ORIGINAL ARTICLE ECG signal analysis for the assessment of sleep-disordered breathing and sleep pattern K. Kesper • S. Canisius • T. Penzel • T. Ploch • W. Cassel Received: 10 March 2011 / Accepted: 10 December 2011 / Published online: 23 December 2011 Ó International Federation for Medical and Biological Engineering 2011 Abstract The diagnosis of sleep-disordered breathing (SDB) usually relies on the analysis of complex poly- somnographic measurements performed in specialized sleep centers. Automatic signal analysis is a promising approach to reduce the diagnostic effort. This paper addresses SDB and sleep assessment solely based on the analysis of a single-channel ECG recorded overnight by a set of signal analysis modules. The methodology of QRS detection, SDB analysis, calculation of ECG-derived res- piration curves, and estimation of a sleep pattern is described in detail. SDB analysis detects specific cyclical variations of the heart rate by correlation analysis of a signal pattern and the heart rate curve. It was tested with 35 SDB-annotated ECGs from the Apnea-ECG Database, and achieved a diagnostic accuracy of 80.5%. To estimate sleep pattern, spectral parameters of the heart rate are used as stage classifiers. The reliability of the algorithm was tested with 18 ECGs extracted from visually scored polysomn- ographies of the SIESTA database; 57.7% of all 30 s epochs were correctly assigned by the algorithm. Although promising, these results underline the need for further testing in larger patient groups with different underlying diseases. Keywords ECG analysis  Heart rate variability  Apnea detection  Estimated respiration  Sleep stages 1 Introduction The assessment of sleep-disordered breathing (SDB) usu- ally requires the recording of various physiological parameters during an overnight investigation in specialized sleep centers. The need for well-trained attending person- nel and the complex recording equipment make sleep studies rather expensive. Hence, a reduction of diagnostic complexity is a promising approach. Looking for alterna- tive methods for sleep assessment, computer-assisted sig- nal analyses capable of gathering the required diagnostic information from a reduced set of biosignals or even a single signal play an increasing role [1, 36, 37]. The electrocardiogram (ECG) is probably the most feasible biosignal for SDB and sleep assessment. The ECG is modulated by respiration, sleep, and the autonomic nervous system, often in a specific manner. Therefore, the reconstruction of a physiological trait from the ECG by means of signal analysis is often possible. Compared to a complete polysomnography, the measurement of the ECG is less stressful for the patient and there is less sleep dis- turbance by the technical equipment. With a signal strength of 1–2 mV, the ECG achieves the best signal-to-noise ratio among all electro-physiological signals. Therefore, the ECG generally provides reliable results, even in unattended measurements in the patients’ domestic environment. It is well known that SDB causes cyclical variations of the heart rate (CVHR) [16] and that parameters extracted from the shape of the signal allow the derivation of a respiratory effort-related curve (ECG-derived respiration, EDR) [25]. More recent advances in signal processing have shown that a reliable detection of SDB based on a single- channel ECG is possible [31]. Additionally, the ECG is influenced by sleep in various ways: The heart rate vari- ability (HRV) [3, 35] and the distribution of LF (low K. Kesper (&)  S. Canisius  T. Ploch  W. Cassel Department for Internal Medicine, Section Respiratory Diseases, Faculty of Medicine, Philipps-University Marburg, Baldingerstr. 1, 35043 Marburg, Germany e-mail: kesperk@med.uni-marburg.de T. Penzel Charite ´ Universitaetsmedizin Berlin, Interdisziplinaeres Schlafmedizinisches Zentrum, Berlin, Germany 123 Med Biol Eng Comput (2012) 50:135–144 DOI 10.1007/s11517-011-0853-9