Federated Fine-tuning of SAM-Med3D for MRI-based Dementia Classification
By: Kaouther Mouheb , Marawan Elbatel , Janne Papma and more
Potential Business Impact:
Helps AI find dementia from brain scans.
While foundation models (FMs) offer strong potential for AI-based dementia diagnosis, their integration into federated learning (FL) systems remains underexplored. In this benchmarking study, we systematically evaluate the impact of key design choices: classification head architecture, fine-tuning strategy, and aggregation method, on the performance and efficiency of federated FM tuning using brain MRI data. Using a large multi-cohort dataset, we find that the architecture of the classification head substantially influences performance, freezing the FM encoder achieves comparable results to full fine-tuning, and advanced aggregation methods outperform standard federated averaging. Our results offer practical insights for deploying FMs in decentralized clinical settings and highlight trade-offs that should guide future method development.
Similar Papers
FedSCA: Federated Tuning with Similarity-guided Collaborative Aggregation for Heterogeneous Medical Image Segmentation
Image and Video Processing
Helps AI learn from many hospital scans safely.
Federated Learning for Predicting Mild Cognitive Impairment to Dementia Conversion
Machine Learning (CS)
Predicts memory loss without sharing patient secrets.
Fine-Tuning Foundation Models with Federated Learning for Privacy Preserving Medical Time Series Forecasting
Machine Learning (CS)
Trains smart health models without seeing private data.