Dear John,
For subcortical volumetric, the tools are pretty good (see eg the Medical Image Analysis and PNAS papers on SynthSeg and SynthSeg+, both by Billot et al, 2023). For cortical: if you take a look at the 2 recent papers by Gopinath et al in Medical Image Analysis (recon-all-clinical, 2025) and Human Brain Mapping (accepted this year), you’ll see that the accuracy of the cortical measurements decreases more than that of the subcortical volumes, as one would expect, but still recover general trends. We have a new tool coming soon that improves the accuracy compared with recon-all-clinical and ReconAny; stay tuned!
Best wishes,
/Eugenio
-- Juan Eugenio Iglesias http://www.jeiglesias.com From: John Anderson jb19790001@gmail.com Date: Monday, August 10, 2026 at 7:06 PM To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Fwd: Question Regarding Synth Tools for Quantitative Atrophy Analysis
External Email - Use Caution
Hello,
Could an expert in the Synth collection of tools please confirm whether these tools are intended for reliable use in patient populations, or primarily in healthy brains?
Specifically, we would like to understand whether the Synth tools can preserve and reproduce cortical atrophy measurements in a manner comparable to the full recon-all pipeline. Practically, it appears that recon-all and the Synth tools may produce different measurements for the same brains, so we would like to clarify the expected use case.
For example, if we have a pooled MRI dataset from patients with different neurological pathologies, including Alzheimer’s disease, amyotrophic lateral sclerosis, and other disorders, each condition may have a distinct cortical atrophy pattern. Are the Synth tools able to perform segmentation while preserving these disease-specific pathological patterns, or should they not be used for quantitative atrophy analysis?
Any advice would be appreciated.
Thank you.
John