Spatio-Temporal Models for Multi-Scale Multi-Modal Data

• Activity Number: 226
• Type: Topic Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Statistics in Mental Health Research
• Author(s): Hernando Ombao*+
• Companies: University of California at Irvine
• Keywords: spatio-temporal data ; multi-scale models ; spectral analysis ; multi-modal time ; brain imaging ; neuroscience

Abstract

This project is motivated from challenges in modeling and analyzing spatio-temporal multi-modal data with different spatial and temporal resolutions. Such data arises in neuroscience, among many others, where one goal to study the spatio-temporal dynamics of brain processes during resting state or in response to a particular stimulus. Here, the underlying human neuronal activity is often not directly observed. Instead, this activity is observed indirectly through different imaging modalities such as functional magnetic resonance imaging (fMRI) and electroencephalograms (EEG). On the one hand, EEGs provide excellent temporal resolution but have poor spatial resolution. On the other hand, fMRI has excellent spatial resolution. The most common practice is to analyze these two data separately. Here, we develop a statistical framework for a simultaneous analysis of these two data. Our approach is based on spatially-varting wavelet stochastic representations of the non-stationary spatio-temporal data. We present statistical inference methods under the proposed model.

This is joint work with John Aston (Warwick University).