MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641111266 A) filed by Vellore Institute Of Technology on September 14, 2026, for Automated Image-Driven Multi-Organ Defect Detection Using Deep Learning And Explainable Artificial Intelligence Framework.
Inventors include Sudha M; and Patil Soham Lalitkumar.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: ABSTRACT AUTOMATED IMAGE-DRIVEN MULTI-ORGAN DEFECT DETECTION USING DEEP LEARNING AND EXPLAINABLE ARTIFICIAL INTELLIGENCE FRAMEWORK A medical image of a subject is received. A routing model (120) comprising a first trained neural network classifies the medical image into one of a plurality of organ classes without operator selection of an organ type. Based on the organ class, the medical image is dispatched to one of a plurality of organ-specialized classifiers (130), each comprising a second trained neural network trained on a respective organ class, which outputs a disease classification and a severity grade. An explainability module (140) segments the medical image into superpixel regions, perturbs the superpixel regions, evaluates responses of the organ- specialized classifier to the perturbed superpixel regions, identifies a subset of the superpixel regions most influential on the disease classification, and generates an overlay image identifying the subset. A report generation module (150) generates a natural-language clinical report from the organ class, the disease classification, the severity grade, and the subset. (Fig. 1)
Disclaimer: Curated by HT Syndication.
