APPLICATION OF THE HAUSDORF METRICS FOR DETERMINING ATYPICAL CARDIOCYCLES IN THREE-DIMENSIONAL VECTORCARDIOGRAM PHASE SPACE COORDINATE

Abstract

Background. An electrocardiogram (ECG) is a cyclic but not a periodic signal. On the electrocardiogram, similar but not identical cycles can be distinguished. Therefore, diagnostic algorithms use averaging procedures. Real cardiograms may have significant distortion of individual cardiac cycles. We call these cycles atypical. These cycles should be distinguished and not used in averaging.

Materials and methods. To detect atypical cycles in the electrocardiogram, we form a three-dimensional phase space of coordinates using three ECG leads. ECG signals are pre-filtered and divided into cardiac cycles and elements. Each ECG cycle is presented as a three-dimensional curve arc. Each point of a curve has three coordinates. These coordinates are the values of the ECG in three leads at a certain point of time. For all pairs of three-dimensional curves, we calculate the Hausdorff metrics - the distance between each two curves. We consider atypical the cycles whose Hausdorf metrics will be significantly larger.

Results. The algorithm was tested experimentally on 142 individuals who were in different emotional and physical condition. Three ECGs have been recorded for each person. The first three ECG leads have been used to form curve arcs in the three-dimensional coordinate space. After calculating the Hausdorff metric, atypical cycles have been detected and removed from further calculations.

Conclusions. The Hausdorff metric in the three-dimensional phase space of coordinates is an effective tool for detecting atypical ECG cycles. The removal of atypical cardiac cycles from further calculations allows improving the results of identification of a person by his electrocardiogram.