External Email - Use Caution
Hi Doug,
I have been running mri_gtmpvc RBV-PVC and have noticed that in a few regions the RBV corrected values are negative. This occurs when using –no-rescale so I don’t think it is a reference region issue. I also tried using the –wm-annot option in mri_gtmseg and this resulted in slightly more regions with a negative value following RBV PVC. This suggested to me that the matrix may be ill-conditioned, but I was wondering what you think and if you have seen this before? I included a screen shot showing a couple regions with negative values. The raw counts are slightly lower in these regions but not significantly below the average.
The data is from an HR+ scan (~6mm FWHM) from AD subject so the cortex is pretty thin and regions that this is occurring in are fairly small (e.g. left caudal anterior cingulate, right vessel, right accumbens, right transverse temporal).
If you think it is the case that the matrix is ill-conditioned is there a way to reduce the number of regions by combining them automatically? I can generate a new parcellation and feed it into mri_gtmseg, but I wasn’t sure if you had something set up already.
Best, Jon
