High correlations between MRI brain volume measurements based on NeuroQuant^® and FreeSurfer

Area of Science:

• Neuroimaging analysis
• Medical image processing
• Brain volume quantification

Background:

• NeuroQuant (NQ) and FreeSurfer (FS) are standard automated tools for MRI brain volume measurement.
• Previous studies noted high reliability but significant effect size discrepancies between NQ and FS.
• Large effect size differences can complicate comparative analyses and interpretations.

Purpose of the Study:

• To investigate the efficacy of linear transformations in reducing effect size differences between NQ and FS brain volume measurements.
• To develop and compare methods for transforming FreeSurfer volumes to be more comparable with NeuroQuant volumes.
• To enhance the reliability and comparability of automated brain volumetry data.

Main Methods:

• Utilized brain volume data from 60 subjects, including healthy controls and patients with neurological conditions (TBI, Alzheimer's disease).
• Applied two statistical approaches: traditional linear regression and Bayesian linear regression, to derive FS-to-NQ volume transformations.
• Performed regression analyses to establish linear transformation models for harmonizing FS and NQ measurements.

Main Results:

• Linear transformations using traditional linear regression reduced effect sizes to small or moderate levels.
• Bayesian linear regression-based transformations resulted in trivially small effect sizes, indicating high concordance.
• This study presents the first method for transforming FS to NQ data, achieving both high reliability and minimal effect size differences.

Conclusions:

• Linear transformations, particularly those employing Bayesian regression, are effective in reconciling discrepancies between NeuroQuant and FreeSurfer brain volume measurements.
• Bayesian regression offers a superior approach for harmonizing neuroimaging data, potentially improving the accuracy and consistency of brain volume analysis.
• The developed transformation method facilitates more reliable comparisons across studies and patient cohorts utilizing different automated volumetry software.