Monitoring the brain's longitudinal changes in clinical trials for Alzheimer's disease: a robust and reliable non-rigid registration framework

polycarbonate. The average signal intensity of tissue surrounding the lateral ventricles was measured in 3T T1-weighted MRI images from ten subjects with AD in the ADNI database. The mixture found to replicate this tissue was made of 2% (w/v) agar dissolved in water, containing 0.01% (w/v) NaCl and a gadopentetate dimeglumine concentration of 0.0375 mM. A brain-shaped mold was filled with tissue-mimicking solution and the ventricle positioned in the middle of the brain. The completed phantom was scanned using the ADNI-specific MP-RAGE sequence. Images were analyzed using a fully-automated segmentation tool. Results: The signal intensity difference between the ventricle and agar solution successfully matched in-vivo signal intensity differences. At a resolution of 1.0 x1.0 x1.0 mm, a volume of 46.6 cm was reported, which is only 2.3% smaller than the actual volume. This result illustrates that the software used does a reasonable job of estimating ventricle volume. The use of the ventricle phantom allowed us to also pinpoint ventricular sub-regions where the algorithm failed. Conclusions: A life-sized MRI-compatible brain ventricle phantom was successfully created. Images acquired are available from the authors to groups wishing to use this data for validation. As enlargement of the ventricles is further established as a marker of disease progression and incorporated into clinical trials, careful validation using gold standards must be performed to ensure the integrity of the study.