Hi, I am trying to use fslmaths to filter my data before I plug it into a resting state analysis. However, once I obtain a pcc map, all correlation coefficients are <0.1, even the autocorrelation with the seed voxel. The data look fine when I open them in Matlab using MRIread; the correlations between individual voxel time courses obtained in Matlab is about 0.6.
My analysis stream is as follows: a) I filter the data using "fslmaths input -bptf highpass lowpass output"; from this I obtain a filtered nii.gz file b) I extract a seed timecourse from the filtered data in Matlab c) I run mkanalysis-sess with -notask d) I run selxavg3-sess e) I overlay the pcc.nii
I have tried this with smoothed and unsmoothed data, with and without nuisance regressors, with different seeds, with mri_convert after step a. It is not a problem with the overlay since the max values in pcc are all <0.1. I noticed that there are some differences in the nifti header (as read with MRIread) before and after filtering, but it seemed to me that they are gone after I used mri_convert. I do find much higher correlations with unfiltered data. Any advice what could be going wrong here? I am using Freesurfer 5.1 and FSL 4.1.9 on Linux. Thanks, Caspar
Is this doing all of those seeds simultaneously or one seed at a time?
On 05/07/2012 02:25 PM, Caspar M. Schwiedrzik wrote:
Hi, I am trying to use fslmaths to filter my data before I plug it into a resting state analysis. However, once I obtain a pcc map, all correlation coefficients are <0.1, even the autocorrelation with the seed voxel. The data look fine when I open them in Matlab using MRIread; the correlations between individual voxel time courses obtained in Matlab is about 0.6.
My analysis stream is as follows: a) I filter the data using "fslmaths input -bptf highpass lowpass output"; from this I obtain a filtered nii.gz file b) I extract a seed timecourse from the filtered data in Matlab c) I run mkanalysis-sess with -notask d) I run selxavg3-sess e) I overlay the pcc.nii
I have tried this with smoothed and unsmoothed data, with and without nuisance regressors, with different seeds, with mri_convert after step a. It is not a problem with the overlay since the max values in pcc are all <0.1. I noticed that there are some differences in the nifti header (as read with MRIread) before and after filtering, but it seemed to me that they are gone after I used mri_convert. I do find much higher correlations with unfiltered data. Any advice what could be going wrong here? I am using Freesurfer 5.1 and FSL 4.1.9 on Linux. Thanks, Caspar
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Caspar M. Schwiedrzik -
Douglas N Greve