Enabling Factor Analysis on Thousand-Subject Neuroimaging Datasets

TL;DR

This paper enhances the scalability of multi-subject neuroimaging factor analysis methods, enabling processing of large datasets through optimized algorithms and distributed computing, thus facilitating advanced brain activity analysis.

Contribution

The paper introduces optimized, scalable implementations of the Shared Response Model and Hierarchical Topographic Factor Analysis for large neuroimaging datasets.

Findings

Achieved 99x and 1812x speedups on single-node implementations.

Demonstrated strong scaling with 3.3x and 5.5x efficiency on 20 nodes.

Supported weak scaling on synthetic data with 1024 subjects across 1024 nodes.

Abstract

The scale of functional magnetic resonance image data is rapidly increasing as large multi-subject datasets are becoming widely available and high-resolution scanners are adopted. The inherent low-dimensionality of the information in this data has led neuroscientists to consider factor analysis methods to extract and analyze the underlying brain activity. In this work, we consider two recent multi-subject factor analysis methods: the Shared Response Model and Hierarchical Topographic Factor Analysis. We perform analytical, algorithmic, and code optimization to enable multi-node parallel implementations to scale. Single-node improvements result in 99x and 1812x speedups on these two methods, and enables the processing of larger datasets. Our distributed implementations show strong scaling of 3.3x and 5.5x respectively with 20 nodes on real datasets. We also demonstrate weak scaling on a synthetic dataset with 1024 subjects, on up to 1024 nodes and 32,768 cores.