Hello!

I had a couple of questions about how mri_cnr computes the contrast to noise ratio.

From previous posts, a command such as mri_cnr $FREESURFER_HOME/subjects/bert/surf $FREESURFER_HOME/subjects/bert/mri/norm.mgz

gives the lh cnr, rh cnr, and total cnr for bert.

1) What exactly is happening when you reference the /surf directory here?

2) Are the CNR values for the norm.mgz volume being computed by using the lh white + lh pial; and rh white + rh pial surfaces?

3) Are the subcortical gm structures somehow involved in this calculation as well?

I'm asking question 3 because when you add a -label $FREESURFER_HOME/subjects/bert /label/lh.cortex.label $FREESURFER_HOME/subjects/bert/label/rh.cortex.label the lh cnr, rh cnr, and total cnr values that you get are now different.

The outputs for bert for each case is as follows (FreeSurfer v6 bert was used here):

1) With /surf alone

white = 96.6+-6.0, gray = 75.7+-13.5, csf = 55.3+-16.6 gray/white CNR = 1.983, gray/csf CNR = 0.907 lh CNR = 1.445

white = 96.4+-6.0, gray = 76.0+-13.3, csf = 55.6+-16.2 gray/white CNR = 1.952, gray/csf CNR = 0.943 rh CNR = 1.448

total CNR = 1.446

2) With /surf and -label

white = 96.6+-6.0, gray = 73.1+-18.9, csf = 50.8+-20.5 gray/white CNR = 1.389, gray/csf CNR = 0.644 lh CNR = 1.016

white = 96.4+-6.0, gray = 73.4+-18.8, csf = 51.0+-20.3 gray/white CNR = 1.353, gray/csf CNR = 0.652 rh CNR = 1.002

total CNR = 1.009

Any inputs or thoughts on this matter are greatly appreciated!

Thanks,

Prad