MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641051590 A) filed by Bgs College Of Engineering And Technology, Bengaluru, Karnataka, on April 22, for 'a method and system for generating a feature map for retinal images.'
Inventor(s) include Ravikumar Guralamata Krishnegowda; Sampath Subbe Gowda; Chethan Hirikythanahalli Venkatesh; and Manjula Gangadhariah.
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: "Retinal diseases such as diabetic retinopathy and glaucoma are major causes of vision loss, wherein conventional diagnostic approaches are limited by noise sensitivity, class imbalance, and inadequate feature representation, leading to reduced accuracy and computational inefficiency. The present disclosure provides a method and system for generating feature maps from retinal images using an integrated processing pipeline. The system performs adaptive preprocessing including noise reduction, data augmentation, and class balancing, followed by hierarchical multi-level feature extraction using a multi-scale convolutional architecture. Region-based localization using a region proposal network and ROI alignment enables precise identification of pathological regions, while a spatial attention mechanism enhances diagnostically relevant features. The generated feature maps are utilized for classification and regression of retinal conditions. The disclosed approach improves accuracy, reduces error rates, and enhances computational efficiency, thereby enabling reliable and scalable retinal disease detection."
Disclaimer: Curated by HT Syndication.
