SVM Based fMRI Data Classification: An Approach to Decode Mental State

Recently, a growing number of studies have shown that machine learning technologies can be used to decode mental state from functional magnetic resonance imaging (fMRI) data. Two feature extraction methods are compared in this paper, one is based on the cumulative change of blood oxygen level dependent (BOLD) signal of activated brain areas, and the other is based on the values at each time point in the BOLD signal time course of each trial. The authors collected the fMRI data while participants were performing a simplified 4×4 Sudoku 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 approach 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 extraction based on serial signal change of BOLD effect can predict cognitive state better because it could use abundant and typical information kept in BOLD effect data. By ranking accuracy of every single-voxel based classifier, it is interesting that voxels with higher accuracy are anatomically located in PPC, PFC, ACC and other brain regions closely related to problem solving, which is consistent with previous studies in cognitive neuroscience. The methods might shed light on brain computer interface (BCI).