Do manual and voxel-based morphometry measure the same? A proof of concept study.

Niels K Focke,Peter Falkai,Walter Paulus,Oliver Gruber,Sarah Trost

doi:10.3389/fpsyt.2014.00039

Abstract

Voxel-based morphometry (VBM) is a commonly used method to study volumetric variations on a whole brain basis. However, it is often criticized for potential confounds, mainly based on imperfect spatial registration. We therefore aimed to evaluate if VBM and “gold standard” manual volumetry are measuring the same effects with respect to subcortical gray matter volumes. Manual regions-of-interest were drawn in the hippocampus, amygdala, nucleus accumbens, thalamus, putamen, pallidum, and caudate nucleus bilaterally. Resulting volumes were used for a whole brain VBM correlation analysis with Statistical Parametric Mapping (SPM8). The hippocampus, amygdala, putamen, and caudate nucleus were correctly identified by SPM using the contemporary high-dimensional normalization (DARTEL toolbox). This strongly suggests that VBM and manual volumetry both are indeed measuring the same effects with regard to subcortical brain structures.

Highlights

Since its first description in the late 1990s [1, 2] voxel-based morphometry (VBM) has gained much attention in the neuroscience community and has been applied to pathological and physiological conditions alike
There are several reports in the literature that VBM findings could not be replicated by repeat studies or manual validation, e.g., in schizophrenia [4], which can be regarded as lack of robustness of the method
One study has reported a superiority of region of interest (ROI) volumetry in physiological aging with an overestimation of age-related differences in regional brain volumes by VBM [9]; another study found VBM to be specific in detecting local volumetric alterations in expected regions and capable of detecting remote volume loss in Huntington disease patients [10]

Summary

Introduction

Since its first description in the late 1990s [1, 2] voxel-based morphometry (VBM) has gained much attention in the neuroscience community and has been applied to pathological and physiological conditions alike. There are several reports in the literature that VBM findings could not be replicated by repeat studies or manual validation, e.g., in schizophrenia [4], which can be regarded as lack of robustness of the method. Manual region of interest (ROI)based methods are still regarded as gold standard by many authors but these are much more time-consuming, subject to operator biased, and require a priori anatomical constraints. One study has reported a superiority of ROI volumetry in physiological aging with an overestimation of age-related differences in regional brain volumes by VBM [9]; another study found VBM to be specific in detecting local volumetric alterations in expected regions and capable of detecting remote volume loss in Huntington disease patients [10]

Objectives

Methods

Results

Conclusion