Cardiac Arrhythmia Detection using Laplacian Eigenmaps and Wavelet Transform

Cardiac Arrhythmia is the most common causes of death .These abnormalities of heart may cause sudden cardiac arrest or cause damage to heart. This paper demonstrates theapplication of the Laplacian Eigenmaps (LE) and wavelet transform to the task of cardiac arrhythmia detection. LaplacianEigenmaps is a dimension reduction method which combines the benefits of latent variable models with spectral manifold learningmethods-no local optimum, ability to unfold nonlinear manifolds, and excellent practical scaling to latent spaces of high dimensions.In this research, two dimensional wavelet transform was appliedon ECG signal, and then a modified Laplacian eigenmap mapping was used to reduce the final feature vector size. Finally, a feedforwardneural network is used to classify ECG signal beats. Proposed Laplacian eigenmap were compared with other commonused Laplacian Eigenmaps. Results authenticate superiority of the proposed modified Laplacian eigenmap. Also, some waveletfunctions were tried to see their effect on the overall results. In thisstudy, we achieved average positive predictive accuracy as 99.14%, total accuracy as 99.13% and average specificity as 99.83% on MIT-BIH arrhythmia database

Cardiac Arrhythmia Detection using Laplacian Eigenmaps and Wavelet Transform Keywords:

Cardiac arrhythmia , Discrete wavelet transform , Laplacian eigenmaps , Feed forward neural network