Signal processing techniques for removal of various artifacts from obstructive sleep apnea signals

Abstract

Sleep is judgmental to health and well-being. Deficient quality sleep is similar with a wide range of negative outcomes varies from schizophrenia to cardiovascular disorders. Obstructive sleep apnea is one of the sleep disorders. In order to identify the various syndromes the signals are need to record by using the sensors. Sleep signals are recorded by using the polysomnography (PSG) labs which is the old traditional and gold standard for recording the sleep signals. PhysioNet is a large online medical database that consists of a large collection of recordings of various physiological signals. PhysioNet database consist of sleep apnea database available. Physionet website is a universal service, physionet resource supported by the national institute of health’s National Institute of Biomedical Imaging and Bioengineering (NIBIB) and National Institute of General Medical Sciences (NIGMS). This survey paper aims to bring the different Signal Processing Techniques for Removal of Various Artifacts from Obstructive Sleep Apnea Signals to identify sleep apnea syndrome, because pre-processing is most effective and efficient to reduce unwanted signals from the original sleep signals. While recording the sleep apnea signals various artifacts records along with raw signals either directly or indirectly due to the internal and external sources like Power line interference, Muscle contractions, Electrode contact noise, Motion Artifacts, Baseline wandering, Noise generated by electronic circuits, while breathing and coughing, body position movements etc, and they need to be eliminated in order to acquire genuine health information. So in order to remove there artificats from the sleep signals the signal processing techniques (filtering techniques) are predominantly used for pre-processing of the sleep signals and have been executed in a wide variety of systems for analysis. Filtering of the sleep signal is contingent and should be implemented only when the required statistics remains cryptic.