Electronics, Journal Year: 2025, Volume and Issue: 14(9), P. 1741 - 1741

Published: April 24, 2025

Accurately detecting and classifying brain tumors in magnetic resonance imaging (MRI) scans poses formidable challenges, stemming from the heterogeneous presentation of need for reliable, real-time diagnostic outputs. In this paper, we propose a novel multi-path convolutional architecture enhanced with channel-wise attention mechanisms, evaluated on comprehensive four-class tumor dataset. Specifically: (i) design parallel feature extraction strategy that captures nuanced morphologies, while refines salient characteristics; (ii) employ systematic data augmentation, yielding balanced dataset 6380 MRI to bolster model generalization; (iii) compare proposed against state-of-the-art models, demonstrating superior performance 97.52% accuracy, 97.63% precision, 97.18% recall, 98.32% specificity, an F1-score 97.36%; (iv) report inference speed 5.13 ms per scan, alongside higher memory footprint approximately 26 GB, underscoring both feasibility clinical application importance resource considerations. These findings collectively highlight framework’s potential improving automated detection workflows prompt further optimization broader deployment.

Language: Английский