MUMBAI, India, June 22 -- Intellectual Property India has published a patent application (202621049427 A) filed by Moresh Madhukar Mukhedkar; Ruturaj Mane; Soham Chaware; Sakshi Admane; Pratiksha Khokale; and Anurag Kumar on April 17, 2026, for Computer-Aided Diagnosis System For Invasive Oral Cancer Detection Using Deep Learning Techniques.
Inventors include Moresh Madhukar Mukhedkar; Ruturaj Mane; Soham Chaware; Sakshi Admane; Pratiksha Khokale; and Anurag Kumar.
The application for the patent was published on June 12, 2026, under issue no. 24/2026.
Abstract: This project presents a Computer-Aided Diagnosis (CAD) system for automated detection of invasive oral cancer from oral cavity images using deep learning. The system leverages transfer learning with MobileNetV2, pre-trained on ImageNet, to classify images into Cancer and Normal categories. Multiple publicly available datasets from Kaggle are merged, cleaned, and preprocessed to form a unified dataset of 2183 images. Data augmentation, class weighting, and threshold tuning are applied during model training to improve sensitivity and reduce false negatives. The trained model achieves 95% test accuracy, 96% recall for the Cancer class, and an F1-score of 95%. The system is deployed as a web application built with Streamlit, where users upload oral cavity images and receive real-time predictions with confidence scores and risk level interpretation. This solution is lightweight, cost-effective, and suitable for academic demonstration and early oral cancer screening research.
Disclaimer: Curated by HT Syndication.
