ECG Preprocessing, linear prediction coefficients, mel frequency cepstral
coefficients, morphological features, Y-coordinates, R amplitude, heart beat, histogram
of oriented gradients, row pixel count, support vector machine, auto associative
neural network, gaussian mixture model, ECG classification, template matching, cardiac
arrhythmia, myocardial infarction, conduction blocks, supraventricular tachycardia,
atrial fibrillation, ventricular tachycardia, anteroseptal infarction, anterior infarction,
inferior infarction, atrioventricular blocks, left bundle branch block, right bundle
branch block, physiobank
The objective of an ECG classification system is to classify the Electrocardiogram
(ECG) signal using one or more characteristics associated with the Electrocardiogram.
This thesis presents a method for ECG signal and ECG image preprocessing, feature
extraction and classification using features extracted from normal ECG and nine categories
of cardiac diseases. ECG preprocessing refers to denoising of the ECG signal
and resizing of the ECG image. ECG classification system classifies the ECG signal
and image into one of the predefined categories namely normal and Supraventricular
tachycardia, Atrial fibrillation, Ventricular tachycardia, Anteroseptal infarction, Anterior
infarction, Inferior infarction, Atrioventricular blocks, Left bundle branch blocks,
Right bundle branch blocks.
