Objectives. Cheyne-Stokes Breathing (CSB) is a type of sleep-disordered respiration characterized by a crescendo-decrescendo pattern of ventilation, alternating hyperventilation and central hypopneas/apneas. CSB is mainly prevalent in patients with severe heart failure (ventricular ejection fraction less than 30%) and can be associated with a worse prognosis. Overnight multi-channel polysomnography is commonly used for diagnosis but as it is a labor-intensive method can prevent from identifying patients with periodic breathing preceding CSB. Our objective is to test the feasibility of detecting early stage of CSB based on a single electrocardiogram signal.
Methods. Holter recordings and respiratory signals from 3 patients presenting heart failure with low ejection fraction (<30%; HFrEF) were analyzed. Each patient showed different types of breathing (normal, periodic breathing and severe CSB annotated by an expert). Two ECG-derived respiration (EDR) signals: (1) Heart-Rate Variability (HRV) and (2) R-Wave Amplitude (RWA) were computed and jointly used to estimate respiratory events (classified as central, mixed or obstructive apneas/hypopneas), respiratory rate and amplitude modulation (main peak frequency extracted using Fast Fourier Transform).
Results. Our algorithm achieves good performance for the detection of breathing cycles compared with the ventilation signal (Se = 97.43%; Pr = 100% for normal respiration and Se = 91.7%; Pr = 74.5% for CSB). Final classification based on decision tree using respiratory events, Apnea/Hypopnea Index (AHI) and main peak frequency detected in HRV and RWA, shows exact correlation with the expert, even if AHI is overestimated in EDR signals as hypopneas tend to be missed by the detection and classified as apneas.
Conclusion. This paper shows the feasibility to detect early stages of CSB using a single electrocardiogram. The proposed method has now to be validated on a wider range of HFrEF patients.