There is not an easy way to do this. FS expects an intact whole brain, and it tries to fit such into every brain, ie, it is trying to force a whole cerebellum where there is none. What some people do is to fill the lesion with "reasonable" intensities, eg, from the contralateral side. Eg, you could left-right reverse the subject: mri_convert orig.mgz --left-right-reverse-pix orig.lrrev.mgz mri_coreg --mov orig.lrrev.mgz --targ orig.mgz --reg lrrev.lta mri_vol2vol --mov orig.lrrev.mgz --targ orig.mgz --reg lrrev.lta --o orig.lrrev-in-orig.mgz Then replace the voxels in the lesion in orig.mgz with voxels from orig.lrrev-in-orig.mgz and re-run Let me know how it goes
On 9/18/19 6:52 PM, Sam W wrote: > > External Email - Use Caution > > Thanks Doug. The problem is that for some patients the lesion affects > the segmentation dramatically. For example I have one patient with a > large lesion in the right cerebellum which is partly in GM and partly > in WM. If I load the aseg file I see that a portion of the right > cerebellar hemisphere is assigned to the left cerebellar hemisphere > (the right white matter is also being assigned to the left hemisphere) > and the whole cerebellum looks distorted. I suppose there is no way to > inform freesurfer about the lesion during recon-all? > I followed the steps in the FsTutorials for the cerebral cortex, but > how can I do the same for the cerebellum patient? What is the > recommended way to fix a bad segmentation due to cerebellum lesions? > Control points, white matter edits, or something else? > Sam > > On Fri, Sep 6, 2019 at 4:28 PM Greve, Douglas N.,Ph.D. > <dgr...@mgh.harvard.edu <mailto:dgr...@mgh.harvard.edu>> wrote: > > You can get the volume from the aseg.stats file. Unfortunately, we > do not separate the lesions into left and right. You could do > something like > mri_binarize --i aseg.mgz --match 77 --o wmlesions.mgz > mri_volcluster --i wmlesions.mgz --sum lesions.sum.dat --thmin 0.5 > --regheader subject > This will output a list of lesions clusters in lesions.sum.dat and > their coordinates. you can see which ones are left and with ones > are right. > > > > On 9/3/2019 5:17 PM, Sam W wrote: >> >> External Email - Use Caution >> >> Hello, >> I have run recon-all on T1 scans of patients with WM lesions. I >> noticed however that for some patients the lesion is excluded >> from aparc.a2009s+aseg.mgz but for other patients it is included >> (and labelled as non-lesion). >> Ultimately I'd like to extract a) volume information in WM and b) >> volume information of the WM lesion. I think I can get the the WM >> volume from the wmparc.stats file. For the lesion volume I think >> I can take the WM-hypointensities from the aseg file right? >> However I noticed that if a lesion is on the right hemisphere, >> the Right-WM-hypointensities shows 0s in all columns, which >> cannot be right. >> I have a mask of the lesion (1s where lesion occurs, 0s >> elsewhere) in anatomical space, can I use this mask somehow in FS >> to inform recon-all where the lesion occurs? >> Thanks in advance! >> Sam >> >> _______________________________________________ >> Freesurfer mailing list >> Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer