Hi All,
I want to run an SPM first level GLM using data that is denoised using Conn's CompCor. I loaded my preprocessed images into Conn 17f and ran through the denoising step, and got the resultant niftiDATA .nii files.
But when I take those denoised .nii files back to SPM and try to feed them into the GLM as if they were the smoothed, warped, realigned preprocessed data, I get an error when running spm_spm.
The error is:
Error using spm_est_non_sphericity (line 196)
Please check your data: There are no significant voxels.
I looked at the volumes in the niftiDATA files and I have the correct # that SPM is expecting, and I looked at the data with fsleyes and they all look normal - in that they are all there. But, the values in the voxels are wildly different than what would be found if I input a typical smoothed warped bold image. in the denoised data, there are a lot of negative values (as expecting based upon the way the distribution of voxel values changes after denoising), which I suppose could be causing a problem in the estimation in SPM?
Has anyone tried doing this and run into this problem?
Perhaps there are defaults I need to shut off when running the SPM GLM if I feed in the denoised data?
I have attached a screenshot showing the spm error.
Thanks for any advice!
Best,
Emily
I want to run an SPM first level GLM using data that is denoised using Conn's CompCor. I loaded my preprocessed images into Conn 17f and ran through the denoising step, and got the resultant niftiDATA .nii files.
But when I take those denoised .nii files back to SPM and try to feed them into the GLM as if they were the smoothed, warped, realigned preprocessed data, I get an error when running spm_spm.
The error is:
Error using spm_est_non_sphericity (line 196)
Please check your data: There are no significant voxels.
I looked at the volumes in the niftiDATA files and I have the correct # that SPM is expecting, and I looked at the data with fsleyes and they all look normal - in that they are all there. But, the values in the voxels are wildly different than what would be found if I input a typical smoothed warped bold image. in the denoised data, there are a lot of negative values (as expecting based upon the way the distribution of voxel values changes after denoising), which I suppose could be causing a problem in the estimation in SPM?
Has anyone tried doing this and run into this problem?
Perhaps there are defaults I need to shut off when running the SPM GLM if I feed in the denoised data?
I have attached a screenshot showing the spm error.
Thanks for any advice!
Best,
Emily