Difficulty with tensor2metric

Hi there!
I’m trying to generate maps of FA, RD, and MD. I used dwi2tensor, and then tensor2metric on my dwi_denoised_deringed_preproc_unbiased_normalized_upsampled.mif with the -adc, -rd, and -fa options. Only the -fa option yielded an actual brain image. the -rd and -adc did not resemble a brain at all. Any advice would be greatly appreciated!

This is probably simply a matter of setting the intensity range more appropriately. There’s a few voxels at the edge of your image that give completely unrealistic values, and that sets the intensity range that mrview then uses automatically (it just uses min to max intensity in the image slice by default). Use the view tool to set it to something closer to the expected range – should be something like 0 to 0.001 or thereabouts.

We’ll probably need to to do something smarter about this in future since it is a recurrent issue. We’re thinking about using the 10th to 90th percentile, which should be much more robust. But someone would need to invest time to do this…

Yes, that was it! Thanks a lot, I should’ve been able to figure that one out!