Abstract
We investigated the decrease in intersubject functional variability in the horizontal meridian (HM) of the primary visual area (V1) before and after individual anatomical variability was significantly reduced using a high-resolution spatial normalization (HRSN) method. The analyzed dataset consisted of 10 normal, right-handed volunteers who had undergone both an O-15 PET study, which localized retinotopic visual area (V1), and a high-resolution anatomical MRI. Individual occipital lobes were manually segmented from anatomical images and transformed into a common space using an in-house high-resolution regional spatial normalization method called OSN. Individual anatomical and functional variability was quantified before and after HRSN processing. The reduction of individual anatomical variability was judged by the reduction in gray matter (GM) mismatch and by the improvement in overlap frequency between individual calcarine sulci. The reduction in intersubject functional variability of HM was determined by measurements of the overlap frequency between individual HM areas and by improvement in intersubject Z-score maps. The HRSN processing significantly reduced the individual anatomical variability: GM mismatch was reduced by a factor of two and the mean calcarine sulcus overlap frequency was improved from 37 to 68%. The reduction in functional variability was more subtle. However, both HM mean overlap (increased from 18 to 28%) and the average Z-score (increased from 2.2 to 2.55) were significantly improved. Although, functional registration was significantly improved by matching sulci, there was still residual variability. This is believed to be the variability of individual areas within the calcarine sulcus, and cannot be resolved by sulcal match. Thus, the proposed methodology provides an efficient, unbiased, and automated way to study structure-functional relationship in human brain.