Local Volume Change Maps in Nonrigid Registration: When Are Computed Changes Real?

Measures of brain change can be computed from sequential MRI scans, providing valuable information on disease progression. Tensor-based morphometry (TBM) creates maps of these brain changes, visualizing the 3D profile and rates of tissue growth or atrophy. In this paper, we examine the reproducibility and stability of different techniques in TBM. In particular, we compare matching functionals (sum of squared differences and mutual information), and registration schemes (unbiased large-deformation registration and viscous fluid registration) using serial MRI scans of nine normal elderly subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Our results show that the unbiased large-deformation method has higher reproducibility. When coupled with unbiased registration, sum of squared differences outperforms mutual information. In contrast, when coupled with fluid registration, mutual information outperforms sum of squared difference. Moreover, the regions with least stability, due to both spatial distortion and intensity inhomogeneity, are the brain stem, thalamus, and ventricles.