MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641094736 A) filed by Dhanalakshmi Srinivasan College Of Engineering And Technology on August 05, 2026, for Ai-Assisted Medical Image Segmentation Using U-Net And Transformers.
Inventors include Dr. K. John Peter; B. Rakshitha; T. Sivamathu; and K. Thirumurugan.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present invention relates to an Artificial Intelligence (AD-assisted medical image analysis system for the automated classification and segmentation of kidney and brain tumors using advanced deep learning techniques. The invention integrates a Vision Transformer (ViT) model for accurate tumor classification and a U-Net model for precise pixel-level tumor segmentation, thereby providing an intelligent computer-aided diagnostic solution for healthcare applications. The system receives medical images, including Magnetic Resonance Imaging (MRI) scans for brain tumor analysis and Computed Tomography (CT) scans for kidney tumor analysis. The acquired images undergo preprocessing operations such as resizing, normalization, and image enhancement to improve image quality and prepare the data for deep learning analysis. The preprocessed images are first analyzed by the Vision Transformer model, which extracts global contextual features using a self-attention mechanism to classify the medical images into predefined disease categories. The classified images are subsequently processed by the U-Net segmentation model, which accurately identifies and delineates tumor regions by generating pixel-level segmentation masks. The invention further includes a Django-based web application that enables authorized users to upload medical images, perform automated analysis, and visualize both classification results and segmented tumor regions through a secure and user-friendly interface. The integrated framework combines image preprocessing, classification, segmentation, result visualization, and data management into a unified platform, thereby eliminating the need for multiple standalone diagnostic systems.
Disclaimer: Curated by HT Syndication.
