MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641096927 A) filed by R Nandhana; Renu Gopu; and Dr. S. Ramesh on August 11, 2026, for Multimodal Ensemble Framework For Automated Melanoma Screening.
Inventors include R Nandhana; Renu; and Dr. S. Ramesh.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: This invention relates to an automated melanoma screening system utilizing a multimodal ensemble deep learning framework for reliable and interpretable skin lesion classification from dermoscopic images. The system addresses critical challenges in medical image analysis including image noise, class imbalance, model overfitting, and lack of interpretability. The framework comprises three key technical components: (1) Intelligent Preprocessing: Automated hair removal via DullRazor algorithm, color normalization, lesion region-ofinterest (ROI) extraction, and data augmentation to enhance image quality and address class imbalance. (2) Ensemble Architecture: Parallel deployment of three CNN architectures (EfficientNet-BO, EfficientNet-B4, DenseNet121) with transfer learning, combined through soft-voting ensemble and feature-level fusion to leverage complementary strengths of each model. (3) Clinical Interpretability: Integration of Grad- CAM explainable AI techniques providing visual heatmaps highlighting decision-critical image regions, building clinician trust. Experimental evaluation on the ISIC dataset demonstrates superior performance: 89.6% classification accuracy, 0.90 precision, 0.89 recall, 0.89 F 1 -score, and 0.98 AUC, significantly outperforming individual baseline models. The system classifies dermoscopic images into seven skin lesion categories (Melanoma, Nevus, BCC, Benign Keratosis, Dermatofibroma, Actinic Keratosis, and Vascular lesion) with high reliability and clinical relevance. The invention enables early melanoma detection, reduces diagnostic workload on dermatologists, and supports clinical decision- making through transparent, explainable AT outputs—directly contributing to improved patient outcomes and reduced mortality rates aligned with Sustainable Development Goal 3 (Good Health and Well-being).
Disclaimer: Curated by HT Syndication.
