P‐038: MRI system tracking and correction using the adni phantom

Background: MRI systems used primarily for clinical practice are also used in clinical imaging research. Clinical considerations typically drive scanner maintenance schedules. The Alzheimer’s Disease Neuroimaging Initiative (ADNI) is a large multi-center natural history study. Approximately 800 subjects will be tracked longitudinally. Employing 80 different scanners over five years, upgrades and recalibrations will be unavoidable. Objective(s): To track and compensate for scanner upgrades in serial image data using a phantom. Methods: As part of the ADNI MRI scanning protocol a phantom is scanned immediately after each human study. Comparison of the phantom as imaged to the physically constructed phantom allows the determination of dimensional scale factors in 3D which may be used to track scanner performance. The same scalings may be applied to the human images. Results: For each phantom scan on a single representative 1.5T scanner, the linear scale factors are plotted versus scan date (Figure 1). Ideal scale factors are unity. Dashed lines indicate maintenance or upgrade dates. Slow drifts are observed as well as discrete changes associated with upgrade/maintenance. If the phantom is capturing scanner drift then application of the scaling factors to human images should improve the spatial consistency of the images. Co-registration (including scaling) of pairs of scans for individual subjects with and without phantom-based scaling was carried out. For each subject, pairs of images were acquired on a single scanner. Data from 19 pairs of images acquired on multiple 1.5T scanners from a single MRI vendor are shown in histograms of co-registration scaling factors in the logical X (R/L) direction for pairs of images with (without) phantom scaling correction (Figure 2). The distribution of scale factors after phantom-based correction is narrower and centered closer to unity than the uncorrected distribution. Similar results (not shown) are found for the remaining two cardinal directions. Phantom based correction reduces the width (RMS) of the co-registration scale factor distributions by one quarter to one half. Conclusions: The ADNI phantom may be used to capture longitudinal changes in scanner performance. Our findings suggest that the phantom reduces measurement error, and thus may be expected to reduce variance in clinical trials and increase power. P-039 TELOMERE SHORTENING AND/OR “ABSENCE” MAY INDICATE DEMENTIA/AD STATUS IN OLDER INDIVIDUALS WITH DOWN SYNDROME