There is now the option to experiment with the atlas in SPM12, labels_Neuromorphometrics.nii, which is made from 30 subjects according to the notes. This at least labels all the brain as far as I can see rather than having gaps that are in many atlases.
Simon
> -----Original Message-----
> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
> Behalf Of Helmut Nebl
> Sent: 02 October 2014 14:11
> To: [log in to unmask]
> Subject: Re: [SPM] Confirming that masks made in WFU PickAtlas have been
> transformed into MNI space
>
> Dear Alexander,
>
> good luck with your paper :-) Yes, as long as sufficient information is provided
> this should be alright of course. Unfortunately it seems quite common to just
> briefly mention a certain atlas, but not providing details how this atlas was
> used (peak coordinate, % of activated voxels, ...), as it seems to be trivial at
> first sight. If the atlas was used for labeling purpose this is less of an issue, as
> you still have figures. However, sometimes a cluster is classified as a certain
> fine-grained region based on the peak coordinate and then the whole story is
> about this particular region, ignoring that it's maybe just a small part of the
> cluster. For a proper ROI analysis based on a probabilistic brain atlas it might
> be of high relevance which way the maps were generated and thresholded, as
> there might be a high variability in the data the atlas is derived from and thus a
> low overlap, possibly resulting in very small ROIs when using a certain
> threshold (e.g. some of the cyctoarchitectonic maps in Anatomy toolbox)
> and/or making it impossible to properly differentiate between different
> subregions. Thus, having a good atlas is one issue, but it always depends on
> how you use it.
>
> Best,
>
> Helmut
|