Caroline A. Sage1, Wim Van Hecke2, Ronald Peeters1, Jan Sijbers2, Wim Robberecht3, Judith Verhoeven4, Sabine Deprez1, Paul Parizel5, Guy Marchal1, Alexander Leemans6, Stefan Sunaert1
1Radiology, University Hospitals of the Catholic University of Leuven, Leuven, Belgium; 2Vision Lab, University of Antwerp, Wilrijk, Belgium; 3Neurology, University Hospitals of the Catholic University of Leuven, Leuven, Belgium; 4Pediatrics, University Hospitals of the Catholic University of Leuven, Leuven, Belgium; 5Radiology, University of Antwerp, Antwerp, Belgium; 6CUBRIC, School of Psychology, Cardiff University, Cardiff, UK
Results of voxel-based analyses of DTI data are dependent on various factors such as the coregistration technique and choice of reference system. In this study, non-rigid coregistration and a population-based DTI atlas that contains the complete diffusion information are used to validate our previously published results of voxel-based analyses in which DTI data of 28 patients with amyotrophic lateral sclerosis were compared to those of 26 healthy controls. We demonstrate that, as the residual variance after spatial normalization to the atlas is lowered, the results of the voxel-based analyses are improved, especially for the mean diffusivity.