1 Bayesian Analysis of Morphological Changes Associated with Mild Cognitive Impairment: A Cross-Sectional Study Hanchuan Peng * , Susan M. Resnick ‡ , Dinggang Shen † , Christos Davatzikos † , and Edward H. Herskovits † * NERSC Division, Lawrence Berkeley National Laboratory, MS.50F, One Cyclotron Rd, Berkeley, CA, 94720. Email: hpeng@lbl.gov † Section on Biomedical Image Analysis, Department of Radiology, School of Medicine, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA, 19104. Email: herskovi@rad.upenn.edu , dgshen@rad.upenn.edu , christos@rad.upenn.edu ‡ Laboratory of Personality and Cognition, National Institute on Aging, National Institutes of Health, Baltimore, Maryland 21224-6825. Email: resnick@lpc.grc.nia.nih.gov ABSTRACT In this paper we apply a new morphology-function analysis method, a Bayesian Morphometry Algorithm (BMA), to a set of cross-sectional magnetic resonance images of subjects in the Baltimore Longitudinal Study of Aging, some of whom have very mild cognitive impairment. Based on Bayesian model selection, this new method is able to test a series of hypotheses about morphology-function associations and determine morphological changes associated with clinical variables. INTRODUCTION The purpose of this paper is to describe the application of a new Bayesian method for morphological analysis, which can be used to determine associations among structural and clinical