B0473
Title: A Bayesian regression framework for brain imaging data with multiple structural- and network-valued predictors
Authors: Rajarshi Guhaniyogi - University of California Santa Cruz (United States)
Aaron Scheffler - University of California, San Francisco (United States) [presenting]
Rene Marquez - University of California Santa Cruz (United States)
Abstract: Clinical researchers collect multiple images from separate modalities (sources) to investigate questions of human health that are inadequately explained by considering one image source at a time. Viewing the collection of images as multi-objects, the successful integration of multi-object data produces a sum of information greater than the individual parts. Still, this integration can be hindered by the data complexity. Each image contains structural information, indexing spatial information, or network information, indexing connectivity among the image, which reinforce each other but are challenging to merge. We propose a Bayesian regression framework that provides inference and prediction for a scalar outcome as a function of a multi-object predictor. Our framework will accommodate multiple image predictors with different structures and identify image regions that influence the response jointly via efficient hierarchical prior structures that scale to high-resolution image data volume. A working example is provided to predict language comprehension scores from multi-object image data to explore the neural underpinnings of language loss in primary progressive aphasia patients.
