Key points are not available for this paper at this time.
In recent years, fMRI at ultrahigh magnetic field strengths (≥7T) has shifted from group analyses to probing neural processing in the individual brain. Identifying neurosignatures requires detection of BOLD effects with high sensitivity and spatial accuracy. Yet, it remains a challenge to enhance the sensitivity of fMRI for the BOLD effect without blurring the spatial details. Here, we assess the quality of the Gaussian, spatial adaptive non-local means (SANLM) and the adaptive weights smoothing (AWS) filters by employing a synthetic fMRI dataset as ground truth. AWS provides superior localization of the BOLD activations with high sensitivity at reasonable noise levels.
Ceja et al. (Wed,) studied this question.