Mental Stress Assessment of ECG Signal using Statistical Analysis of Bio-Orthogonal Wavelet Coefficients

It is observed that the stress level is function of various statistical parameters like standard deviation, entropy, energy, , mean, Co- variance and power of the ECG signals of two states i. e. normal state of mind and stressed state of mind. Further, it is observed that the features extracted are directly from the ECG in frequency domain using db4 wavelet. However, db4 introduces some error on account of db4 wavelet shape. This error in turn amplifies while measuring the features as mentioned above. In order to reduce this error, we propose a Bior 3.9 wavelet family to decompose the ECG signal. The decomposed ECG signal is now analyzed statistically to extract the above feature. . Further, in the existing work, all the ECGs are taken by using the 12 leads method. This factor also adds some undue stress to the person under scanner. And this is time consuming too. Therefore, to reduce this complexity, we propose to analyze all above features by using a two lead ECG. Further, a back propagation neural network is trained using the above features as input neurons in order to classify the stress level.