A Comparison of Freesurfer and Automatic Segmentation of Hippocampal Subfields for Estimating Hippocampal Volumes among Preschoolers

Poster Session C - Sunday, April 14, 2024, 5:00 – 7:00 pm EDT, Sheraton Hall ABC

Zehua Cui¹ (zcui12@umd.edu), Jade Dunstan¹, Tracy Riggins¹; ¹University of Maryland

The hippocampus is a complex structure comprised of multiple internal circuits (i.e., subfields), including the Cornu Ammonis (CA) fields 1-4, dentate gyrus (DG), and subiculum that subserves memory across the lifespan (Amaral & Lavenex, 2007). Hippocampal subfields are thought to undergo extended postnatal development (Lavenex & Lavenex, 2013), however studies with human children are limited, partially due to methodological limitations. Our study aims to compare hippocampal subfield volumes derived from two automated software packages, Freesurfer and Automatic Segmentation of Hippocampal Subfields (ASHS), among a sample of 3- to 5-year-old children. A total of 33 children (N = 33, Mage = 4.26 ± 0.61 years, 60.6% female) provided a whole-brain T1-weighted .9mm isotropic scan for Freesurfer 7.1.1 (Fischl, 2012) and a T2-weighted scan (.4mmx.4mmx2mm) of the medial temporal lobe for ASHS processing (Yushkevich et al, 2014). Volumes derived from the two methods were robustly correlated across each bilateral subregion and subfield (subiculum, CA1, and CA2-4/DG), average r(31) = 0.58 (rleft = 0.68, rright = 0.47). Paired sample t-tests showed significantly smaller Freesurfer segmentation volumes in all subfields except bilateral CA1-head ( = 151.11mm3, p < .001) and right subiculum-body ( = -8.28mm3, p = .18). Our findings revealed correlations and differences in hippocampal volumes derived from Freesurfer and ASHS. Our future steps include 1) comparing both methods to manual tracing, 2) examining potential divergences in group comparisons based on children’s sex, socioeconomic and nap status, and 3) investigating differences in estimates of developmental changes in subfields between the two methods.

Topic Area: METHODS: Neuroimaging