MUMBAI, India, Oct. 12 -- Intellectual Property India has published a patent application (202641118746 A) filed by Srm Trp Engineering College on October 05, 2026, for Development Of An Acquisition-Aware Deep Learning System For Reliable Intracranial Aneurysm Detection From Multimodal Brain Imaging.
Inventors include Vanmughilan S. K; Santhosh Pandi C; Yashwanth V; and Priya S.
The application for the patent was published on October 09, 2026, under issue no. 41/2026.
Abstract: The present invention discloses a system that leverages deep learning to detect cerebral intracranial aneurysms based on multimodal brain imaging data and simultaneously accounting for the imaging condition of the input scan. The scanning images, as DICOM files, are inputted to the system and then transformed to standardised 3D image volumes via a preprocessing method. From such image volumes, distinctive image descriptors are extracted through a deep learning 3D model while the acquisition related features, e.g. imaging modality, imaging orientation, voxel spacing, slice spacing, resolution and image-quality characteristics of the imaging data are also derived from the images. The generated feature set is a hybrid of the image features and the acquisition-related features that is fed to an estimator for the prediction of the existence of an aneurysm. The system then performs a reliability assessment to give a reliability indicator of the prediction and a possibility to flag which predictions still need manual verification. The system may also be validated with controlled variations in noise blur resolution and slice spacing to investigate the impact of the variation of images characteristics on the performance of the aneurysm detection task and prediction reliability.
Disclaimer: Curated by HT Syndication.
