Laparoscopy Image Enhancement

I was a part of the ViGIL research group at IIT Bombay, where I pursued my thesis under the guidance of Prof. Suyash Awate and Prof. Shabbir Merchant. We worked on removing random noise, surgical smoke, and speckles from the laparoscopy images. These corruptions perturb the texture and color of the organs, and make the surgical experience unpleasant, and reduce the efficacy of automated algorithms. We modelled the system using Markov random fields, and designed potential functions using color distribution priors and dictionary priors (to model texture). We used variational Bayesian factorization to make the optimization tractable using expectation-maximization (EM) algorithm.

Graphical Abstract

We had an oral presentation at IEEE ISBI 2017 (paper). The full thesis can be found here. I was awarded the Undergraduate Research Award 03 (URA-03) for my work.