Cepstral based approach for online quantification of ECG quality in freely moving subjects

We present an algorithm for assessing the ECG signal quality in real time. The algorithm is designed for small systems with low computational power. The method estimates the ECG cepstrum on a running window of 10 s and calculates the power of the highest peak in the “quefrency” band of the mean cardiac interval (between 0.25 and 2.0 s). If the ECG is corrupted by noise, the power of this cepstral peak is just a fraction of the total power. The ECG quality index (QI) is defined as ratio between the power of the cepstral peak and the total cepstral power. Examples with ECG of different signal quality from different leads illustrate the method. Moreover, two overnight ECG recordings were analyzed by setting the threshold for ECG acceptability at QI>;0.40. ECG signals were classified acceptable for 99.8% and 60% respectively of the recordings time. These percentages were similar to those scored visually by an expert operator (100% and 67% respectively).