Compared the reproducibility of methods for measuring hippocampal atrophy rates.
Compared FreeSurfer, FSL/FIRST, MAPS-HBSI, AdaBoost and manual.
Back-to-back MPRAGEs at baseline and year one for N=562 ADNI1 1.5 T subjects.
Used a novel but simple statistical test that is robust to outlying data points.
MAPS-HBSI should require half the subjects in a clinical trial than others tested.