A hybrid high-resolution anatomical MRI atlas with sub-parcellation of cortical gyri using resting fMRI

Brain atlases provide reference parcellations of the brain that are essential for population neuroimaging studies. We present a new high-resolution, single-subject atlas labeled into two parcellation schemes: 1) the anatomical BCI-DNI atlas, which was labeled based on known morphological and anatomical features; and 2) the hybrid USCBrain atlas, which used additional functional information to guide the sub-parcellation of cerebral cortex. Particular attention was paid to the image acquisition, processing and labeling methods to capture fine anatomical details, accommodating for the high-quality data common in recent imaging studies. A single-subject, high-resolution T1-weighted image was acquired and was then processed by an expert neuroanatomist using semi-automated methods in BrainSuite. The brain’s features were meticulously extracted with manual corrections to bias-field and masking steps, thereby providing accurate tissue classification and anatomical surface modeling. Guided by sulcal and gyral landmarks, labeled anatomical regions were drawn manually on coronal single-slice images to generate the BCI-DNI atlas, which contains 66 cortical and 29 noncortical regions. The cortical regions were further sub-parcellated based on connectivity analysis of resting fMRI data from multiple subjects in the Human Connectome Project (HCP) database, which were coregistered to the single subject. The resulting USCBrain atlas contains a total of 130 cortical and 29 noncortical regions. In addition to the anatomical and functional parcellations, we also provide a delineation between sulcal valleys and gyral crowns, which offer an additional set of 26 sulcal subregions per hemisphere. The intended use of the USCBrain atlas is to label individual brains through structural coregistration. To assess utility, we computed the adjusted Rand indices between individual sub-parcellations obtained through structural-only coregistration to the USCBrain atlas and sub-parcellations obtained directly from each subject’s resting fMRI data. The gyral sub-parcellations generated by atlas-based registration show variable but generally good overlap with the resting fMRI-based subdivisions. In addition to the crisp parcellations, a probabilistic map is included to provide users a quantitative measure of reliability for each gyral subdivision. Both atlases can be used with the BrainSuite, FreeSurfer, and FSL software packages.