Information Technology Reference
In-Depth Information
Chapter 5
Comparison of Supervised Classification
Methods with Various Data Preprocessing
Procedures for Activation Detection in fMRI
Data
Mahdi Ramezani and Emad Fatemizadeh
Abstract
In this study we compare five classification methods for detecting ac-
tivation in fMRI data: Fisher linear discriminant, support vector machine, Gaus-
sian nave Bayes, correlation analysis and
k
-nearest neighbor classifier. In order
to enhance classifiers performance a variety of data preprocessing steps were em-
ployed. The results show that although
k
NN and linear SVM can classify active and
nonactive voxels with less than 1.2% error, careful preprocessing of the data, in-
cluding dimensionality reduction, outlier elimination, and denoising are important
factors in overall classification.
5.1 Introduction
Studying the functionality of the brain with versatile noninvasive tools has boost
enormously in recent years. It is widely believed that blood oxygen level, the ratio
of oxygenated to deoxygenated hemoglobin in the blood at the corresponding in
the brain, is influenced by local neural activity. Based on the blood oxygen level-
dependent (BOLD) principle, functional magnetic resonance imaging (fMRI) has
become one of the typical tools in the neurological disease diagnosis and human
brain research. This imaging method can quantify hemodynamic changes induced
by neuronal activity in human brain at high-spatial resolution during sensory or
cognitive stimulations. fMRI technology offers the promise of revolutionary new
approaches to studying human cognitive processes, provided we can develop ap-
propriate data analysis methods to make sense of this huge volume of data. The
