Classification of hypnotisable groups based on normal EEG signals using the Recurrence Quantification Analysis and Support Vector Machine

Hypnosis is an interesting brain state for researchers and can be used as hypnotherapy. The efficiency of hypnotherapy profoundly depends on the hypnotisability of subjects to reach depth of hypnosis. In this research, the normal EEG signals have been used in three mental tasks include: relax state with closed eyes (base line), mental multiplication and geometric figure rotation to evaluate groups with low, medium and high hypnotisability. Most previous researchers used hypnotic EEG signals whereas in our work the normal EEG signals have been used. Furthermore, features are extracted using RQA (Recurrence Quantification Analysis) method then best features were selected by Scaled Class Separability Selection algorithm and were applied to the SVM (Support Vector Machine) classifier. The performance of classifier is evaluated using leave-one-out cross-validation method. Groups are separated with 69.69% in the first mental task, 66.66% and 78.78% in second and third mental tasks respectively.