Privacy-Preserving Brain Tumor Classification using VIT, Eigencam and Federated Learning
International Journal of Science, Strategic Management and Technology · 27.03.2026 · 02 (03) · 1
DOI: 10.55041/ijsmt.v2i3.273 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Science, Strategic Management and Technology, online erschienen am 27.03.2026. Verfasst von Suresh kumar und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Brain tumor diagnosis from MRI remains a critical challenge due to high inter-class morphological similarity, limited annotated data, and stringent patient privacy requirements in multi-institutional environments. This paper presents a novel unified framework integrating Vision Transformer (ViT) classification, EigenCAM-based Explainable AI (XAI), and Federated Learning (FL) with FedAvg aggregation. The ViT achieves 95% validation accuracy through frozen-backbone transfer learning from ImageNet-1k. EigenCAM generates spatially-precise attention heatmaps enabling tumor region localization with quantitative size estimation. The FL framework across three virtual hospital nodes achieves 93% accuracy in compliance with GDPR and HIPAA. Comparative evaluation against twelve state-of-the-art methods demonstrates that the proposed framework uniquely addresses interpretability, privacy, and classification accuracy simultaneously.
Zitieren
APA
kumar, S., & B, S. (2026). Privacy-Preserving Brain Tumor Classification using VIT, Eigencam and Federated Learning. International Journal of Science, Strategic Management and Technology, 02(03), 1. https://doi.org/10.55041/ijsmt.v2i3.273
Vancouver
kumar S, B S. Privacy-Preserving Brain Tumor Classification using VIT, Eigencam and Federated Learning. International Journal of Science, Strategic Management and Technology. 2026;02(03):1. doi:10.55041/ijsmt.v2i3.273
BibTeX
@article{kumar2026,
author = {kumar, Suresh and B, SanakaraLakshmi},
title = {Privacy-Preserving Brain Tumor Classification using VIT, Eigencam and Federated Learning},
journal = {International Journal of Science, Strategic Management and Technology},
year = {2026},
volume = {02},
number = {03},
pages = {1},
doi = {10.55041/ijsmt.v2i3.273}
}
EUVIMED