Department of Electrical and Computer Engineering, University of Science and Technology of Mazandaran, Behshahr, Iran
10.22091/jdaid.2026.16857.1103
Abstract
Accurate diagnosis of brain tumors from Magnetic Resonance Imaging (MRI) plays a critical role in treatment planning, prognosis assessment, and clinical decision-making. However, reliable tumor classification remains challenging due to significant tumor heterogeneity, variations in imaging characteristics, and the subjective nature of manual interpretation by radiologists. To address these challenges, this paper presents a novel hybrid multi-stream deep learning architecture that combines the complementary strengths of Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and Graph Convolutional Networks (GCNs) within a Feature Pyramid Network (FPN) backbone. The proposed framework is designed to simultaneously capture local texture patterns, long-range global dependencies, and topological relationships among extracted features from MRI scans. In addition, a compound loss function is employed to alleviate the impact of class imbalance and improve classification robustness. To enhance model transparency and clinical trustworthiness, Grad-CAM is utilized to generate interpretable attention maps highlighting diagnostically relevant regions. Experimental evaluation on a publicly available CE-MRI dataset demonstrates that the proposed approach consistently outperforms several state-of-the-art methods across multiple performance metrics, including accuracy, precision, recall, and F1-score. The results confirm that the proposed hybrid architecture provides an accurate, robust, and interpretable solution for automated brain tumor classification.
Mirpour Marzuni, S. (2026). Interpretable Brain Tumor Classification via Triple-Stream Fusion of Local, Global, and Topological Features. Journal of Data Analytics and Intelligent Decision-making, (), -. doi: 10.22091/jdaid.2026.16857.1103
MLA
Saeed Mirpour Marzuni. "Interpretable Brain Tumor Classification via Triple-Stream Fusion of Local, Global, and Topological Features". Journal of Data Analytics and Intelligent Decision-making, , , 2026, -. doi: 10.22091/jdaid.2026.16857.1103
HARVARD
Mirpour Marzuni, S. (2026). 'Interpretable Brain Tumor Classification via Triple-Stream Fusion of Local, Global, and Topological Features', Journal of Data Analytics and Intelligent Decision-making, (), pp. -. doi: 10.22091/jdaid.2026.16857.1103
VANCOUVER
Mirpour Marzuni, S. Interpretable Brain Tumor Classification via Triple-Stream Fusion of Local, Global, and Topological Features. Journal of Data Analytics and Intelligent Decision-making, 2026; (): -. doi: 10.22091/jdaid.2026.16857.1103