MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641007493 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Jan. 26, for 'a system and method for domain-adaptive multi-stage diagnosis of retinopathy of prematurity using attention-based analysis.'
Inventor(s) include A Anny Leema; P Balakrishnan; Haritha R; Prasanna Kumar K S; and Sowmiya K.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to an automated system and method for stage-wise diagnosis of Retinopathy of Prematurity (ROP using retinal fundus images. The invention provides a domain-adaptive, attention-enhanced diagnostic pipeline configured to operate across heterogeneous fundus imaging devices. The system integrates a preprocessing module employing PCAr-CLAHE with adaptive gamma correction for enhancing clinically relevant retinal features, a class-balanced learning strategy to address dataset imbalance, and an attention-enhanced deep learning model incorporating convolutional block attention modules for accurate multi-stage ROP classification. An explainability module generates visual explanation maps identifying retinal regions contributing to the diagnostic decision, thereby improving clinical interpretability and trust. The invention achieves improved generalization, diagnostic accuracy, and reliability compared to conventional approaches and is suitable for deployment in neonatal screening and tele-ophthalmology environments, particularly in resource-constrained clinical settings."
Disclaimer: Curated by HT Syndication.
