Impact of different realignment algorithms on the SPM analysis of [/sup 11/C]raclopride PET studies

Abstract

In this paper we address the issue of how the preprocessing steps of a typical neuroreceptor PET study may influence the outcome of an SPM analysis examining inter-condition differences. Two preprocessing steps and their possible interaction were investigated: 1) the realignment of dynamic images to correct for patient's motion and 2) the effect of the template in the normalization step required by SPM. Methods: The impact of different realignment algorithms was tested by performing the SPM analysis on a six patient - two conditions /sup 11/C-raclopride (RAC) study where the dynamic images of each subject were re-aligned with four algorithms. Once realigned, the images were normalized to the standard cerebral blood flow (CBF) template. available in SPM, and to a RAC template. The SPM pattern showing the inter-condition differences obtained in each case was compared with the expected one. The accuracy of the four alignment algorithms used was further tested using a set of motion-free RAC studies. Results/Conclusion: We found that 1) different realignment algorithms may create different inter-condition difference patterns, 2) such differences in the results may be due to a failure in accuracy of some realignment algorithms, and 3) the use of a template that closely matches the spatial distribution of the tracer under investigation can minimize these problems.