MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641076466 A) filed by Sr University on June 20, 2026, for An Automated 3d Tumor Analysis In Medical Imaging Using Transformer- Based Deep Learning Models.
Inventors include Tuniki Krishnaveni; and Sreedhar Kollem.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: AN AUTOMATED 3D TUMOR ANALYSIS IN MEDICAL IMAGING USING TRANSFORMER-BASED DEEP LEARNING MODELS The invention relates to an automated 3D tumor analysis system for medical imaging using transformer based deep learning models. The system preprocesses volumetric data from MRI, CT, and PET scans, applies a transformer backbone with patch embeddings and self attention mechanisms to capture global spatial dependencies, and generates voxel level segmentation maps of tumor regions. A hybrid encoder decoder architecture integrates multi scale features, while training employs Dice and cross entropy losses with data augmentation and transfer learning. Outputs include 3D segmentation masks, confidence maps, and attention heatmaps, providing accurate, efficient, and explainable tumor detection. The system reduces computational overhead compared to 3D CNNs and is compatible with clinical imaging formats, enabling real time deployment in hospital PACS systems.
Disclaimer: Curated by HT Syndication.
