Inter-Software Consistency on the Estimation of Subcortical Structure Volume in Different Age Groups

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Imaging Analysis

Background:

• Limited research exists on the consistency of subcortical volume estimates across different neuroimaging software packages.
• Understanding inter-software consistency is crucial for the reproducibility and interpretation of neuroimaging findings.
• The impact of age on the consistency of these estimates remains largely unexplored.

Purpose of the Study:

• To evaluate the consistency of subcortical volume estimates generated by three popular automated software packages: FreeSurfer, CAT, and FSL.
• To investigate whether age influences the consistency of these subcortical volume estimates.
• To assess the validity of each software package by examining its ability to predict age from brain structure volumes.

Main Methods:

• Analysis of T1-weighted structural MRI data from two public datasets, encompassing four age groups from childhood to late adulthood.
• Estimation of volumes for seven subcortical structures (thalamus, caudate, putamen, pallidum, hippocampus, amygdala, accumbens) using FreeSurfer, CAT, and FSL.
• Quantification of inter-software consistency using intraclass correlation coefficient (ICC) and Pearson correlation coefficient (PCC), with comparisons across age groups.

Main Results:

• Good inter-software consistency was observed for the thalamus, caudate, putamen, and hippocampus.
• Moderate consistency was found for the pallidum, while poor consistency was noted for the amygdala and accumbens.
• No significant age-related differences in inter-software consistency were detected for most structures; FreeSurfer demonstrated superior age prediction accuracy compared to CAT and FSL.

Conclusions:

• Inter-software consistency in subcortical volume estimation varies significantly by brain structure but is generally independent of age.
• These findings highlight the importance of considering software choice and anatomical region when interpreting neuroimaging results.
• The study underscores the need for standardized protocols and further research to enhance the reliability and reproducibility of subcortical volume analyses in neuroscience.