Using SVM to Predict High-Level Cognition from fMRI Data: A Case Study of 4*4 Sudoku Solving - Brain Informatics - page 171

Using SVM to Predict High-Level Cognition
from fMRI Data: A Case Study of 4*4 Sudoku
Solving

Jie Xiang 1 , 2 , Junjie Chen 1 ,HaiyanZhou 2 , Yulin Qin 2 , 3 ,
Kuncheng Li 4 , and Ning Zhong 2 , 5

1 College of Computer and Software, Taiyuan University of Technology, China
2 The International WIC Institute, Beijing University of Technology, China
3 Dept of Psychology, Carnegie Mellon University, USA
4 Dept of Radiology, Xuanwu Hospital Capital University of Medical Sciences, China
5 Dept of Life Science and Informatics, Maebashi Institute of Technology, Japan
[e-mail], [e-mail]

Abstract. In this study, we explore the approach using Support Vec-
tor Machines (SVM) to predict the high-level cognitive states based on
fMRI data. On the base of taking voxels in the brain regions related
to problem solving as the features, we compare two feature extraction
methods, one is based on the cumulative changes of blood oxygen level
dependent (BOLD) signal, and the other is based on the values at each
time point in the BOLD signal time course of each trial. We collected
the fMRI data while participants were performing a simplified 4*4 Su-
doku problems, and predicted the complexity (easy vs. complex) or the
steps (1-step vs. 2-steps) of the problem from fMRI data using these
two feature extraction methods, respectively. Both methods can produce
quite high accuracy, and the performance of the latter method is better
than the former. The results indicate that SVM can be used to predict
high-level cognitive states from fMRI data. Moreover, the feature extrac-
tion based on serial signal change of BOLD effect can predict cognitive
states better because it can use abundant and typical information kept
in BOLD effect data.

1 Introduction

Methods in machine learning, such as classification, have been introduced into
fMRI (functional Magnetic Resonance Imaging) data mining to investigate neu-
ral cognitive mechanism and decode mind states. In early years, Haxby and
colleagues used Support Vector Machines (SVM) to train classifiers to predict
whether a participant was watching a shoe or a bottle [1], and a human face or
an object [2], which provided more information about semantic representation
in human brain. Since then, more and more researchers have investigated the
methods of machine learning to analyze fMRI data collected in visual or audi-
tory perception tasks. Kamitani and Tong's work showed that different angles
of gratings in vision could be distinguished through fMRI data [3]. By analyzing