Meeting Banner
Abstract #0684

Robust High-Dimensional Morphological Metric: Application to the ADNI Multi-Centric Dataset

Nicolas Robitaille1, Abderazzak Mouiha1, Simon Duchesne1,2

1Centre de recherche Universit Laval Robert-Giffard, Qubec, QC, Canada; 2Radiology, Universit Laval, Quebec, QC, Canada

Structural MRI has been proposed to fulfill the role of quantitative biomarker in Alzheimers disease. We proposed a high-dimensional morphological metric extracted from T1-weighted MRI and now wish to demonstrate its robustness in a multi-centric setting. To form our metric we used data from two different studies, totaling 300 subjects. We tested the metric over the 797 subjects of the ADNI dataset, and found an average scan/repeat scan distance of 1.7%. In order to detect a 15% difference between groups, this minimum precision threshold results in an increase from 59 to 75 participants to reach identical power in a trial.