Stable and convenient spatial registration of stand-alone NIRS data through anchor-based probabilistic registration

Abstract

For functional neuroimaging with near-infrared spectroscopy (NIRS), we recently introduced a probabilistic registration method that uses a reference magnetic resonance image (MRI) database instead of the subject's own MRI, and probabilistically registers the NIRS optode or channel positions onto a canonical brain template in the standard stereotactic brain coordinate systems. As an alternative method, we devised an anchor-based registration method utilizing roughly obtained anchor positions on the scalp instead of strictly defined landmarks such as 10/20 landmarks. This method uses a spherical coordinate system to seek a position in the reference MRI database that corresponds to the anchor position, and eventually presents NIRS optode and channel positions in the standard stereotactic brain coordinate system. For comparison against conventional probabilistic registration, we simulated NIRS optode holder placement on 100 synthesized virtual heads, and found holistic tendencies for probe position estimations were similar between the two methods. Comparison among anchor-based probabilistic registration, conventional probabilistic registration, and SPM-based registration via co-registration to a subject's own MRI revealed that intra-method variability was comparable to a small inter-method variability. Thus, anchor-based registration is a practical alternative, especially to avoid burdening a subject and to reduce experimental time.