MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641063707 A) filed by Dayananda Sagar University Devarakaggalahalli, Harohalli, Kanakapura Road, Bengaluru South Dt, Karnataka on May 20, 2026, for Hybrid Quantum Classical System For Brain Tumor Diagnosis.
Inventors include Sushma Ds Assistant Professor, Cse, Dayananda Sagar University, Harohalli, Kanakapura Road, Bengaluru South Dt, Karnataka; Manish Nandy Assistant Professor, Department Computer Science And Engineering Dayananda Sagar University, Harohalli; Aishwarya K, Assistant Professor, Department Computer Science And Engineering Dayananda Sagar University, Harohalli, Kanakapura Road, Bengaluru South Dt, Karnataka; Suhita Biswas,assistant Professor, Department Computer Science And Engineering Dayananda Sagar University, Harohalli; Sonali Bairagi; Assistant Professor, Department Computer; Joydeep Patar, Assistant Professor, Department Computer; Bhanuteja M Engcs; Computer Science Department; Abhishta Mp Engcs Computer Science Department Dayananda Sagar University, Bengaluru Karnataka; B Krishna Charan Engcs. Computer Science Department Dayananda Sagar University, Bengaluru Karnataka; and Tejas H R Engcs, Computer Science Department Dayananda Sagar University, Bengaluru Karnataka.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: Q-Neuro is a reliability-driven, hybrid diagnostic framework designed for four-class brain MRJ classification (glioma, meningioma, pituitary tumor, and no tumor), with a strong emphasis on robustness, interpretability, and clinical usability. The pipeline begins with heuristic out-of~ distribution (OOD) screening to identify and filter anomalous or low-confidence inputs, ensuring that only clinically relevant data proceeds through the system. Subsequently, automated region-of-interest (ROI) extraction isolates salient brain regions, reducing noise and improving downstream model focus. A Res Net IS-based deep learning backbone perfonns primary classification, capturing hierarchical imaging features. This prediction is further enhanced through a four-qubit variational quantum circuit (VQC) refinement module, which models complex feature interactions and contributes to improved decision sensitivity in challenging cases. For spatial precision, the system integrates U-Net-based segmentation to accurately delineate h1mor regions, while Grad-CAM visualization generates interpretable heatmaps that highlight diagnostically significant areas, thereby supporting clinician trust and transparency. A conservative decision fusion strategy aggregates outputs from classification, quantum refinement, and segmentation stages to minimize false positives and false negatives, prioritizing diagnostic reliability. The entire framework is deployed as a web- based prototype, enabling scalable, accessible, and efficient decision support for brain tumor diagnosis.
Disclaimer: Curated by HT Syndication.
