Multi-site fMRI analysis using privacy-preserving federated learning and domain adaptation: ABIDE results

Abstract

Deep learning models have shown their advantage in many different tasks, including neuroimage analysis. However, to effectively train a high-quality deep learning model, the aggregation of a significant amount of patient information is required. The time and cost for acquisition and annotation in assembling, for example, large fMRI datasets make it difficult to acquire large numbers at a single site. However, due to the need to protect the privacy of patient data, it is hard to assemble a central database from multiple institutions. Federated learning allows for population-level models to be trained without centralizing entities' data by transmitting the global model to local entities, training the model locally, and then averaging the gradients or weights in the global model. However, some studies suggest that private information can be recovered from the model gradients or weights. In this work, we address the problem of multi-site fMRI classification with a privacy-preserving strategy. To solve the problem, we propose a federated learning approach, where a decentralized iterative optimization algorithm is implemented and shared local model weights are altered by a randomization mechanism. Considering the systemic differences of fMRI distributions from different sites, we further propose two domain adaptation methods in this federated learning formulation. We investigate various practical aspects of federated model optimization and compare federated learning with alternative training strategies. Overall, our results demonstrate that it is promising to utilize multi-site data without data sharing to boost neuroimage analysis performance and find reliable disease-related biomarkers. Our proposed pipeline can be generalized to other privacy-sensitive medical data analysis problems. Our code is publicly available at: https://github.com/xxlya/Fed_ABIDE/.

Keywords: ABIDE; Data sharing; Domain adaptation; Federated learning; Privacy; Rs-fmri.

Fig. 12:

Interpreting brain biomarkers associated with identifying Male among ASD subjects from federated learning model (Fed) and using single site data for training (Single). The colors stand for the relative importance scores of the ROIs and the values are denoted on the color bar. The names of the strategies and sites are denoted on the left-upper corners of each subfigure. Each row shows the results of NYU, UM, UCLA site from top to bottom.

Fig. 1:

fMRI distribution of different sites

Fig. 2:

The simplified example of privacy-preserving federated learning strategy for fMRI analysis.

Fig. 3:

Domain adaptation strategies for our proposed federated learning setup.

Fig. 4:

Investigate communication pace τ vs accuracy

Fig. 5:

Investigate Gaussian mechanism vs accuracy

Fig. 6:

Investigate Laplace mechanism vs accuracy

Fig. 7:

Different classification strategies

Fig. 8:

t-SNE visualization of latent space.

Fig. 9:

The histogram of MoE gated values assigned to federated global model.

Fig. 10:

Interpreting brain biomarkers associated with identifying HC from federated learning model (Fed) and using single site data for training (Single). The colors stand for the relative importance scores of the ROIs and the values are denoted on the color bar. The names of the strategies and sites are denoted on the left-upper corners of each subfigure. Each row shows the results of NYU, UM, USM, UCLA site from top to bottom.

Fig. 11:

Interpreting brain biomarkers associated with identifying ASD from federated learning model (Fed) and using single site data for training (Single). The colors stand for the relative importance scores of the ROIs and the values are denoted on the color bar. The names of the strategies and sites are denoted on the left-upper corners of each subfigure. Each row shows the results of NYU, UM, USM, UCLA site from top to bottom.