MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641074983 A) filed by Sr University on June 17, 2026, for A Computer-Based System For Automated Detection And Classification Of Brain Tumors Using Mri Scans.
Inventors include Maddileti Reddy V; and Dr. Pramod Kumar Poladi.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: A COMPUTER-BASED SYSTEM FOR AUTOMATED DETECTION AND CLASSIFICATION OF BRAIN TUMORS USING MRI SCANS The invention discloses a multi-scale attention-guided ensemble deep learning framework for automated detection and classification of brain tumors using MRI scans. The system integrates preprocessing techniques including histogram equalization, Gaussian filtering, and skull stripping to enhance image quality. Tumor segmentation is achieved through a ResAtU-Net model augmented with residual and attention modules for improved boundary localization. A hybrid feature extraction pipeline combines radiomics descriptors, wavelet-based textural features, and CNN-based deep features to provide rich tumor representations. Feature optimization is performed using a novel Hybrid Pangulture Optimization Algorithm (POA), which balances exploration and exploitation to select the most relevant features. Classification is carried out by a Multi-Scale Attention-Guided Ensemble Network (MSAE-Net) that adaptively fuses outputs from Vision Transformer (ViT) and MobileNet. Benchmark testing on BraTS datasets demonstrates superior performance, achieving accuracy of 98.7%, sensitivity of 98.3%, specificity of 99.1%, and real-time inference averaging 1.8 seconds per scan. The invention provides a clinically practical, interpretable, and highly accurate solution for brain tumor detection and classification, surpassing existing deep learning approaches.
Disclaimer: Curated by HT Syndication.
