An Explainable MRI-based Brain Tumor Diagnosis Framework with Grad-CAM Visualization
Journal of Soft Computing Paradigm · 07.07.2026 · 8 (3) · 239-252
DOI: 10.36548/jscp.2026.3.004 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Soft Computing Paradigm, online erschienen am 07.07.2026. Verfasst von Mouthika G. und 4 weiteren Autor:innen. Die Arbeit nennt 14 Literaturangaben.
Abstract (Originalsprache)
This research work presents an explainable deep learning approach to automatically detect, classify, localize, and estimate the severity of the brain tumors using MRI (Magnetic Resonance Imaging) images. It involves a VGG16-based transfer learning model with a two-stage approach of classification including first binary tumor detection and then multiclass classification of tumors. Preprocessing and data augmentation of brain MRI images is carried out before classification. Grad-CAM is applied to the proposed approach to visualize the activation map highlighting the most relevant areas of the image used by the classifier. The activation maps can be further used to approximate the localization and pixel-based severity estimation of the tumor. The experiments were conducted using a publicly available MRI dataset, and 97.56% accuracy for the binary classification and 90.38% for the multiclass classification were achieved.
Zitieren
APA
G., M., V., H., Srinivasan, G., L., A. S., & A.K., J. (2026). An Explainable MRI-based Brain Tumor Diagnosis Framework with Grad-CAM Visualization. Journal of Soft Computing Paradigm, 8(3), 239-252. https://doi.org/10.36548/jscp.2026.3.004
Vancouver
G. M, V. H, Srinivasan G, L. AS, A.K. J. An Explainable MRI-based Brain Tumor Diagnosis Framework with Grad-CAM Visualization. Journal of Soft Computing Paradigm. 2026;8(3):239-252. doi:10.36548/jscp.2026.3.004
BibTeX
@article{G2026,
author = {G., Mouthika and V., Hariganesh and Srinivasan, Ganesh and L., Ashika Sharon and A.K., Jayanthy},
title = {An Explainable MRI-based Brain Tumor Diagnosis Framework with Grad-CAM Visualization},
journal = {Journal of Soft Computing Paradigm},
year = {2026},
volume = {8},
number = {3},
pages = {239--252},
doi = {10.36548/jscp.2026.3.004}
}
EUVIMED