Add statistics

[jcohenadad]

After each plot it would be important to compute the following distribution statistics, and display as a table:

• row: stats mean +/- SD; column: qMRI metric
• stats:
  • Intra-subject COV: Compute COV per subject and then display average and SD across subjects.
  • Inter-subject COV: Compute mean per subject and then compute inter-subject COV (based on the mean per subject)

[mathieuboudreau]

Statistics tables were added to the Jupyter Book in their respective sections. By the looks of it, the CoVs for quantitative measures are much worse in brain than spine; I'm wondering if this could be due to bad segmentations more than biological/mri variability.

[jcohenadad]

I assume this is not in percent?

could you pls put it in percent (fairly common) and make it explicit in the column: "intrasubject CoV mean [%]" thx

[jcohenadad]

By the looks of it, the CoVs for quantitative measures are much worse in brain than spine

spine:

brain:

i'm not sure i agree: COV for MTR and T1 is lower in the brain, and similar for MTsat. No?

[mathieuboudreau]

i'm not sure i agree: COV for MTR and T1 is lower in the brain, and similar for MTsat. No?

Sorry, I wasn't clear enough with what I meant. I meant the COVs within the brain (MTR, MTsat, and T1) vary much more than the COVs within the spine. eg in the brain, intrasubject MTR COV is 0.008 and MTsat COV is 0.064 (8x more), whereas in the spine MTR COV is 0.051 and MTsat is 0.079 (only ~1.5 more)

[mathieuboudreau]

If you look at the figures, this intrasubject variabilities in the WM for non-MTR metrics become much more apparent.

MTR

MTsat

(also I just noticed I forgot to do a table for brain GM, I'll add that too).

[mathieuboudreau]

Also MTR vs MTsat, note that MTR measurements are a subset of MTsat, so there's something really questionable with the much larger variability for MTsat...

[mathieuboudreau]

I guess a few more acquisitions are included in MTsat actually, because B1 is added too in MTsat

[jcohenadad]

i see, yes, i agree. It would be worthwhile:

• checking the brain masks used to compute MTR
• making sure the MTR acq params were correct for subj4 (looks suspiciously different)

[mathieuboudreau]

✅ Done converting COVs to [%]

Brain

Spinal cord