It was the only explanation of why it would suddenly stop working that I could think of. I was reasoning that the headers may put the images out of alignment with each other, but that does not seem to be the case. I don't think that a change in datatype could cause such a problem, but I may be wrong. Maybe NaNs have entered into the data. I'm not sure how the code deals with them. I'll be away from the mailing list for a few days, but email me again in a week if you haven't figured out the cause of the problem. Best regards, -John On Mon, 2009-06-29 at 15:07 -0500, Guofan Xu wrote: > Thanks for the quick answer. Those N3-correction was done with freesurfer and converted back to nifti format. They looks well in spm5 "display" & "check-reg" functions. Is there any specific head info or field I should check? > > guofan > 6.29 > > >>> John Ashburner <[log in to unmask]> 6/29/2009 3:06 PM >>> > It sounds like a file format issue to me, as I'm not sure whether there > is a NIfTI compatible version of N3. Use Check Reg to make sure that > the bias corrected images are in alignment with each other. > > Best regards, > -John > > > On Mon, 2009-06-29 at 11:53 -0500, Guofan Xu wrote: > > Hi, Nikos: > > Actually I got the same error from High Dimensional Warping from my data and the results (Jacobians) is NOT good. I used N3-correction for my T1-MPRAGE data and SPM HDW gave me the same complain as you had. Before the N3-correction, HDW runs OK and I got a brain-shaped Jacobian file. I am curious why HDW gave me such error as well. > > > > > > Guofan > > 6.29 > > > > > > Guofan Xu > > Department of Medicine > > Geriatrics & Gerontology Section > > Veterans Administration Hospital > > 2500 Overlook Terrace > > Madison, WI 53705 > > > > > > >>> Nikos Koutsouleris <[log in to unmask]> 6/29/2009 4:04 AM >>> > > Dear John & VBM experts, > > > > I have two question regarding HDW: > > > > 1) When running spm_hdw I get the following warning during the differential > > bias field correction step : > > > > Warning: Matrix is singular, close to singular or badly scaled. > > Results may be inaccurate. RCOND = NaN. > > > In spm_hdw>bias_correction at 208 > > In spm_hdw>run_warping at 30 > > In spm_hdw at 14 > > In spm_jobman>run_struct1 at 1585 > > In spm_jobman>run_struct1 at 1595 > > In spm_jobman>run_struct1 at 1595 > > In spm_jobman>run_job at 480 > > In spm_jobman at 89 > > > > is this a serious problem that may affect my results or can I safely ignore it. > > > > 2) Is there any possibility to parallelize spm_warp.c by means of OpenMP, > > e.g. using a simple #pragma omp parallel directive at some for loop in the > > source code? > > > > Any help would be greatly appreciated! > > > > Best, > > > > Nikos > > -- John Ashburner <[log in to unmask]>