Dear FSL experts,

I attached the PNG files. The highres2standard_OK and highres2standard_strange files show differences in frontal and parietal convexities, though they may be a bit difficult to distinguish. The overlay file shows strange result (gray scale) overlaid on normal result (red). This would be easier to see the difference.

Best regards,

Jun

2016/02/15 13:52、Jun Miyata <[log in to unmask]> のメール：

Dear FSL experts,

I have run Melodic group ICA for the same rsfMRI data twice, but I had different results for highres2standard registration. For the 1st trial it worked well, but for the 2nd the final highres2standard.nii.gz was skewed like diamond shape.
I performed ICA from Melodic GUI. The parameters were the same except for the number of subjects. I confirmed that highres2standard_linear.nii.gz, highres2standard.mat, and standard2highres.mat files are identical between two trials, whereas highres2standard_warp.nii.gz and highres2standard.nii.gz are different. Thus I think the problem is in nonlinear registration.
I appreciate if anyone could explain the reason. I would like to attach the files but it exceed the maximum size. 

Thanks in advance,

Jun

<highres2standard_OK.png><highres2standard_strange.png><Overlay.png>