Hi Ehsan,

I took a look at your data and I found the image volume to be in the RAS orientation, the gradient table to be in LPS, and the voxel dimensions to be 1 x 1 x 2 mm^3. I was a little concerned that the voxel dimensions were anisotropic, but the tractography portion of the script within DKE works fine (attached), though the data is noisy, which is likely due to the high resolution of your scan. For our experiments, we typically don't go below about 2 x 2 x 2 mm^3 for human imaging at 3T since were are concerned about SNR. I would double check SNR.

I also think this could be causing the error you experienced. I used the 'low' resolution option to try to visualize the ODFs but the output dki.fib file was almost 3GB! It was able to save, but this is a very large file, and I suspect a local memory issue was responsible for your error writing the outputs. I would suggest setting the save_odfs option to 0. In the future we will hopefully be able to reduce this file size further. I can also confirm the very long processing times for your data.

I tried running fiber tracking in DSI Studio as well and the results looked kind of funny to me. I did not look into this further, but I would check with Frank or DSI Studio documentation to make sure it will work with the 1 x 1 x 2 mm^3 voxel dimensions.

Best,

Russell