Hello Gabor,
a few quick comments:
i) Although we typically aim for maximizing both, accuracy and reproducibility are different things. In a longitudinal study you are often particularly interested in the reproducibility of the findings, in the sense of their repeatability across different time points if there are no significant biological changes in your sample. T
ii) Ideally you would have included in your experimental design a group of controls that you expect to not show longitudinal effects, and used them to measure and estimate the test-retest reproducibility error of your experimental setup (i.e., acquisition + analyses pipeline). This information can then be used to estimate a lower limit of the effect size you may be able to measure with your setup.
iii) We have looked at reproducibility errors of the longitudinal Freesurfer cortex in healthy elderly volunteers from standard 3T MPRAGE and found that it was generally above 2% in the structures we looked at (https://www.ncbi.nlm.nih.gov/pubmed/23668971, Table 7). We did not look at the cerebellum, but I imagine its reproducibility will not be higher. However, if your acquisition protocol and subject population are very different then these results might not apply well as reference for you.
Cheers, jorge
On 13/12/2016 11:41, Gabor Perlaki wrote:
I've found the article "Within-subject template estimation for unbiased longitudinal image analysis" by Reuter et al. It only examines a limited number of structures for the reproducibility of longitudinal Freesurfer. Are there any other paper that examines the cerebellum as well? Any suggestion from the authors of Freesurfer about accuracy of longitudinal Freesurfer for the cerebellum in healthy subjects?
Best, Gabor
2016-12-06 11:44 GMT+01:00 Gabor Perlaki <petzinger.gabor@gmail.com mailto:petzinger.gabor@gmail.com>:
Dear all, We've run longitudinal Freesurfer on 30 healthy subjects. We have two subgroups (n=15) and we found a significant longitudinal change in the left and right cerebellar cortex in one of our subgroups. However, this change is very small: mean=0.67% range: -1.61-2.3% for the right cerebellar cortex; mean=0.86% range: -1.64-3.7% for the left cerebellar cortex. Although statistics indicate significant cerebellar cortex increase, we are sceptical that Freesurfer's accuracy allows reliable detection of such small differences. Is there any article on how accurate the longitudinal Freesurfer for cerebellum segmentation or any suggestion on how to decide whether our results are reliable? Best, Gabor-- Gabor Perlaki research associate Diagnostic Center of Pécs H-7623 Pécs, Rét str. 2. Tel.: 0036-30-2084367 E-mail: petzinger.gabor@gmail.com mailto:petzinger.gabor@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.