Validity of automated FreeSurfer segmentation compared to manual tracing in detecting prenatal alcohol exposure-related subcortical and corpus callosal alterations in 9- to 11-year-old children

Abstract

Highlights Corpus callosum and caudate nuclei show excellent reliability across methods. Consistent PAE-related volume changes across methods if ICC consistency >0.70. FreeSurfer performs similarly on the smaller brains of PAE children as in controls. FreeSurfer segmentation can be used in easy-to-segment regions in pre-adolescence.

High‐resolution diffusion tensor imaging identifies hippocampal volume loss without diffusion changes in individuals with prenatal alcohol exposure

“…This study demonstrates whole-hippocampus volume reductions of ~18% in the PAE group relative to controls, as measured from manual segmentation on DWIs. Volume reductions in the hippocampus have likewise been found in many previous human MRI studies, which report whole structure volume reductions of 5% to 25% in children, adolescents, and young adults with PAE (Astley et al, 2009;Biffen et al, 2018Biffen et al, , 2020Coles et al, 2011;Dodge et al, 2020;Dudek et al, 2014;Gross et al, 2018;Krueger et al, 2020;McLachlan et al, 2020;Nardelli et al, 2011;Roussotte et al, 2012;Treit et al, 2013Treit et al, , 2017Uban et al, 2020;Willoughby et al, 2008). However, the majority of these studies (11/15;Biffen et al, 2020;Coles et al, 2011;Dodge et al, 2020;Gross et al, 2018;Krueger et al, 2020;McLachlan et al, 2020;Nardelli et al, 2011;Roussotte et al, 2012;Treit et al, 2013Treit et al, , 2017Uban et al, 2020) measure hippocampal volume from automated segmentation on anatomical MRI (T1 or T2), whereas our study found no significant group differences with automated segmentation on T1-weighted images.…”

“…However, this study and one other (which used automated segmentation) found that participants with pFAS/FAS had smaller hippocampi than those who were heavily exposed but did not receive a diagnosis of pFAS/FAS (Biffen et al, 2020;Dodge et al, 2020). As such, the lack of group differences in automated segmentation volumes may reflect that our PAE group was less severely affected than previous samples, given none here had a diagnosis of pFAS or FAS, whereas just over half of the participants with PAE (27/52) had a diagnosis of pFAS/FAS in Biffen et al (2020). Indeed, automated T1-based segmentation did not yield significant group differences in the whole brain, gray or white matter volumes between groups here, again suggesting that our sample is less impaired than previously published samples, while also suggesting that manual tracing methods may be more sensitive to less severely affected PAE populations.…”

Charting the human amygdala development across childhood and adolescence: Manual and automatic segmentation

“…After controlling the quality of scan, the segmentation accuracy increased with amygdala volume, and then remained stable when the amygdala has reached a large enough size (around 1.3–1.4 ml). Previous studies have found that smaller brain structures were associated with greater automatic segmentation errors due to their sizes and shapes differing from adults ( Schoemaker et al, 2016 , Biffen et al, 2020 , Sánchez-Benavides et al, 2010 ). Our results are consistent with that neuro-anatomical and geometric features could systematically influence the accuracy of their automatic segmentation.…”

The Paternal Transition Entails Neuroanatomic Adaptations that are Associated with the Father’s Brain Response to his Infant Cues

“…These steps resulted in 3D triangular meshes for the white and pial surfaces (inner and outer surfaces of the cortex, respectively) and vertex-wise maps of CV, CT, and CA. We restricted our analysis to the cortical mantle as the estimates for subcortical regions typically involved in maternal behavior, such as the ventral striatum, are not very reliable with automatic methods such as FreeSurfer ( Biffen et al. 2020 ).…”