Cardiac decision making using higher order spectra

摘要: Abstract The electrocardiogram (ECG) is the P-QRS-T wave representing information about condition of heart. shape and size ECG signal may contain useful nature disease afflicting However, these subtle details cannot be directly monitored by human eye indicate a particular cardiac abnormality. Also, highly subjective, symptoms appear at random in time scale. Hence computer assisted methods can help physicians to monitor health easily accurately. nonlinear non-stationary nature. These variations captured using non-linear dynamical Higher Order Statistics (HOS) techniques. Bispectrum third order spectra which captures beyond mean standard deviation. In this work we have analyzed five types beats namely: Normal, Right Bundle Branch Block (RBBB), Left (LBBB), Atrial Premature Contraction (APC) Ventricular (VPC). extracted bispectrum features are subjected principal component analysis (PCA) for dimensionality reduction. components were fed four layered feed forward neural network Least Square-Support Vector Machine (LS-SVM) automated pattern identification. our work, obtained highest average accuracy 93.48%, sensitivity specificity 99.27% 98.31% respectively LS-SVM with Radial Basis Function (RBF) kernel. Our system clinically ready run on large amount data sets.