Hi all,

I have a number of DKI NIFTI images (.nii extension) that I will need to preprocess before using them in DKE. For pre-processing, I plan on using FSL functions eddy and topup to remove distortions, and then perform skull stripping on the dataset.

One issue I have with my dataset is that I have numerous NIFTI volumes for b = 0 data. As such, I am having trouble determining a way to average all of these NIFTI images to get a single b = 0 NIFTI volume, of which I can use in the DKE software. I have heard about co-registration with SPM, but am confused by the procedure to get these images. Any clarification on the methodology and implementation of this procedure (or another averaging procedure) would be greatly appreciated.

Another question I have is if there should be any additional pre-processing I should do prior to submitting my images to DKE (aside from that aforementioned above). The MRI scanner used was of higher resolution (4.7 T), so I am not sure how effective or useful an additional denoising procedure in my pre-processing pipeline will be, as the images are fairly clean as they are currently.

Thanks in advance!

Hi Michael, 

Since you already seem familiar with fsl look into using FLIRT for coregistration. I am not positive but eddy might actually have motion correction build in. 
Once your images are coregistered a simple averaging procedure using either matlab or FSL (FSL maths - Tmean) is sufficient before running DKE.  

Other preprocessing steps you can look into is denoising and correction for gibbs artefacts using mrtrix. If you follow the preprocessing suggested here (http://mrtrix.readthedocs.io/en/0.3.16/w...) you will be in good shape! 

Hope this helps, 
Emilie

Hi Emily,

Ok great, thank you for your help! I will definitely check out those resources.

Best,
Michael