Automatic Extraction of HRV Sequences from Noisy ECG Data for Reliable Analysis and Telediagnosis

P. Padmanabhan, Z. Lin, M.E.H. Ong, W. Ser, and G.-B. Huang (Singapore)


Heart rate variability, electrocardiogram, noise, signal processing.


Heart rate variability (HRV) parameters are affected by the presence of noise and non-sinus rhythm, resulting in errors in analyses that could have important clinical applications. This paper presents techniques to automatically extract beat-to-beat (RR) interval sequences from noisy electrocardiogram (ECG) signals for reliable HRV analysis. The system consists of pre-processing stages for noise reduction, as well as algorithms for automatic beat detection and isolation of ectopics and outliers with minimal user intervention. The algorithms were tested against manually annotated data from the MIT-BIH database and were found to detect and isolate non-sinus rhythm with a high sensitivity (99.8%) and specificity (99.4%). The results have important applications in ambulatory HRV analysis and telediagnosis where HRV parameters often have to be calculated in high levels of noise.

Important Links:

Go Back