Modeling and Nonlinear Analysis in fMRI via Statistical Learning - Advance Image Processing in Magnetic Resonance Imaging - page 551

18

Modeling and Nonlinear
Analysis in fMRI via
Statistical Learning

Yongmei Michelle Wang

CONTENTS

18.1 Introduction ...........................................................................................565
18.2 Background ............................................................................................566
18.2.1 Nonlinearities in fMRI............................................................566
18.2.2 Features of fMRI Data ............................................................566
18.2.3 fMRI Data Modeling ..............................................................567
18.2.4 Overview .................................................................................567
18.3 Statistical Learning Theory ...................................................................568
18.4 fMRI Data Analysis and Modeling through SVR ................................571
18.4.1 Data Representation ................................................................571
18.4.2 Temporal Modeling.................................................................571
18.4.3 Multiresolution Signal Analysis..............................................573
18.4.4 Merging Model-Driven with
Data-Driven Methods ..............................................................575
18.4.5 Generalization to Multisession Studies ..................................576
18.4.6 Testing on Real fMRI Data ....................................................578
18.5 Conclusions and Discussions ................................................................581
Acknowledgments .............................................................................................582
References .........................................................................................................582

18.1 INTRODUCTION

This chapter focuses on functional magnetic resonance imaging (fMRI) data analysis
and modeling using statistical learning techniques. fMRI is a powerful technique for
mapping brain function by using the blood oxygenation level dependent (BOLD)
effect (32); however, the small signal change due to the BOLD effect is very noisy
and susceptible to artifacts such as those caused by scanner drift, head motion, and
cardiorespiratory effects. Although a task or stimulus can be repeated over and over
again, there are limits due to time constraints, habituation effects, etc. Therefore,

565