IC‐P‐147: Combining white matter density and atrophy information to improve the characterization of axonal loss in Alzheimer's disease

Medical images provide two sources of quantitative information to investigate pathology: 1) image intensity, 2) image morphology. Degeneration of axons in Alzheimer's disease (AD) is often detected using surrogate MRI measures for white matter 'integrity' (e.g. Fractional Anisotropy, FA). These measures are based entirely on image intensity and discard useful information provided by morphology. Morphology information is pertinent to investigating AD, as cellular debris is cleared after degeneration and axon loss largely manifests as white matter atrophy. We applied a new diffusion MRI measure called Apparent Fibre Density (AFD) [1] to compare 46 AD patients with 94 age-matched healthy volunteers (Australian Imaging Biomarkers & Lifestyle study). AFD is proportional to the image intensity, which is related to the quantity of intra-axonal water, and is therefore sensitive to the number of axons within a voxel (density). To perform voxel-wise comparisons of AFD, images were non-linearly warped to a common template image. We incorporated morphology information by modulating the AFD in each voxel based on the change to a fibre bundle's cross-sectional area (atrophy). Modulation ensures the AFD in template space encapsulates both sources of axonal loss (axon density as measured by the MRI signal, and volume loss due to atrophy). We investigated whole-brain group AFD differences with and without modulation using novel tractography-based statistics. Significant AFD decreases in AD were observed in white matter bundles known to be involved in language and memory (Fig.1). As shown in columns 1 & 2, the group AFD difference is more extensive when morphology information is included via AFD modulation. Since AFD differences are associated with a voxel and direction (shown by the direction-colour-encoded tractogram in Fig.1) population differences can be attributed to a specific fibre bundle in regions with crossing-fibres (e.g. Arcuate Fasciculus). The absence of differences in fibre density (column 1) within regions previously detected using FA suggests that FA decreases detected in AD are largely caused by atrophy-induced partial volume changes. By incorporating morphology information, AFD is potentially a more sensitive and interpretable biomarker of axon degeneration compared to existing diffusion MRI measures based on image intensity alone.[1] Raffelt et al.2012,59(4):3976–94. Significant AFD decreases in AD vs healthy controls (p<0.05, corrected for multiple comparions). Left: Significant AFD decreases without modulation (a reduction in fibre axon density). Middle: Significant AFD decreases with modulation (axonal loss due to density of fibers within the bundle and reduction in the volume of the bundle itself). Right: Whole-brain population-average tractogram included as an antomical reference image. Tracks are colour-coded by direction (red: left-right, green: anterior-posterior, blue: inferior-superior).