LAUSR

Editorial for “Reliability of Changes in Brain Volume Determined by Longitudinal Voxel-Based Morphometry” Voxel-based morphometry (VBM) is a powerful technique that identifies neuroanatomical differences between groups by mapping brain images into a common space. VBM is automated, easy to use, and assesses the entire brain. Hence, it gained wide popularity in the study of many neurological and psychiatric disorders. VBM requires accurate segmentation of the brain tissue and nonlinear registration to a common space. Both operations are challenged by the natural variability in brain morphology, variations in imaging hardware and scan parameters, and differences between the analysis pipelines. These factors may reduce the reliability of the VBM results. Multiple cross-sectional studies have assessed the reliability of VBM. However, studying the longitudinal reliability of VBM has been complicated by the natural age-related changes in brain tissue which would be inaccurately interpreted as lower reliability. In this issue of JMRI, Takao et al address the longitudinal reliability of VBM by analyzing back-to-back scan data performed at baseline and at 2-year follow-up. The data were acquired in the Alzheimer’s disease neuroimaging initiative (ADNI) study, and included a total of 68 subjects in three groups of healthy elderly subjects, patients with mild cognitive impairment (MCI), and patients with Alzheimer’s disease (AD). Applying longitudinal VBM analysis of pairs of images, the authors found good-to-excellent reliability in 92% of the voxels in all subjects. Importantly, reliability was lower in MCI and AD patients (83% compared to 99% in healthy subjects). These results indicate that VBM is generally a reliable tool for longitudinal analysis of brain morphometry, but is differentially affected by the characteristics of the study populations. The authors explain the lower reliability in the patient groups by the lower image quality, attributed to a possible higher degree of head motion. If confirmed, better motion management and/or motion correction techniques will be crucial for reliable VBM studies. A limitation of this study is the relatively small sample and that site differences were not evaluated. Larger studies are needed to account for the multiple sources of variability, and to determine the robustness of VBM more accurately in various patient populations. These findings from Takao et al shed light on the possible group-specific effects resulting from head motion that may compromise the reliability and interpretation of the VBM results. This may in turn help to optimize data acquisition and analysis procedures to account for these effects and improve the reliability of VBM.