Hello,
Our University's MRI facility recently upgraded from an 8 channel to a 32 channel head coil. I have been doing some QC analyses to determine if this will affect our results with T1 enough to warrant a regression for the number of RF channels in the final study analysis. I ran FreeSurfer recon-all cortical measurements two times on the 8 channel T1's (I always do the recon-all with two consecutively acquired T1's), and two times on the 32 channel T1's. I did it twice on each because some members in our lab have mentioned that in the past they have re-ran the same analysis and got different results. This was not the case for me as I got identical results for both runs.
I ran a t-test on all of the cortical results for 32 vs 8 channels and there is a significant difference p<.019 between the two in the LH aparc measurements, but no significant difference between the 8 and 32 channel scans in the RH aparc measurements (p<.280). Is there anything inherent to the method FreeSurfer uses for its' cortical measurements that would result in an inconsistency only in the left hemisphere, or is it safe to chalk this up to a difference between head coils? Also of note, the two sets of scans were done only 6 months apart on the same individual (me).
Here is what I ran if it helps at all:
recon-all -s FShc12_redo -i /data/birc/RESEARCH/DFITZGERALD_projects/664-2008/HC12/acqfiles/DICOM/T1-1/DICOM/o20101113_104703IM_0001T1W3DTFESENSE664-2008-HC12s601a1006.nii.gz -i /data/birc/RESEARCH/DFITZGERALD_projects/664-2008/HC12/acqfiles/DICOM/T1-2/DICOM/o20101113_104703IM_0001T1W3DTFESENSE664-2008-HC12s701a1007.nii.gz -autorecon-all
and then,
aparcstats2table --subjects FShc12_redo --hemi rh --meas thickness --tablefile aparc_rh_thickness_stats_May_10_2011.txt
and
aparcstats2table --subjects FShc12_redo --hemi lh --meas thickness --tablefile aparc_lh_thickness_stats_May_10_2011.txt
Thank you in advance,
Joe
Hi Joe,
my guess is it has more to do with coil geometry than FS. The 8 channel coils are not very homogeneous in the cortex. The 32 are in general much better for surface recons. You could look at SNR or CNR maps and see if they correlate with reliability
cheers Bruce
On Tue, 10 May 2011, Gullett,Joseph M wrote:
Hello,
Our University's MRI facility recently upgraded from an 8 channel to a 32 channel head coil. I have been doing some QC analyses to determine if this will affect our results with T1 enough to warrant a regression for the number of RF channels in the final study analysis. I ran FreeSurfer recon-all cortical measurements two times on the 8 channel T1's (I always do the recon-all with two consecutively acquired T1's), and two times on the 32 channel T1's. I did it twice on each because some members in our lab have mentioned that in the past they have re-ran the same analysis and got different results. This was not the case for me as I got identical results for both runs.
I ran a t-test on all of the cortical results for 32 vs 8 channels and there is a significant difference p<.019 between the two in the LH aparc measurements, but no significant difference between the 8 and 32 channel scans in the RH aparc measurements (p<.280). Is there anything inherent to the method FreeSurfer uses for its' cortical measurements that would result in an inconsistency only in the left hemisphere, or is it safe to chalk this up to a difference between head coils? Also of note, the two sets of scans were done only 6 months apart on the same individual (me).
Here is what I ran if it helps at all:
recon-all -s FShc12_redo -i /data/birc/RESEARCH/DFITZGERALD_projects/664-2008/HC12/acqfiles/DICOM/T1-1/DICOM/o20101113_104703IM_0001T1W3DTFESENSE664-2008-HC12s601a1006.nii.gz -i /data/birc/RESEARCH/DFITZGERALD_projects/664-2008/HC12/acqfiles/DICOM/T1-2/DICOM/o20101113_104703IM_0001T1W3DTFESENSE664-2008-HC12s701a1007.nii.gz -autorecon-all
and then,
aparcstats2table --subjects FShc12_redo --hemi rh --meas thickness --tablefile aparc_rh_thickness_stats_May_10_2011.txt
and
aparcstats2table --subjects FShc12_redo --hemi lh --meas thickness --tablefile aparc_lh_thickness_stats_May_10_2011.txt
Thank you in advance,
Joe
Hi Joseph,
you are using a single subject? So how do you get the pvalue? And 6 month is not a small amount of time.
I guess you need several subjects before and after the upgrade (if possible within hours or days).
Martin
On Tue, 2011-05-10 at 11:31 -0400, Gullett,Joseph M wrote:
Hello,
Our University’s MRI facility recently upgraded from an 8 channel to a 32 channel head coil. I have been doing some QC analyses to determine if this will affect our results with T1 enough to warrant a regression for the number of RF channels in the final study analysis. I ran FreeSurfer recon-all cortical measurements two times on the 8 channel T1’s (I always do the recon-all with two consecutively acquired T1’s), and two times on the 32 channel T1’s. I did it twice on each because some members in our lab have mentioned that in the past they have re-ran the same analysis and got different results. This was not the case for me as I got identical results for both runs.
I ran a t-test on all of the cortical results for 32 vs 8 channels and there is a significant difference p<.019 between the two in the LH aparc measurements, but no significant difference between the 8 and 32 channel scans in the RH aparc measurements (p<.280). Is there anything inherent to the method FreeSurfer uses for its’ cortical measurements that would result in an inconsistency only in the left hemisphere, or is it safe to chalk this up to a difference between head coils? Also of note, the two sets of scans were done only 6 months apart on the same individual (me).
Here is what I ran if it helps at all:
recon-all -s FShc12_redo -i /data/birc/RESEARCH/DFITZGERALD_projects/664-2008/HC12/acqfiles/DICOM/T1-1/DICOM/o20101113_104703IM_0001T1W3DTFESENSE664-2008-HC12s601a1006.nii.gz -i /data/birc/RESEARCH/DFITZGERALD_projects/664-2008/HC12/acqfiles/DICOM/T1-2/DICOM/o20101113_104703IM_0001T1W3DTFESENSE664-2008-HC12s701a1007.nii.gz -autorecon-all
and then,
aparcstats2table --subjects FShc12_redo --hemi rh --meas thickness --tablefile aparc_rh_thickness_stats_May_10_2011.txt
and
aparcstats2table --subjects FShc12_redo --hemi lh --meas thickness --tablefile aparc_lh_thickness_stats_May_10_2011.txt
Thank you in advance,
Joe
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Gullett,Joseph M -
Martin Reuter