Mohanasankar S
Type: PHD seminar

Updated: