Ensemble Empirical Mode Decomposition of Photoplethysmogram Signals for Assessment of Ventricular Fibrillation

Ventricular fibrillation is a type of cardiac arrhythmia which is responsible for several cases of sudden cardiac arrests. As many cases of arrhythmia result to fatality, it is the goal of this research to develop a method to analyze this condition through the use of ensemble empirical mode decomposition (EEMD). EEMD is a variant of empirical mode decomposition (EMD) which solves its weakness in terms of mode mixing. EEMD results to the decomposition of a signal into its intrinsic mode functions(IMFs). The IMFs, together with their power spectral densities (PSDs) of photoplethysmogram (PPG) signals are analyzed for cases with and without ventricular fibrillation. Also, IMFs and PSDs are used as the features for classifying the presence of this condition. Principal component analysis (PCA) is used to reduce the large dimension of the features. In classifying, k-NN classifier was used. It was found that the IMFs of a signal with and without ventricular fibrillation resampled at 250 Hz and at window length of 1000 has most of its signal energy at the 5
^th
to 8
^th
siftings. The highest overall classification accuracy of 83.75%was achieved with noise width of 0.1. Thus, the ensemble empirical mode decomposition of PPG signals was successfully used for assessment of ventricular fibrillation and further modifications with the parameters and pre-processing techniques may be done to improve classification accuracy based on this feature.