Pediatric neuroimaging is a quickly developing field that still faces important methodological challenges. One key challenge is the use of many different atlases, automated segmentation tools, manual edits in semiautomated protocols, and quality control protocols, which complicates comparisons between studies. In this article, we present our semiautomated segmentation protocol using FreeSurfer v6.0, ENIGMA consortium software, and the quality control protocol that was used in FinnBrain Birth Cohort Study. We used a dichotomous quality rating scale for inclusion and exclusion of images, and then explored the quality on a region of interest level to exclude all regions with major segmentation errors. The effects of manual edits on cortical thickness values were minor: less than 2% in all regions. Supplementary materials cover registration and additional edit options in FreeSurfer and comparison to the computational anatomy toolbox (CAT12). Overall, we conclude that despite minor imperfections FreeSurfer can be reliably used to segment cortical metrics from T1-weighted images of 5-year-old children with appropriate quality assessment in place. However, custom templates may be needed to optimize the results for the subcortical areas. Our semiautomated segmentation protocol provides high quality pediatric neuroimaging data and could help investigators working with similar data sets.

Developing accurate subcortical volumetric quantification tools is a crucial issue for neurodevelopmental studies, as they could reduce the need for challenging and time-consuming manual segmentation. In this study the accuracy of two automated segmentation tools, FSL-FIRST (with three different boundary correction settings) and FreeSurfer were compared against manual segmentation of subcortical nuclei, including the hippocampus, amygdala, thalamus, putamen, globus pallidus, caudate and nucleus accumbens, using volumetric and correlation analyses in 80 5-year-olds. Both FSL-FIRST and FreeSurfer overestimated the volume on all structures except the caudate, and the accuracy depended on the structure. Small structures such as the amygdala and nucleus accumbens, which are visually difficult to distinguish, produced considerable overestimations and weaker correlations with all automated methods. Larger and more readily distinguishable structures such as the caudate and putamen produced notably lower overestimations and stronger correlations. Overall, the segmentations performed by FSL-FIRST's Default pipeline were the most accurate, while FreeSurfer's results were weaker across the structures. In line with prior studies, the accuracy of automated segmentation tools was imperfect with respect to manually defined structures. However, apart from amygdala and nucleus accumbens, FSL-FIRST's agreement could be considered satisfactory (Pearson correlation > 0.74, Intraclass correlation coefficient (ICC) > 0.68 and Dice Score coefficient (DSC) > 0.87) with highest values for the striatal structures (putamen, globus pallidus and caudate) (Pearson correlation > 0.77, ICC > 0.87 and DSC > 0.88, respectively). Overall, automated segmentation tools do not always provide satisfactory results, and careful visual inspection of the automated segmentations is strongly advised.