MUMBAI, India, April 17 -- Intellectual Property India has published a patent application (202531104580 A) filed by Asansol Engineering College; Jis University; and Abacus Institute Of Engineering & Management, Asansol, West Bengal, on Oct. 30, 2025, for 'quality-aware contrastive learning for retinal image assessment using pyramid networks.'
Inventor(s) include Sumanta Karmakar; Dr. Jyotirmoy Chatterjee; Dr. Sambit S Mondal; Dr. Ashmi Chakraborty; Dr. Khushi Banerjee; Dr. Goutam Kumar Das; Anirban Kundu; and Kedarnath Paul.
The application for the patent was published on April 17, under issue no. 16/2026.
According to the abstract released by the Intellectual Property India: "Accurate retinal image quality assessment (RIQA) is critical for reliable diagnosis and screening of ocular diseases such as diabetic retinopathy and glaucoma. Variations in image quality due to blur, poor illumination, or defocus can compromise diagnostic accuracy and clinical workflow. This invention proposes a deep learning framework that combines a pyramid network architecture with a quality-aware contrastive loss function for automated RIQA. The pyramid network captures multi-scale features, including both local and global image details, while the contrastive loss enforces clear separation between high-quality and low-quality images in the feature space. Experimental evaluations demonstrate that the proposed approach robustly identifies diagnostically usable retinal images, even under challenging imaging conditions. By filtering out low-quality images, the system enhances screening efficiency and supports timely clinical decision-making, providing a reliable tool for ophthalmologists and telemedicine platforms to improve diagnostic accuracy and patient outcomes."
Disclaimer: Curated by HT Syndication.
