MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641097964 A) filed by Sri Eshwar College Of Engineering on August 12, 2026, for Explainable Hybrid Deep Learning System For Early Melanoma Detection And Clinical Decision Support.
Inventors include Ms. K. Parimalam; and Dr. Suganthi Evangeline C.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: Melanoma is one of the most dangerous types of skin cancer and timely and accurate diagnosis can improve the patient's survival rate. Conventional AI-based diagnostic systems are mostly single-modality picture processors, and have low clinical acceptability due to the lack of interpretability techniques. The current invention proposes an Explainable Multi-Modal Deep Learning Framework for improved accuracy, reliability and interpretability in melanoma diagnosis using clinical skin scans, dermoscopic images and patient metadata. The proposed system utilizes a robust feature extraction phase followed by a hybrid ensemble of convolutional neural networks and an adaptive attention-based fusion module which dynamically fuses multimodal data according to their diagnostic relevance. The inventors envision an architecture with clear decision support provided to doctors through an explainable AI module that generates feature importance scores and visual interpretation maps. The framework includes an explainable AI module that offers feature importance scores and visual explanation maps, allowing healthcare professionals to make more informed and transparent diagnostic decisions. The system has a confidence estimation algorithm to recognize questionable predictions and calls for expert review when required. Also, an automatic clinical report generation module provides a structured summary of diagnostic findings, lesion categorization, confidence score, and visual explanations. The invention helps to build confidence amongst clinicians while bringing down false-positive and false-negative classification for effective use cases and improves diagnostic accuracy. It also facilitates realtime deployment of the skin cancer screening on cloud, portable and telemedicine platforms. The proposed framework provides a scalable, intelligent, and clinically effective method for identifying and treating melanoma at the earliest stages.
Disclaimer: Curated by HT Syndication.
