ONE IMPLEMENTATION OF MATHEMATICAL MORPHOLOGY IN MEDICAL (ECG) APPLICATION

Abstract

Acquired ECG signal is degraded by low-pass EMG drift modifying its baseline; and by power net interference, high-pass noise and A/D conversion compromising its morphology. One of the main problems in the automated ECG analysis is to filter out the baseline wander, and extract the isoelectric reference, thus enabling accurate measurements and morphology recognition. Using Mathematical Morphology (MM) binary primitive operators, accurate baseline extraction is performed in three steps. First is to determine the exact heart rate (HR), which is done by high-pass MM filtering and R-wave detection. In the second step, preliminary baseline estimation is performed using low-pass MM filtering, thus allowing accurate morphology recognition. In the third step, the baseline is corrected using the ending points of TP intervals (P-start) which are considered as truly isoelectric. This procedure allows accurate baseline extraction and recognition of the complete morphology. The Mathematical Morphology offers a reliable solution for the baseline extraction problem, allowing ECG analysis for holter and monitoring applications.