Objective: Fractional anisotropy (FA) is a powerful measure to study the integrity of the cerebral white matter in vivo. However, because clinical FA assessments are frequently based on single slice evaluations, intra- and interindividual comparisons are highly dependent on image alignment. We attempted to develop an observer-independent, fully automated technique for quantitative FA assessment.
Materials and methods: We employed whole brain diffusion tensor imaging at 3 T with an echo planar imaging sequence (isotropic spatial resolution 1.8 mm) on 4 patients (2x Alzheimer disease, 1x microangiopathy, 1x paraneoplastic disease) and 2 normal control groups (group "young," age 19-32 years; group "old," age 59-69 years). The images were spatially normalized to the standard brain template of the Montreal Neurologic Institute. We introduced a fractional anisotropy index (FAI) as a single measure for the mean tissue anisotropy in certain brain regions of interest. The regions of interest were defined by masks in relation to the Montreal Neurologic Institute coordinate space. We varied the spatial extent of the masks. Confidence intervals of the FAIs for both control groups were calculated.
Results: We found the resulting FAIs to be highly robust against considerable mask variations (product-moment correlation: r > 0.97). The FAIs of the 4 patients presented with neurologic conditions associated with white matter alterations significantly fell outside the confidence intervals for normal FA.
Conclusion: FAIs based on mean fractional anisotropy values obtained from isotropic whole-head high-field diffusion tensor imaging by fully automated algorithms represent a robust and observer-independent measure for the comparative assessment of white matter integrity, ideally suited for further statistical treatments.