Effectiveness of Tuned Q-factor Wavelet Transform in Emotion Recognition Among Left-brain Damaged Stroke Patients

Abstract

Emotion recognition is impaired among stroke patients due to brain injury. It has given negative impact towards stroke patients because of the difficulty in expressing themselves. Hence, the vision is to overcome this problem by creating a platform to predict the emotion of stroke patients for them so that recurrent stroke events can be avoided. EelectroEncephaloGraph (EEG) of 19 Left Brain Damage patients (LBD) are used as database. The objective of this paper is to compare the accuracy between Tuned Q-factor Wavelet Transform (TQWT) and Wavelet Packet Transform (WPT). The collected raw EEG signals are de-noised by using 6 th order Buttterworth filter. Then, filtered signals were fed into time-frequency analysis tools namely TQWT and WPT to transform the time domain signals into time-frequency domain signals. Hurst exponent feature is extracted from the corresponding time-frequency domain signals before enhancement by using principal component analysis (PCA). Lastly, classification is done through K-nearest neighbour (KNN), probabilistic neural network (PNN) and random forest (RF) in order to evaluate the performance of the recognition system. From the result, it is found that the classification accuracy is consistently higher in TQWT method.