Federated Learning and Meta-Learning for Medical Image Classification: Effective Aggregation Techniques and Adaptive Learning Rate Approaches
Date
2024-08-23
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
Type
Thesis
Degree Level
Masters
Abstract
Computer-aided diagnosis (CAD) has the potential to revolutionize medical diagnosis through advanced
machine learning (ML) algorithms. However, the efficacy of CAD systems relies heavily on extensive, diverse,
and annotated datasets for robust training. This dissertation proposes a multifaceted strategy to address data
scarcity, privacy concerns, and diverse data distribution across healthcare institutions by harnessing two major
types of learning methods: federated learning (FL) and model-agnostic meta-learning (MAML). In the first
learning method, FL enables CAD systems to train on diverse data across multiple institutions, overcoming
individual dataset limitations. However, data heterogeneity (e.g., due to non-independent and identically
distributed data) among clients causes client model divergence, making it challenging to achieve convergence
during model aggregation. To address these challenges, we propose different aggregation techniques: randomsampling federated maximum (FedRSMax), random-sampling federated median (FedRSMed), and federated
rank-based client selection aggregation (FedRSA). In the second learning method, MAML has the potential to
facilitate swift adaptation to varying medical imaging datasets, effectively navigating discrepancies in imaging
protocols and disease presentations. However, MAML struggles with fixed learning rates, limiting adaptation
to varying task complexities. To address this challenge, we propose a task similarity-driven meta-adaptive
learning rate mechanism. Overall, this research pursues a two-fold strategy: (1) to develop an efficient
aggregation technique for medical image classification using FL; and (2) to employ MAML within a few-shot
learning framework, optimizing adaptation to datasets with limited samples. The proposed methodologies
are evaluated using a comprehensive set of performance metrics, including accuracy, convergence rate, and
adaptation loss. By integrating these cutting-edge techniques, this dissertation advances the frontier of CAD
systems, paving the way for more accurate, adaptable, and scalable solutions in medical image analysis.
Description
Keywords
Medical Imaging, Federated Learning, Meta Learning, Classification
Citation
Degree
Master of Science (M.Sc.)
Department
Biomedical Engineering
Program
Biomedical Engineering