Classification of electrocardiogram using SOM, LVQ and beat detection methods in localization of cardiac arrhythmias

The work investigates a set of efficient methods to extract important features from the ECG data applicable in the localization of cardiac arrhythmia. The work involves the segmentation of the ECG signal and the extraction of important features like QRS and ST segments. Further classification follows the learning process where the SOM (Self Organizing Maps) units organize in such a way that similar map sequences of the ECG data are represented in particular areas of the SOM. Eventual unsupervised learning (UL) time traces are achieved during the training and forwarded to the LVQ (Learning Vector Quantization). Here a set of supervised learning (SL) is followed by a smart beat detection system that further enhances the signal performance and correct localization for arrhythmia detection.