Figure 1

Analysis schemes for cross-sectional comparison. (A) Schematic example for an iterative template-specific MNI-normalization: after a 1^st normalization step based on landmarks, first templates T₁ ((b = 0) template and FA-template) were obtained by arithmetic averaging of DTI-data I₀. Analyses were performed with or without quality control (QC) and subsequent gradient direction elimination. Subsequently, in an iterative procedure, normalized DTI-data I₁ were obtained by non-linear normalization to the previously defined templates (T₁). From these newly normalized DTI-data I₁, new templates (T₂) were derived which again could be used for normalization. This iterative process is stopped when a predefined coincidence (measure by correlation) between DTI-data and templates was reached. (B) Scheme for whole brain-based spatial statistics: FA-maps are calculated from normalized DTI data and a smoothing filter to the individual normalized FA-maps is applied. In a consecutive step, voxelwise statistical comparison between the patient groups and the corresponding control group is performed. Final steps are correction for multiple comparisons using the false-discovery-rate (FDR) algorithm and a clustering procedure for further reduction of type I and type II errors.