Signal to noise ratio for a 3D image

Hello @jcohenadad!

I have been trying to use the sct_compoute_snr command to get the spatial Signal to noise ratio on the motion corrected image averaged across time (it is a 3D image, so I intend to use -method single) and I don’t quite understand what the noise make (-m-noise) stands for.

Can you please help me out with that?
Thanks in advance!


-m-noise points to a NIfTI file that defines a binary mask over a region where the noise standard deviation is to be computed, usually it is defines in the ‘air’ (ie: background). Please note however that the “single” approach assumes that the noise is i.i.d, and follows either a Gaussian or Rayleigh (in which case use flag -rayleigh) distribution. When averaging across multiple scans, the noise distribution across pixels in a noise mask is not i.i.d anymore.

I recommend you read the following article for more details: Dietrich et al., Measurement of signal-to-noise ratios in MR images: Influence of multichannel coils, parallel imaging, and reconstruction filters. J Magn Reson Imaging 2007; 26(2): 375-385.


Thank you very much, @jcohenadad!