[FieldTrip] AAL atlas interpolation and parcellation

Fri Dec 25 02:09:29 CET 2020

Thank you so much! I really appreciate the help -- the cortical atlas was
exactly what I was looking for. You mentioned that the atlas is defined
from the same topological mesh as the individual subject data, but I
noticed that the .pos fields are different between the two, and FieldTrip
is asking that I interpolate. Is it appropriate to just set
source.pos=atlas.pos and move forward? I'm getting results that seem
potentially feasible, but I'm hesitant whether that step was done
appropriately. Thanks again!

Best,
Philip

On Sun, Dec 20, 2020 at 4:42 AM Schoffelen, J.M. (Jan Mathijs) <
jan.schoffelen at donders.ru.nl> wrote:

> Dear Philip,
>
> You’re right: since the source model that you used for inverse modelling
> is strictly constrained to the neocortex, the inverse result does not cover
> subcortical structures.
> In general, I’d recommend against using a volumetrically defined
> parcellation (as is the AAL atlas shipped with FieldTrip) for a
> cortically-constrained source reconstruction (or the other way around).
>
> I suggest you use the finer-grained Brodmann area constrained cortical
> atlas as available in the “Sensory modality independent activation of the
> brain network for language” data sharing collection on the Donders data
> repository. I noticed your name on the history of this collection, so I’d
> assume you know where to find it. The relevant file is the one whose name
> starts with “atlas”, in the metadata&templates folder. This atlas is
> defined on exactly the same topological mesh as the individual subject
> data, which makes the parcellation step very straightforward.
>
> Next to this, I’d recommend you to have a look at the relatively new
> ‘ft_virtualchannel’ function. There is a little bit of documentation on the
> website about this. The bottom line is, that this function supports a
> parcellation (in addition to a source-level structure with spatial filters
> and sensor level single trial data) in the input. This allows you to obtain
> parcel-level single trial time series, which is not possible with
> ft_sourceparcellate.
>
> Best wishes,
> Jan-Mathijs
>
>
> > On 20 Dec 2020, at 09:06, Philip Cho <pnc9 at georgetown.edu> wrote:
> >
> > Dear FieldTrip Community,
> >
> > I am currently trying to obtain the source-level signal in a few
> anatomically defined ROIs, and so I'm interpolating the AAL atlas onto the
> 8196-vertex cortical surface template (since I previously used that to
> inverse-normalize onto my subject MRIs), and I'm using ft_sourceparcellate
> to obtain my results. However, I'm facing an error, which I believe stems
> from the fact that some labels (e.g. putamen, pallidum, cerebellum, vermis)
> are not present in my tissue after interpolation. Please let me know if any
> advice is available. Thank you very much.
> >
> > My code is as follows:
> >
> > function parc_source = mous2_sourceParcellate(sublist)
> >
> > %load atlas
> > atlas =
> ft_read_atlas('C:\fieldtrip-20201205\template\atlas\aal\ROI_MNI_V4.nii');
> >
> > %load sourcemodel (to get its .pos field)
> > ftpath   = 'C:\fieldtrip-20201205';
> > sourcemodel = ft_read_headshape(fullfile(ftpath,
> 'template\sourcemodel\cortex_8196.surf.gii'));
> >
> > cfg_int = [];
> > cfg_int.interpmethod = 'nearest';
> > cfg_int.parameter = 'tissue';
> > atlas_int = ft_sourceinterpolate(cfg_int,atlas,sourcemodel);
> >
> > for sub = sublist
> >     %load source
> >     load(['D:\MOUS2\results\sourceanalysis_low_depth\source_low_depth_'
> sub{1} '.mat']);
> >     source_allWords = ft_datatype_source(source_allWords);
> >     source_allWords.pos = sourcemodel.pos;
> >
> >     %parcellate
> >     cfg = [];
> >     cfg.method = 'mean';
> >     cfg.parameter = 'all';
> >     parc_source = ft_sourceparcellate(cfg,source_allWords,atlas_int);
> > end
> >
> >
> >
> > And the last part of the command line message, including the error, is
> as follows:
> >
> > there is no "Putamen_L" in "tissue"
> > there is no "Pallidum_L" in "tissue"
> > there is no "Pallidum_R" in "tissue"
> > there is no "Cerebellum_Crus1_R" in "tissue"
> > there is no "Cerebellum_Crus2_L" in "tissue"
> > there is no "Cerebellum_Crus2_R" in "tissue"
> > there is no "Cerebellum_3_L" in "tissue"
> > there is no "Cerebellum_3_R" in "tissue"
> > there is no "Cerebellum_4_5_L" in "tissue"
> > there is no "Cerebellum_6_L" in "tissue"
> > there is no "Cerebellum_6_R" in "tissue"
> > there is no "Cerebellum_7b_L" in "tissue"
> > there is no "Cerebellum_7b_R" in "tissue"
> > there is no "Cerebellum_8_L" in "tissue"
> > there is no "Cerebellum_8_R" in "tissue"
> > there is no "Cerebellum_9_L" in "tissue"
> > there is no "Cerebellum_9_R" in "tissue"
> > there is no "Cerebellum_10_L" in "tissue"
> > there is no "Cerebellum_10_R" in "tissue"
> > there is no "Vermis_1_2" in "tissue"
> > there is no "Vermis_3" in "tissue"
> > there is no "Vermis_4_5" in "tissue"
> > there is no "Vermis_6" in "tissue"
> > there is no "Vermis_7" in "tissue"
> > there is no "Vermis_8" in "tissue"
> > there is no "Vermis_9" in "tissue"
> > there is no "Vermis_10" in "tissue"
> > Warning: using "tissue" for the parcellation
> >  In ft_sourceparcellate at line 117
> >
> > there are in total 8196 positions, 8196 positions are inside the brain,
> 7222 positions have a label
> > 7222 of the positions inside the brain have a label
> > 7222 of the labeled positions are inside the brain
> > 974 of the positions inside the brain do not have a label
> > creating 116 parcels for parameter mom by taking the mean
> > computing parcellation
> > Index exceeds the number of array elements (0).
> >
> > Error in ft_sourceparcellate>cellmean1 (line 488)
> > y = x{1};
> >
> > Error in ft_sourceparcellate (line 226)
> >           tmp(j,:,:) = cellmean1(dat(seg==j));
> >
> > Error in mous2_sourceParcellate (line 25)
> >     parc_source = ft_sourceparcellate(cfg,source_allWords,atlas_int);
> > _______________________________________________
> > fieldtrip mailing list
> > https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> > https://doi.org/10.1371/journal.pcbi.1002202
>
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20201224/6c86e5c7/attachment.htm>