Dwi2mask output is suboptimal

1

Dear experts,

I used dwi2mask to generate the binary brain mask. A substantial portion of the brain was missing:

image
image1398×1292 140 KB

Then I used dwibiascorrect (ant) to correct the bias field, hoping to improve the brain mask’s quality. The mask generated using the bias-field-corrected image was much better, but still a considerable portion of the brain was excluded:

image
image1380×1284 127 KB

What else can I do to improve the brain mask’s quality?

To my knowledge dwi2mask of newer MRtrix versions allows the selection of algorithm. To my understanding the dwi2mask of the dev version has this option, but I couldn’t install it as I don’t have QT on my Mac.

Thank you so much in advance!!

2

In my experience you have to iterate dwi2mask and dwibiascorrection a few times. Because a good mask relies on the absence of a bias field and a good bias correction relies on a good mask.

2 Likes
3

Thanks a lot for your suggestion. I’ll give it a try now :slight_smile: