Dwidenoise multi-band, multi-shell noise.mif output

Hi All -

I’ve read another thread, that was a bit older, but I wasn’t able to glean the information I was hoping to acquire. I was hoping someone may help me resolve my concern.
I am using the BATMAN tutorial for our MRtrix DWI preprocessing/analyses. Our acquisition is: Hyperband sequence TR 4s, Minimum TE, 140x140, multiple b-values (0, 500, 1000, 2000, 3000), 102 directions.

For the dewidenoise script, we run

dwidenoise raw_dwi.mif dwif_den.mif -noise noise.mif

When reviewing the noise.mif output, there are some visible outlines of what looks like a brain (example attached). I am curious whether this level of non-randomness is normal/abnormal for this acquisition?

Any thoughts would be appreciated!

1 Like

Hi Michael.
Your output looks fine. The noise is symmetrical and there is almost always more noise in the middle of the scan.
I assume by “hyperband” that you are using a GE scanner? What was your hyperband factor (I am assuming a factor of 3 based on your TR of 4)? Which head coil are you using?

Jerome, thank you very much for your reply. Yes, you are spot on! We are using a GE 750, with a multi-channel standard adult head coil. From my conversations with some folks, it seems with GE we’re getting a little bit more artifacts, but helpful to know that this type of noise is standard.

Hi Michael. When you say standard adult head coil, is that a GE coil? I always use a Nova 32 channel with MR750 and MR750w scanners and the artifacts are minimal. Using DV25 (or newer) also helps.