Reproducibility of amygdala activation in facial emotion processing at 7T

Despite its importance as the prime method for non-invasive assessment of human brain function, functional MRI (fMRI) was repeatedly challenged with regards to the validity of the fMRI-derived brain activation maps. Amygdala fMRI was particularly targeted, as the amygdala’s anatomical position in the ventral brain combined with strong magnetic field inhomogeneities and proximity to large vessels pose considerable obstacles for robust activation mapping. In this high-resolution study performed at ultra-high field (7T) fMRI, we aimed at (1) investigating systematic replicability of amygdala group-level activation in response to an established emotion processing task by varying task instruction and acquisition parameters and (2) testing for intra- and intersession reliability. At group-level, our results show statistically significant activation in bilateral amygdala and fusiform gyrus for each of the runs acquired. In addition, while fusiform gyrus activations are consistent across runs and sessions, amygdala activation levels show habituation effects across runs. This amygdala habituation effect is replicated in a session repeated two weeks later. Varying task instruction between matching emotions and matching persons does not change amygdala activation strength. Also, comparing two acquisition protocols with repetition times of either 700 ​ms or 1400 ​ms did not result in statistically significant differences of activation levels. Regarding within-subject reliability of amygdala activation, despite considerable variance in individual habituation patterns, we report fair to good inter-session reliability for the first run and excellent reliability for averages over runs. We conclude that high-resolution fMRI at 7T allows for robust mapping of amygdala activation in a broad range of variations. Our results of amygdala 7T fMRI are suitable to inform methodology and may encourage future studies to continue using emotion discrimination paradigms in clinical and non-clinical applications. •fMRI response to repeated runs of a facial emotion discrimination task was assessed at ultra-high field (7 ​T).•Systematic replicability and reliability for 2 sessions of 3 runs each.•Highly reproducible group-level amygdala activations maps across all runs, instruction variants and acquisition protocols.•Statistically significant intra-session habituation effects in amygdala in both sessions.•High reliability of amygdala activation levels based on ICC using multiple runs.