Voxel-based morphometry and automated lobar volumetry: the trade-off between spatial scale and statistical correction

Area of Science:

• Neuroimaging
• Brain Anatomy
• Quantitative MRI

Background:

• Voxel-based morphometry (VBM) and automated lobar region of interest (ROI) volumetry are used for brain anatomy analysis.
• Discrepancies exist between VBM and ROI volumetry in detecting gray matter differences.

Purpose of the Study:

• To compare the sensitivity of VBM and automated lobar ROI volumetry in detecting gray matter differences.
• To investigate the influence of spatial scale and multiple comparison corrections on VBM and ROI volumetry.

Main Methods:

• Assessed VBM and automated lobar ROI volumetry on simulated gray matter deficits.
• Evaluated both methods on a dataset of 22q11-deletion syndrome patients with schizophrenia (22q11DS-SZ) and controls.
• Performed theoretical calculations of detection sensitivities for both techniques.

Main Results:

• VBM demonstrated higher sensitivity to simulated focal gray matter deficits than automated ROI volumetry.
• VBM detected global cortical deficits equally well as automated ROI volumetry.
• Automated ROI volumetry showed reduced sensitivity with increasing ROI size, more so than VBM.

Conclusions:

• VBM is more sensitive to focal gray matter differences compared to automated lobar ROI volumetry.
• Averaging over larger volumes in ROI methods impairs sensitivity to focal differences more than multiple comparison corrections in VBM.
• VBM is recommended over automated lobar-scale ROI volumetry for group-based gray matter volume analysis.