imaging data pulled: 2019-12-04
clinical data pulled: 2019-12-04
code written: 2019-11-28
last ran: 2019-12-04
website: http://rpubs.com/navona/NM_LC_derivateScore
github: https://github.com/navonacalarco/CurAge/blob/master/scripts/analysis/06_NM_ROI_SN_LC_derivateScore.Rmd


This analysis stems from the Sibille lab meeting presentation on Monday November 18th, at which Drs. Diniz and Banasr suggested that we account for the effect of volume when looking at CNR.


Correlations

First, we should see if there are correlations between volume and CNR in individual participants. The intuition is that, it might make sense for higher CNR to exist in LLD in the absence of higher volume, but higher CNR combined with higher volume in LLD is harder to interpret, as we don’t expect higher volumes in LLD.

Raw data

SN.

LC.

Correlation matrix

The values represent the Pearson correlation coefficient value, the circles are a visualization of the same. We see that, in LLD, there is a strong correlations between LC volume and CNR (r=.57), and a moderate correlation between SN volume and CNR (r=.41). In HCs, the correlations between LC volume and CNR (r=.66), and SN volume and CNR (r=.84) are higher than in LLD, and considered strong.


Derivate score

Because we see moderate/strong correlations between volume and CNR, in both structures, in both groups, it may make sense to control for volume in the CNR calculation, i.e., make a “derivate score”. The “derivate score” is calculated simply as CNR over volume, for all participants.

LLD, n=16 (52%) HC, n=15 (48%) p
SN derivate 0.02 (0.01) 0.03 (0.03) .277
LC derivate 0.37 (0.13) 0.80 (1.51) .263