Dear Chaitanya,
I'd say the emphasis in the manual is on "The most interesting parameters to change here *may be*". You *might* want to use custom tissue probability maps in some instances (particular patient groups with large atrophy, children, lateralisation analyses to compare the two hemispheres), in which you don't want to rely on the DARTEL algorithm and use the default segmentation approach instead, which should be less accurate in general. In these cases you would have to adjust some of the settings.
Going with DARTEL is alright, see for example the paper by Bezzola, Mérillat, Gaser, and Jäncke (2011, J Neurosci) "Training-Induced Neural Plasticity in Golf Novices".
Best,
Helmut
|