Title :
Automated Sleep-Wake Detection in Neonates from Cerebral Function Monitor Signals
Author :
Eklund, J. Mikael ; Fontana, N. ; Pugh, E. ; McGregor, Carolyn ; Yielder, Paul ; James, Ashish ; Keyzers, Matthew ; Hahn, Christoph ; McNamara, Paul
Author_Institution :
Dept. of Electr., Univ. of Ontario Inst. of Technol., Oshawa, ON, Canada
Abstract :
Amplitude-integrated electroencephalography (aEEG), a time compression technique, compresses the time scale of the conventional electroencephalogram (EEG) which is advantageous for presenting long EEG recordings. Cerebral function monitors use a reduced number of electrodes from the standard 10-20 montage and displays both an EEG and aEEG trace from one or two channels. Sleep-wake cycling is defined as a state of continuous normal voltage and the presence of both wakefulness or active sleep and quiet sleep with a minimum of two or three consecutive sleep state changes on aEEG for a duration of twenty minutes during a three-four hour period. Sleep-wake cycling in infants is often used as an indicator of the patient´s neurological development and response to brain injury. There has been complete absence of algorithm development for the automated detection of changes in normal neonatal sleep-wake cycling patterns displayed by cerebral function monitor. This study will incorporate a unique, robust algorithm for incorporation into a multidimensional data analysis environment, which is capable of capturing multiple individual streams of physiological data from bedside monitors and processing them. The framework supports the acquisition, collection, transmission, real-time processing, storage and retrospective analysis of wave form and physiological data streams combined with supporting clinical information including clinical observations and the results of laboratory investigations. The development of a robust, accurate and reliable algorithm that detects sleep-wake cycling in newborn infants greater than 29 weeks gestational age should provide valuable information for meaningful clinical decision support for physicians caring for critically ill neonates. This paper describes the development and results of the algorithm, its upper and lower boundary detection, low pass filtration, and threshold classification on an individual patient´s data set.
Keywords :
data analysis; data compression; electroencephalography; low-pass filters; medical signal processing; neurophysiology; paediatrics; signal classification; aEEG trace; amplitude-integrated electroencephalography; bedside monitors; boundary detection; cerebral function monitor signals; clinical decision support; clinical information; critically ill neonates; electroencephalogram; gestational age; low pass filtration; multidimensional data analysis; neonatal sleep-wake cycling patterns; neonates automated sleep-wake detection; physiological data stream processing; threshold classification; time compression technique; Algorithm design and analysis; Biomedical monitoring; Electroencephalography; Monitoring; Pediatrics; Real-time systems; Sleep; amplitude-integrated electroencephalography; cerebral function monitor; sleep-wake detection;
Conference_Titel :
Computer-Based Medical Systems (CBMS), 2014 IEEE 27th International Symposium on
Conference_Location :
New York, NY
DOI :
10.1109/CBMS.2014.36