CMStatistics 2016: Start Registration
View Submission - CMStatistics
B1159
Title: Association of structural brain imaging markers with alcoholism using structural connectivity via a regularized approach Authors:  Jaroslaw Harezlak - Indiana University School of Public Health-Bloomington (United States) [presenting]
Mario Dzemidzic - Indiana University School of Medicine (United States)
Joaquin Goni - Purdue University (United States)
David Kareken - Indiana University School of Medicine (United States)
Marta Karas - Indiana University Fairbanks School of Public Health (United States)
Abstract: Brain imaging studies collect multiple imaging data types, but most analyses are done for each modality separately. Statistical methods that simultaneously utilize and combine multiple data types can instead provide a more holistic view of brain function. We utilize cortical thickness measures obtained by FreeSurfer software to predict the alcoholism-related phenotypes while incorporating prior information from the structural connectivity between cortical regions. We develop a functional linear model with a penalty operator to quantify the relative contributions of imaging markers obtained from high resolution structural MRI (cortical thickness) as predictors of drinking frequency and risk-relevant personality traits, while co-varying for age. We estimate model parameters by a unified approach directly incorporating structural connectivity information into the estimation by exploiting the joint eigenproperties of the predictors and the penalty operator. We applied the developed methods to a sample of 148 young (21-35 years) social-to-heavy drinking male subjects from several alcoholism risk studies. Structural connectivity model was used to estimate the density of connections between 66 cortical regions based on Desikan-Killiany atlas. Using our method we found the best 11 average cortical thickness markers of the Alcohol Use Disorders Identification Test score.