Dear SPMers,
in a recent paper by Pereira et al. on Resgistration accuracy for VBM
studies, NeuroImage 2010;49:2205-15 DARTEL as implemented
in SPM5 is presented as very insensitive to focal GM changes - in their
work, they used a small Alzheimer Disease sample to
to (amongmany other analyses) check the sensitivity to detect hippocampal
atrophy.
However, sensitivity and regional specificity were completely regained when
three preprocessing steps were performed:
1. skull stripping, 2. bias correction, and 3., "fine tuned BET".
> Has anybody made similar observations on insensitivity of DARTEL?
> Is the SPM Bias correction (default parameters) really so non-optimal?
> Is it the order of steps (first skull stripping, then bias correction)
the crucial difference?
> Is DARTEL in SPM8 in combination with the "New Segment" function more
sensitive?
I think for VBM where DARTEL has (my perception) repeatedly been
recommended as excellent/best registration tool,
it would be very useful if also the preprocessing/optimizations can be
performed within SPM. - It is somewhat
strange that extra preprocessing steps seem necessary outside SPM.
I am very interested in comments/opinions here,
thanks,
Philipp
Max Planck Institute of Psychiatry
NMR Research Group
Kraepelinstr. 2-10
80804 Munich
Mail: [log in to unmask]
Phone: 0049-89-30622-413
|