Comparison of Brain Masking Techniques in Functional Magnetic Resonance Imaging

Exclusion of non-brain matter in functional magnetic resonance imaging is required for performing functional localization in particular, as well as functional image analysis in general. This paper presents a comparison of the commonly used brain masking techniques in functional magnetic resonance imaging. In addition, a masking algorithm based on connected components analysis along with Gaussian smoothing is introduced. The contiguous binary brain mask obtained by the new algorithm is shown to be more effective than the ones obtained by the commonly deployed masking algorithms. The introduced algorithm can also be applied to different image modalities that need to be registered to the space of anatomy or to the native space of echo planar images.