MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641050977 A) filed by Amrita Vishwa Vidyapeetham, Bengaluru, Karnataka, on April 21, for 'skin disease categorization system and method thereof.'
Inventor(s) include Ravichandran, Harini Shree; Kaliyaperumal, Deepa; and Muthuraman, Supriya.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The present disclosure relates to a system (100) and method (400) for automated skin disease categorization using multimodal analysis of dermatological images and textual symptom data. The system (100) acquires a dermatological image and symptom statements, preprocesses the image through artifact reduction and tone normalization, and determines a skin tone category using fuzzy-based typology analysis. Tone-dependent enhancement and lesion delineation are performed using entropy-constrained deep segmentation to extract visual features. In parallel, semantic representations are derived from textual symptoms using a sentence embedding model (S2LTP-BERT) and consensus disease definitions are generated via clustering (GC2MAKMSC). Multimodal fusion is carried out using residual Wasserstein optimisation and eigen-entropy balancing to produce a diagnostic output comprising a predicted skin disease type, confidence score, and explainability cues. The proposed system (100) improves accuracy, robustness, and interpretability in skin disease diagnosis across diverse skin tones and sparse clinical data."
Disclaimer: Curated by HT Syndication.
