Multilevel Classification Framework of fMRI Data: A Big Data Approach

Abstract Different activities of the brain are measured through functional Magnetic Resonance Imaging or functional MRI (fMRI) by perceiving changes related to blood flow. In order to create a detailed image of the body, this test uses a strong magnetic field combined with radio waves. The fMRI is widely used to comprehend various behavioral disorders associated with the nervous system, such as Alzheimer’s disease. fMRI produces an enormous amount of data that have to be evaluated, and thus generates a vast network of outcomes. The existing medical image processing tools and statistical methods are not capable of integrating resources to process this amount of data. To process the large datasets produced by fMRI scans, big data analytics platforms have been developed. Machine learning is one of the big data analysis methods that is suitable for use with fMRI data. In this chapter, we discuss how to train an fMRI dataset with different machine learning algorithms towards the enhancement of the precision of classification. Two classifier training methods are explored: logistic regression (LR) and support vector machine (SVM). A series of both subject-dependent as well as subject-independent experiments are carried out. In cases where the subject is fixed, better scores are achieved. Generally, logistic regression provided better scores compared to support vector machine.