Comparison of multi-echo denoising techniques for auditory brainstem fMRI in the presence of task-correlated motion

Motivation: Task-based brainstem fMRI mapping is crucial to understand whole-brain system-level functioning, but data are plagued by low tSNR and task-correlated artifacts. Goal(s): We tested multi-echo denoising techniques in auditory brainstem data with and without task-correlated noise. Approach: Nine healthy individuals received passive auditory stimuli with (Audio+Motor) and without (Audio-only) a simultaneous shoulder abduction motor task during multi-echo fMRI. Results: Both single-echo and multi-echo techniques revealed clusters of significant activation in the inferior colliculus in Audio-only and Audio+Motor data. Despite greater t-statistics with single-echo data in the Audio+Motor dataset, multi-echo ICA yielded higher tSNR without a dependence on motion, suggesting more robust data quality. Impact: Task-based brainstem fMRI is greatly impacted by numerous confounds, including task-correlated motion. Understanding the capabilities of multi-echo denoising techniques in the brainstem will allow more reliable and robust data quality in clinical cohorts that exhibit higher levels of motion.