Dwidenoise: noise map output

[dchristiaens]

Hi Samuel,

@samuelstjean

Anyway, as mentioned previously, one needs to fix the bias afterwards, but doing so requires a calculation based on SNR.

Yes indeed, Rician bias correction is not yet implemented in MRtrix. It’s on my todo list, but that list is only getting longer… If anyone wants to give it a crack (?), I’m happy to support.

@samuelstjean

However, for multishell data, that won’t really work anymore. The original mrm paper mention correcting directly the signal from the kept eigen value, however that does not really apply here as one only gets a 3D map out of it. So, how do people go about correcting the maps usually in case of multishell data?

I’m not entirely sure I understand your question, but I think the confusion may relate to this discussion. dwidenoise estimates the noise level, which is constant for all b-values and gradient directions. SNR is only affected in different shells because the signal attenuates, not the noise level.