dke-questions > Index exceed matrix dimensions
Showing 1-5 of 5 posts
Jun 28, 2017 07:06 PM | wafaa sweidan
Index exceed matrix dimensions
Hello,
I have (neurite orientation dispersion and density imaging) NODDI data with following b-values:
b=0 (9 averages), b=711 (30 averages) and b=2855 (60), total of 99 gradient vector.
I generated the 4d Nifti for the diffusion image and when I try to fit it using DKE with following parameters (B=0 threshold=50, weak median filtering, tensor constrained linear weighted fitting, gradient vectors: user defined and I upload the bvec file (and i made sure to remove gradients for b0) it gives me the following error:
index exceed matrix dimensions
Error in dke estimate line 188
Error in dke line 182
I think the problem arises from the user defined gradient vectors, but I have to insert my gradient vectors
Can anyone please inform me on the problem and try to help in figuring the solution out?
Regards,
Wafaa
I have (neurite orientation dispersion and density imaging) NODDI data with following b-values:
b=0 (9 averages), b=711 (30 averages) and b=2855 (60), total of 99 gradient vector.
I generated the 4d Nifti for the diffusion image and when I try to fit it using DKE with following parameters (B=0 threshold=50, weak median filtering, tensor constrained linear weighted fitting, gradient vectors: user defined and I upload the bvec file (and i made sure to remove gradients for b0) it gives me the following error:
index exceed matrix dimensions
Error in dke estimate line 188
Error in dke line 182
I think the problem arises from the user defined gradient vectors, but I have to insert my gradient vectors
Can anyone please inform me on the problem and try to help in figuring the solution out?
Regards,
Wafaa
Jun 29, 2017 01:06 AM | Kirk Feindel - University of Western Australia
RE: Index exceed matrix dimensions
Hi Wafaa,
I think there may be a couple of issues.
1) Did you average your 9 B0 images so that there is only one at the beginning of the 4D Nifti?
2) I can't remember 100% if you can use a different number of directions for each b value, but if you can, I think you'll need two separate gradient vector files since you use 30 directions and then 60 directions. (Look at previous posts in this forum - I think this query has been addressed previously).
Also, if you attach your parameters file, 4D Nifti, and bvecs files people can try to debug it for you.
Cheers,
Kirk
I think there may be a couple of issues.
1) Did you average your 9 B0 images so that there is only one at the beginning of the 4D Nifti?
2) I can't remember 100% if you can use a different number of directions for each b value, but if you can, I think you'll need two separate gradient vector files since you use 30 directions and then 60 directions. (Look at previous posts in this forum - I think this query has been addressed previously).
Also, if you attach your parameters file, 4D Nifti, and bvecs files people can try to debug it for you.
Cheers,
Kirk
Jun 29, 2017 02:06 PM | wafaa sweidan
RE: Index exceed matrix dimensions
Dear Krik,
Thank you so much for your reply,
I am a beginner in using this software, so I will be really grateful if you assist me additionally.
I understand we have to do averaging of B0 (which I did not do before), however, concerning the different number of directions, I created 2 separate txt files each corresponding to a b-value gradient, and then I opened a command line to insert my parameters (the ndir, indexes, bvals from my DKE_paramters file) but nothing is defined in the cmd.exe, is this the right way to run dke from command line?
The problem now is that even with another set of data (1 B0, 20 directions b=1000, and 20 directions B=2000), the program is assuming that I have 40 direction for each b-value (as you can see in the SAM_DKE_parameters), so I do not know how to solve this issue. I am simply inserting the bvec file (which have 20 directions for b1000 and 20 for b2000) into the user-defined gradient vectors.
I attached the parameters, 4d nifti and bvecs for the dki data set (1b0, 20 dir b=1000, and 20 dir b=2000) and data for noddi which I am using to estimate kurtosis (9b0, 30 b=711, and 60 b=2855).
Thank you so much in advance,
Regards,
Wafaa
Thank you so much for your reply,
I am a beginner in using this software, so I will be really grateful if you assist me additionally.
I understand we have to do averaging of B0 (which I did not do before), however, concerning the different number of directions, I created 2 separate txt files each corresponding to a b-value gradient, and then I opened a command line to insert my parameters (the ndir, indexes, bvals from my DKE_paramters file) but nothing is defined in the cmd.exe, is this the right way to run dke from command line?
The problem now is that even with another set of data (1 B0, 20 directions b=1000, and 20 directions B=2000), the program is assuming that I have 40 direction for each b-value (as you can see in the SAM_DKE_parameters), so I do not know how to solve this issue. I am simply inserting the bvec file (which have 20 directions for b1000 and 20 for b2000) into the user-defined gradient vectors.
I attached the parameters, 4d nifti and bvecs for the dki data set (1b0, 20 dir b=1000, and 20 dir b=2000) and data for noddi which I am using to estimate kurtosis (9b0, 30 b=711, and 60 b=2855).
Thank you so much in advance,
Regards,
Wafaa
Jun 29, 2017 07:06 PM | wafaa sweidan
RE: Index exceed matrix dimensions
Hello Krik,
It worked through the terminal!
One more thing I would like to ask you about is that how we view the colored map?
in FSL, vector files are usually generated for DTI maps and we use them to view FA colored map for example.
Someone suggested ImageJ, but still DKE gives only maps and doesnt give the principle vectors. So how can we view the colored maps?
Thanks,
Wafaa
It worked through the terminal!
One more thing I would like to ask you about is that how we view the colored map?
in FSL, vector files are usually generated for DTI maps and we use them to view FA colored map for example.
Someone suggested ImageJ, but still DKE gives only maps and doesnt give the principle vectors. So how can we view the colored maps?
Thanks,
Wafaa
Jun 30, 2017 02:06 AM | Kirk Feindel - University of Western Australia
RE: Index exceed matrix dimensions
Hi Wafaa,
Yes, you get parameter maps from DKE, which can be viewed in ImageJ and you can choose whatever color lookup table you like. In FSL, the vector maps you get are after running BEDPOSTX for the tractography? DKE doesn't do fiber tracking - you need to use their Fiber Tracking module and from that output you can look at orientation distribution functions, etc.
If you want to do quick fiber tracking with your data (not making use of the non-Gaussian information), you can also use DSI Studio.
Cheers,
Kirk
Yes, you get parameter maps from DKE, which can be viewed in ImageJ and you can choose whatever color lookup table you like. In FSL, the vector maps you get are after running BEDPOSTX for the tractography? DKE doesn't do fiber tracking - you need to use their Fiber Tracking module and from that output you can look at orientation distribution functions, etc.
If you want to do quick fiber tracking with your data (not making use of the non-Gaussian information), you can also use DSI Studio.
Cheers,
Kirk