It is hard to say. While a p-value changing from .02 to .2 seems like a lot, it does not really take much change in the values. I would look at the CVs for the individual subjects both in cross and long and see which ones are causing the change. My experience is that long is much better than cross for longitudinal data.
On 09/19/2018 01:18 PM, Martin Juneja wrote:
External Email - Use Caution
Hi,
I am using freesurfer-Linux-centos6_x86_64-stable-pub-v6.0.0-2beb96c version of FreeSurfer.
I used cross-sectional pipeline to identify ROIs (from Desikan atlas) with significant difference in cortical volume (CV) between controls and *patients (subjects 1-40, baseline)*, say I get ROIs: R1, R2 and R3.
I am using these ROIs (R1, R2 and R3) for further analysis to determine the effect of two treatments: *T1 (placebo) (for subjects 1 to 20) and T2 (active) (for subjects 21 to 40)*, compared to their corresponding baseline.
I compared CV of these ROIs (R1, R2 and R3): *T1 vs. baseline and T2 vs. baseline* using both cross-sectional as well as longitudinal pipeline.
*Using cross-sectional pipeline:* I found there is significant improvement in CV (p = 0.019) for these ROIs for treatment T2 compared to baseline and no change for placebo treatment T1. *Using longitudinal pipeline: *There is little improvement in CV (p = 0.2) for these ROIsfor treatment T2 compared to baseline and no change for placebo treatment T1.
I was wondering why I am getting results so different between these pipelines, and which pipeline suits better in such comparison when my ROIs are identified using cross-sectional pipeline, but further analysis of these ROIs involved longitudinal comparison.
I would really appreciate any help (as early as possible).
Thank you so much !
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer