Hybrid Cnn–Transformer Federated Learning for Brain Tumor Segmentation Across Multi-Center Mri Datasets
International Journal of Drug Delivery Technology · 19.05.2026 · 16 (30s)
DOI: 10.25258/ijddt.16.30s.28 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 19.05.2026. Verfasst von Anil Saraswat und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Brain tumor segmentation using MRI is a vital step for early diagnosis and treatment; nevertheless, sharing multi-center data is not feasible due to privacy, legislation, and security issues. Traditional centralized deep learning techniques demand a joint dataset; hence, there is a high probability of leakage. Additionally, collaborative learning is not feasible. In light of these limitations, a comprehensive review on Federated Deep Learning (FDL) for brain tumor segmentation using CNN and Transformer architectures is presented. The methodology adopted is based on GAN for augmentation, feature selection using optimized techniques like Ant Lion Optimization (ALO) and Bat-Inspired Algorithm (BIA), federated learning for security, and differential privacy for non-IID conditions. The experimental study shows promising results for FDL; to be specific, CNN architectures like VGG-16 (99.86%), InceptionV3 (99.72%), and Bi-LSTM (99.65%) outperform traditional ML algorithms while achieving high ROC-AUC value up to 0.999. Federated learning ensures high accuracy for CNN-Transformer integration with minimal communication cost. The federated learning framework is highly accurate for brain tumor segmentation while maintaining high security standards for a collaborative setting using CNN-Transformer architectures.
Zitieren
APA
Saraswat, A., & Pandey, N. N. (2026). Hybrid Cnn–Transformer Federated Learning for Brain Tumor Segmentation Across Multi-Center Mri Datasets. International Journal of Drug Delivery Technology, 16(30s). https://doi.org/10.25258/ijddt.16.30s.28
Vancouver
Saraswat A, Pandey NN. Hybrid Cnn–Transformer Federated Learning for Brain Tumor Segmentation Across Multi-Center Mri Datasets. International Journal of Drug Delivery Technology. 2026;16(30s). doi:10.25258/ijddt.16.30s.28
BibTeX
@article{Saraswat2026,
author = {Saraswat, Anil and Pandey, Nageshwar Nath},
title = {Hybrid Cnn–Transformer Federated Learning for Brain Tumor Segmentation Across Multi-Center Mri Datasets},
journal = {International Journal of Drug Delivery Technology},
year = {2026},
volume = {16},
number = {30s},
doi = {10.25258/ijddt.16.30s.28}
}
EUVIMED