The advent of therapies for Alzheimer disease (AD) and related disorders has increased interest in biological markers of early disease and in surrogate measures of disease progression. Magnetic resonance imaging (MRI) has potential in both these areas. Direct in vivo imaging of specific molecular pathological features (eg, β-amyloid plaques and neurofibrillary tangles in the AD brain) is not yet possible. However, MRI can be used to measure a surrogate marker of neuronal damage, that is, atrophy. Imaging studies have shown marked regional (eg, hippocampus and entorhinal cortex) and global (whole brain and ventricles) atrophies to be a consistent feature of AD. Because of the wide normal variation in cerebral structure, there is considerable overlap in these measures when mildly affected patients and matched controls are compared. Rates of atrophy from serial imaging allow individuals to form their own control avoiding problems of intersubject morphological variability.
Registration (positional matching) of serial MRI is now possible to subvoxel (0.2 to 0.4 mm) levels and permits direct measurement of rates of atrophy. The methods are automated and unbiased, independent of ora priori decisions about regions of interest, and avoid laborious manual measurements. Rates of cerebral loss derived in this way are highly reproducible (to 0.2% of brain volume), are significantly raised in AD (2.5% ± 1.1%/y vs controls 0.3% ± 0.4%/y), and are sensitive even in early cases. Rates of loss correlate with cognitive decline in untreated patients with AD. These techniques, together with volumetry, may prove useful both in diagnosis and in tracking disease progression.