A new method for evaluating ECG signal quality for multi-lead arrhythmia analysis

Abstract

A simple and computationally efficient new signal quality measure that is responsive to combinations of both physiological and non-physiological noise has been developed The new signal quality measure is based on the use of the area differences between successive QRS complexes. The signal quality assessment for each lead is made on the basis of the characteristics of the statistical distribution of the area differences obtained over a period of time. The new signal quality measure was evaluated using the 44 non-paced patient records from the MIT-BIH two-channel arrhythmia database. Results presented in histogram and cumulative histogram plots showed that the signal quality can be accurately assessed using this new method. Of the total 44 non-paced records, the new method identified 22 records where one of the two leads had much better signal quality than the other lead When this information was used for arrhythmia analysis, the averaged PVC false positive rate was 0.47% for leads that were selected by the new method, and 2.56% for leads that were not selected These results clearly showed that the new signal quality measure developed can be used to accurately assess the ECG signal quality and can be incorporated easily into existing arrhythmia algorithms for performance improvement.