Hi Alexander - When you average over a larger region, you are potentially reducing noise but you may also potentially wash out an effect that is very localized by including more voxels where the effect is not present. That's the difference between the two measures, so they are at different points along the false positive/false negative trade-off. One other option would be to examine if your effect is localized by looking at the FA along the tract (from the pathstats.byvoxel.txt files) instead of averaging over the whole tract.
Hope this helps, a.y On Mon, 9 Feb 2015, Alexander Tomyshev wrote:
Dear Anastasia and Freesurfer team, I used TRACULA to get FA values for two groups: patients and controls. After that, I ran between group comparison and got circa >0.05 p-values for FA _Avg_Weight values BUT circa 0.001-0.005 p-values for FA_Avg_Center for the same tract. Then I ran correlation analysis and found out that some clinical scores of patients correlate with FA_Avg_Weight values (p-value of c. 0.02) and with FA _Avg_Center values (but with p-values of c. 0.002). So the correlation with FA _Avg_Center is more significant than correlation with FA_Avg_Weight. Once again, when I use FA_Avg_Weight in between-group comparison I got non-significant results (slightly >0.05 p-value), but when I use FA_Avg_Center in such comparison I got significant group difference (c. 0.001-0.005 p-values). My question is – can I use these results and say that there is a statistically significant difference in FA values in some tracts? Of course, I will mention that these difference is significant when we compare average FA values of the highest-probability path only and does not reach statistical significance if we compare average values over the entire tract weighted by the value of the probability distribution at every tract’s voxel. As I understand, theoretically and logically, comparison of average FA values of the highest-probability path only (instead of comparing FA_Avg_Weight values) has more statistical power to detect more subtle differences between groups. And in my case comparison of FA_Avg_Weight values just has not enough statistical power to detect such subtle difference. Am I right? I will very appreciate and will be happy to hear any of your thoughts concerning written above and especially any thoughts on how to interpret such difference between results using FA _Avg_Weight and FA_Avg_Center values. Thank you in advance. Kind regards, Alexander Tomyshev Laboratory of Neuroimaging and Multimodal Analysis, Mental Health Research Center of the Russian Academy of Medical Sciences, 34 Kashirskoe shosse, 115522 Moscow, Russia Email.: alexander.tomys...@gmail.com
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
