Derivation Of Respiratory Signals From Single-lead ECg

Respiration is an important parameter in intensive care monitoring. There is evidence that arterial fibrillation, hypertension, and heart failure may be mediated by respiratory dysfunction. Respiration monitoring is a key function of mobile health care monitor,mainly used for monitoring respiratory rate of high-risk patients and new born babies,detecting obstructive sleep apnea syndrome(OSAS).Traditional methods for detection of respiration are either by directly measure air flow in or out of the lung or by indirectly measure body volume changes. These techniques require the use of cumbersome devices that may interfere with natural breathing. These devices are unmanageable in certain applications such as ambulatory monitoring, stress testing and sleep studies. All of these shortcomings stop them from daily use. ECG is the most accessible and the most commonly used electro diagnostic method. Information included in the ECG signal concern not only the heart function but also other systems, especially the respiratory one. With the development of signal processing techniques, people can derive respiratory information from ECG, so–called ECG-derived respiratory signal (EDR). This technique does not need extra respiration testing equipments, thus greatly reduces the complexity of the equipment and the users' inconvenience, so it can be used for daily monitoring. We completely focus on derive respiratory signal from single-lead ECG. Using the EDR information, we can monitor the real time breathing conditions of the user. If no breathing is detected for a certain time, the alarm equipment will send out altering signal to prevent the user from sudden death.In the first part of this paper, we introduce the background and the significance of this technique. The development and the situation of EDR in and abroad are also introduced. We give a brief explanation of the simulation software and signal used in our research in this part.The second part is signal pre-processing. First, we introduce the basic knowledge of ECG and different kinds of noises. Then we give a brief introduction of wavelet transform, the correlation between wavelet transform maximum or zero points and singular points of signal. In the end, we decompose ECG signal into seven layers using Coif4 wavelet, remove baseline drift using wavelet construction algorithm, then, power line interference and electromyographical interference are removed using an improved threshold method to get a clean ECG signal.The third part is the core part of our EDR algorithm. Here, we introduce two widely used EDR algorithms, one is based on heart rate variability (HRV) and the other based on mean electrical axis. After that, our algorithm is given. We detect the characteristic points of the ECG signal, including R peak value, R peak position, QRS complex starting and ending points. With these characteristic points, we calculate the width of the QRS complex and Lipschitz index to tell the supra ventricular beats from the ventricular beats. At last, we use the second approximation of wavelet transform of supra ventricular beats to get EDR. Using moving average method, we calculate the breathing rates per minutes.In the forth part of the paper, we evaluate the performance of the technique and explore its limitations. The application prospects of the algorithm are also discussed here.