Hi, I followed these steps... I registered each subjects brain to the MNI152 brain first. 1) echo "X Y Z" | std2imgcoord -img $FSLDIR/data/standard/MNI152_T1_1mm -std $FSLDIR/data/standard/MNI152_T1_1mm -vox - 2) write down the numbers from step 1 (or store them in a variable using backquotes if you are familiar with that) 3) fslmaths $FSLDIR/data/standard/MNI152_T1_1mm -mul 0 -add 1 -roi A 1 B 1 C 1 0 1 -kernel sphere 8 -dilF sphereROI I will investigate the registration issue further... It DOES seem very likely that this would be the cause. Thanks, Catherine On 4/17/13, zhang mingxia <[log in to unmask]> wrote: > Hi-According to my experience, it is easy to use the MNI coordinates which > is wrong. You should use the FSL coordinates. Also take care FSL > coordinates are different in 1mm and 2mm space. > Good luck. > > Mingxia > > > On Wed, Apr 17, 2013 at 5:21 AM, Mark Jenkinson < > [log in to unmask]> wrote: > >> Hi, >> >> I agree with Wolf that it may well be a registration issue, with images >> in >> the wrong spaces. It would be helpful if you could post the exact >> commands >> that you were running. >> >> All the best, >> Mark >> >> wolf zinke <[log in to unmask]> wrote: >> >> >> Hi, >> >> Maybe this is a problem with the transformation into the individual >> subject brain space? Otherwise it might be a good idea to mask the >> spherical ROI with a brain mask to ensure that it is not exceeding the >> brain. >> >> good luck, >> wolf >> >> On 04/16/2013 08:28 PM, Nucleos P wrote: >> > Hi Wolf - >> > >> > This helps...however, when I run fslmeants using the ROI sphere mask >> > it shows all zeros for individual subjects. I've done more constrained >> > analysis on the same image (e.g. less than 8mm radius) and gotten >> > values. >> > >> > Thank you, >> > Cat >> > >> > >> > On 4/15/13, wolf zinke <[log in to unmask]> wrote: >> >> Hi, >> >> >> >> fslroi uses voxel, but would not be the right tool for your purpose >> >> anyway, since it changes the image size, cropping it to the defined >> region. >> >> >> >> If you want to create a cubic mask you can use the -roi option of >> >> fslmaths. This option uses voxel coordinates, so you have to adjust >> >> the >> >> values according to your voxel dimension. >> >> >> >> If you want to create a more sphere-like ROI you can do follow this >> >> description in a recent post: >> >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;6ce4bff3.1301 >> >> >> >> I hope this helps, >> >> wolf >> >> >> >> On 04/16/2013 12:52 AM, Nucleos P wrote: >> >>> Hello Experts - >> >>> >> >>> When using fslroi it is unclear if the x size, y size, and z size >> >>> refer to mm or # of voxels around the coordinates? >> >>> >> >>> For instance, if I wanted to create a mask at ROI 125, 97, 107 MNI >> >>> 152 >> >>> voxel coordinates and then extend to points outside (e.g. 8mm) how >> >>> would I set this up? >> >>> >> >>> Thank you, >> >>> Cat >> >