MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641091695 A) filed by Suberiya Begum. S; and Dr. N. Sabiyath Fatima on July 28, 2026, for A Hardware-Integrated Multimodal Ocular Surface Assessment System Using Hierarchical Deep Learning.
Inventors include Suberiya Begum. S; and Dr. N. Sabiyath Fatima.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: The present invention is a hardware-integrated multimodal ocular surface assessment system (100) comprising an imaging enclosure (102), an RGB image sensor (104), an infrared meibography sensor (106), controlled visible and infrared illumination assemblies (108, 110), a polarization filter assembly (112), and an edge processing unit (156). An image quality assessment module (122) rejects blurred, underexposed, or glare-dominated frames. A preprocessing module (124) normalizes accepted images, and a region-of-interest localization module (126) isolates ocular tissue. A multi-attention segmentation engine (130) generates scleral and meibomian-gland masks using spatial and channel attention. A hierarchical feature extraction engine (140) and a Vision Transformer classification engine (142) produce an ocular-surface condition category and severity level. A confidence calibration module (148) cooperates with an output suppression controller (150) to withhold unreliable results. A clinical decision support interface (152) displays masks, confidence, severity, and explainability information for clinician review without autonomously prescribing treatment or issuing a final patient diagnosis. (Accompanied Figure No. 1 to 4)
Disclaimer: Curated by HT Syndication.
