Secure Federated Learning for Neuroimaging

Abstract

The amount of biomedical data continues to grow rapidly. However, the ability to collect data from multiple sites for joint analysis remains challenging due to security, privacy, and regulatory concerns.We present a Secure Federated Learning architecture, MetisFL, which enables distributed training of neural networks over multiple data sources without sharing data. Each site trains the neural network over its private data for some time, then shares the neural network parameters (i.e., weights, gradients) with a Federation Controller, which in turn aggregates the local models, sends the resulting community model back to each site, and the process repeats.Our architecture provides strong security and privacy. First, sample data never leaves a site. Second, neural parameters are encrypted before transmission and the community model is computed under fully-homomorphic encryption. Finally, we use information-theoretic methods to limit information leakage from the neural model to prevent a "curious" site from performing membership attacks.We demonstrate this architecture in neuroimaging. Specifically, we investigate training neural models to classify Alzheimer's disease, and estimate Brain Age, from magnetic resonance imaging datasets distributed across multiple sites, including heterogeneous environments where sites have different amounts of data, statistical distributions, and computational capabilities.

Secure federated learning for Alzheimer's disease detection

“…Recently, Stripelis et al ( 2022 ) have also presented an FL architecture that utilizes Homomorphic Encryption (HE) to train ML models for AD detection. The authors studied several prominent AD datasets, splitting the dataset heterogeneously over clients only in terms of dataset sizes.…”

“…FL has been utilized in the medical imaging field through tasks such as X-ray pneumonia detection (Kaissis et al, 2021 ), whole-brain segmentation in MRI (Roy et al, 2019 ; Rieke et al, 2020 ), COVID-19 detection (Liu et al, 2020 ), and analysis of different neurological diseases, such as Alzheimer's disease (AD) (Stripelis et al, 2022 ). AD is the most common cause of dementia in the elderly (Ritter et al, 2015 ).…”

The pursuit of approaches to federate data to accelerate Alzheimer’s disease and related dementia research: GAAIN, DPUK, and ADDI

“…In federated learning, different data partners/clients train their neural networks, and a central model aggregates the parameters of that model ( Rieke et al, 2020 ). This approach allows the training of large-scale neural network models without the need to access centralized data ( Stripelis et al, 2022 ).…”

Medical Imaging Applications of Federated Learning

“…In 2022, Stripelis et al utilized an FL framework they proposed in their previous works and applied it to MRI datasets to classify Alzheimer's disease and estimate Brain Age [53].…”

“…While the group's previous works detailed their security enhancements and architecture development, this later work was more of a case study implementing their previous works on a heterogeneous dataset to prove the capabilities [31,39,53].…”