Scalable Quantum Non-Local Neural Network Optimized with Tyrannosaurus Algorithm for Brain tumor Detection using MRI Images.
Authors
Affiliations (4)
Affiliations (4)
- Assistant Professor, Department of Computer Science and Engineering, SRM Valliammai Engineering College, SRM Nagar, Kattankulathur, Tamil Nadu 603203, India.
- School of Computer science and engineering, Galgotias University, Greater Noida, India.
- Department of Physics, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Science (SIMATS), Thandalam, Chennai602105, Tamil Nadu, India.
- Project Scientist, National Centre for Assistive Health Technologies (NCAHT), IIT Delhi (Senior Project Scientist) 110016, India.
Abstract
Early and reliable detection of brain tumors from magnetic resonance imaging (MRI) is essential for timely clinical diagnosis and treatment planning. However, automated tumor detection continues to be challenging due to the heterogeneous nature of tumors, the complexity of brain structures, the presence of noise in MRI images, and the limitations of existing deep learning models in effectively capturing both global contextual information and fine-grained tumor features. Many existing approaches also suffer from high computational complexity and limited generalization across datasets. To address these challenges, this study proposes a novel automated framework for brain tumor detection that integrates advanced preprocessing, precise tumor region segmentation, hierarchical feature representation, and optimized classification. The framework is evaluated using two publicly available benchmark datasets, BRATS 2018 and Figshare, containing multi-class brain tumor MRI images with diverse tumor types and grades. Experimental results demonstrate that the proposed approach achieves high classification performance with an accuracy of up to 99.8%, outperforming several state-of-the-art deep learning models. Comparative analysis with existing methods confirms improved precision, recall, and F1-score, indicating robust detection capability across different tumor categories. The proposed framework also shows reduced computational error and improved feature representation for complex tumor structures. These results suggest that the proposed method can serve as a reliable computer-aided diagnostic tool to assist clinicians in early brain tumor detection and decision-making, thereby supporting more efficient and accurate clinical assessment.