Xiujuan Geng1, Hong Gu1, Thomas J. Ross1, Yihong Yang1
1Neuroimaging Research Branch, National Institute on Drug Abuse, National Institutes of Health, Baltimore, MD, USA
We propose an implicit reference-based group (IRG) registration method with a high-dimensional elastic deformation model to align structural or functional images within groups. The performance of the IRG method was evaluated using T1-weighted MR images with pre-defined ROIs and compared with reference-based registration methods. Results show that IRG registration produces better registration performance in terms of higher relative overlaps. IRG registration was also applied to an fMRI study with a motor and visual task. The statistical detection power is improved by the elastic registration compared to the affine approach and IRG registration further increases the power compared to reference-based methods.