 |
 |
Hi, Jesper I think I found the reason: it is the difference of pixdim3 between the imain and mask that caused "eddy" not run. I used "fslcreatehd" in the previous step, which is only accurate up to 5 decimals (1.00004), causing the difference in the headers between the original and mask images on the pixdim3 value. Thanks Longchuan ________________________________ From: Longchuan Li <[log in to unmask]> To: [log in to unmask] Sent: Monday, June 17, 2013 3:51 PM Subject: Re: [FSL] a question of using "Eddy" Hi, Jesper Thank you very much for the information. I checked the data using the command you provided and got the following: for imain: dim0 4 dim1 94 dim2 94 dim3 52 dim4 258 dim5 1 dim6 1 dim7 1 pixdim0 0.0000000000 pixdim1 1.0000000000 pixdim2 1.0000000000 pixdim3 1.0000419617 pixdim4 1.0000000000 pixdim5 0.0000000000 pixdim6 0.0000000000 pixdim7 0.0000000000 phase_dim 0 freq_dim 0 slice_dim 0 and for the mask: dim0 3 dim1 94 dim2 94 dim3 52 dim4 1 dim5 1 dim6 1 dim7 1 pixdim0 0.0000000000 pixdim1 1.0000000000 pixdim2 1.0000000000 pixdim3 1.0000400543 pixdim4 0.0000000000 pixdim5 0.0000000000 pixdim6 0.0000000000 pixdim7 0.0000000000 phase_dim 0 freq_dim 0 slice_dim 0 I realize that they have different dim0, which is something I am never aware of before. Could you explain what it is and whether it is the problem causing eddy to fail? Also, these averaged b0s are actually co-registered together before being averaged. I originally thought that topup may benefit from the increased SNR of the averaged b0s. So you do not think it is a good practice? Thank you again for the help! Longchuan ________________________________ From: Jesper Andersson <[log in to unmask]> To: [log in to unmask] Sent: Monday, June 17, 2013 2:35 PM Subject: Re: [FSL] a question of using "Eddy" Dear Li, I enountered a problem using "eddy" (FSL build 504) and would appreciate your help on it: > >I acquired two repetitions of diffusion data with opposite phase blips and each repetition with diffusion direction of 128 (totally 129, including 1 b0, which is the averaged value of 11 b0 images). Since I used S-T diffusion gradient, I want to test if "eddy" will provide extra benefits over "eddy_correct". > In general I would recommend you putting all 11 b0 in there instead of averaging them. If you do eddy will correct for any motion and you can always average them afterwards if you want to. >However, when I used "eddy" after "topup", I was given an error message like this: > > eddy: msg=--imain and --mask images must have the same dimensions >terminate called after throwing an instance of 'EDDY::EddyExceptio >n' > what(): eddy: msg=--imain and --mask images must have the same >dimensions. >my --imain is the combined images of two repetitions wiht opposite blips, with total volume of 129*2. I checked the dimension information of the imain and mask, they are identical except the fourth dimension. >I searched mailing list and did not find any related posts, so my questions are: > >(1) how could these two images have sam dimensions, since one is 4D and one is 3D? > It is comparing the mask image to the first 3D volume in your 4D data. I just checked the code and I can't see any obvious mistakes so I suspect that there is something different in your mask image compared to in your 4D data. if you do fslhd your_image | grep dim it should show you the relevant information. (2) Do I need to concatemante bvec and bval so that they have 258 volumes? > Yes you do. Eddy doesn't demand different blips, nor does it demand that you scan both blips with the same bvecs so it needs to know the bvec and bvalue for each scan. In your case that means duplicating the values, but in another case it will be all unique values. Jesper >Many thanks in advance > >Longchuan > > >