AI-Driven Classification for Specialized Oral Informatics

Undergraduate Researcher
Guide: Prof. Nirmal Punjabi, KCHD | Indian Institute of Technology Bombay

Oral Health Image Classification Pipeline

Project Overview

During my tenure as an Undergraduate Researcher at IIT Bombay, I focused on addressing automated diagnostic challenges within medical imaging. The primary objective was to develop a robust, deep learning-based diagnostic tool capable of accurately segmenting and classifying specialized oral diseases from visual data, ultimately aiding in faster and more reliable medical informatics.


Technical Implementation

To handle the complexities and strict accuracy requirements of medical image classification, I architected an end-to-end deep learning pipeline focused on robust feature extraction and balanced data handling:


System Output & Impact

The resulting pipeline provided a solution for oral disease classification. By combining the deep feature extraction of ResNet101 with rigorously balanced data handling and augmentation, the model achieved reliable multi-class classification, demonstrating the viability of advanced deep learning architectures in specialized medical diagnostic environments.