In 2014, we published a best-practices/evaluation paper concerning the ANTs cortical thickness pipeline. The authors of the paper in question disregard some of the important details in that paper and basically created a straw man ANTs pipeline for comparison that we would not recommend. For example, they used their own in-house template and priors for segmentation which were not constructed in the manner recommended in our 2014 paper. Their reasoning was “because the ANTs package does not include a stock template or segmentation priors, we used our in-house population-specific template/priors” which is odd because we provided a link (see Table 4) to the templates (and priors) used. In addition, as far as we know, their template isn’t available or, at least, wasn’t available at the time of publication so even following their recommendations would be problematic.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear ANTs community,
I have recently read this paper : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5187496/ and I was wondering if anyone could make a comment on the results presented.
All the best,
Sandro
In 2014, we published a best-practices/evaluation paper concerning the ANTs cortical thickness pipeline. The authors of the paper in question disregard some of the important details in that paper and basically created a straw man ANTs pipeline for comparison that we would not recommend. For example, they used their own in-house template and priors for segmentation which were not constructed in the manner recommended in our 2014 paper. Their reasoning was “because the ANTs package does not include a stock template or segmentation priors, we used our in-house population-specific template/priors” which is odd because we provided a link (see Table 4) to the templates (and priors) used. In addition, as far as we know, their template isn’t available or, at least, wasn’t available at the time of publication so even following their recommendations would be problematic.
Thank you Nick for your reply and the link.
Indeed, their results seem to differ quite a bit from your comparison with freesurfer.