Meeting Banner
Abstract #3014

Improved Longitudinal Gray Matter Atrophy Assessment Via a Combination of SIENA and a 4-Dimensional Hidden Markov Random Field Model

Michael G. Dwyer1, Niels P. Bergsland1, Robert Zivadinov1

1Buffalo Neuroimaging Analysis Center, University at Buffalo, Buffalo, NY, United States

We describe a novel technique for substantially improving the reliability of longitudinal gray matter atrophy measurement through the extension of SIENAXs hidden Markov random field model from 3 dimensions to 4. We validate our approach using both simulation and real clinical data, and show a marked improvement in effect size and statistical power.

Keywords

accurate actual addition address aging agreed aimed allows although ambiguous analyses applicable applied arbitrarily artificially assessment assisted atrophy audience automated avoid basic better brain buffalo called chain classifying clinical clinically clinicians combination combine computer consider constrained contributing controls correlated create dataset datasets determine developed differently difficult dimensional direct discrepancies elements evaluate evaluated evaluation excellent expectation extended extraction fast field fields finally generally gray halfway healthy hidden improved improvements includes independent intensity involved issues joint journal known larger longitudinal longitudinally mask matched mathematical maximization measures model much neighbors null others parallel pathologic patients performance period position power precise prevent progress progressed proven random reduced related reproducibility rescan researchers robust scaled scaling scanner science seeking segmentation serial significantly simulation skull smith sought space springer statistical temporal tissue together tomography tool trans trend trials unbiased understand unified variance volunteers whereas white whole year