A Simple Online Histogram and Pattern Recognition Based ECG Analyzer

Abstract

A combination of histogram and heuristic rule based approaches are used for detection of different patterns of ECG waves and extraction of few important time -plane features, which are used for ECG interpretation and classification. At first, an efficient, simple and new approach for detection of base line and QRS complexes from the horizontal and vertical histogram of ECG images plotted on computer screen is described. The vertical and horizontal histograms are generated from the computation of number of ordinates for a particular abscissa at the vertical and horizontal direction respectively. The base line is determined at the maximum of the horizontal histogram whereas QRS or R peaks are determined from the local maximas of the maximum area zone of vertical histograms. A high accuracy is achieved for both the cases ( 99.5% for QRS and 92% for base line). This method is advantageous for onscreen pictorial analysis because both QRS and base lines can be determined directly from computer screen ECG data without use of complex mathematical models, even when ECGs are tilted due to respiration and disturbed in the presence of power line oscillation. After detection of R points and base points, an efficient algorithm is developed using priory knowledge of different patterns of ECG wave for detection of P, Q, R, S and T waves and their different attributes which are useful for ECG interpretation.