Abstract
492
Objectives Respiratory and cardiac motion impair image quality and quantification in PET. In MRI-based motion correction of PET data, e.g., the motion information contained in PET data remains unused. Yet, both modalities can contribute valuable information to the motion detection process. In the present work we propose a combined registration functional that uses information of both modalities.
Methods We developed a registration functional in which two distance functionals, one for each modality, are combined into one. A weighting factor allows to vary the influence of both modalities. Evaluation is performed on XCAT software phantom data (using simulated MRI, CT, and PET data). Registration results are compared to ground-truth motion data by means of the average end-point error. Recovered PET activity for lesion regions is evaluated.
Results For PET and MRI data, the lowest global error is detected for a purely MRI-weighted registration. For individual regions, however, the lowest error is achieved if both modalities are considered. For PET-CT data, the lowest registration error (globally, and for individual regions) is achieved if both modalities are considered. The recovered activity in the lesion regions is higher when both modalities exert influence in the registration.
Conclusions A benefit of the joint registration approach compared to a registration based on MRI (CT) or PET data alone is observable . A lower registration error as well as a better recovery of lesion activity could be achieved. Currently, we are evaluating the proposed approach on patient data.
Research Support This research has been partly supported by a research grant from Siemens Healthcare, Erlangen, Germany, and the Deutsche Forschungsgemeinschaft (German research funding organisation), SFB 656 MoBil (project B3).