[Zurück]


Diplom- und Master-Arbeiten (eigene und betreute):

S. Schreiner:
"Automatic detection of sleep arousals and investigation on the relation with leg movements";
Betreuer/in(nen): W. Scherrer; Institut für Stochastik und Wirtschaftsmathematik, 2017; Abschlussprüfung: 2017.



Kurzfassung englisch:
This thesis provides a method for the automatic detection of arousals in EEG-signals as well as an analysis of the possible relation between leg movements and arousals. The detection algorithm consists of two main steps: Firstly, the recording is divided in segments of three seconds and these are classified with a Support Vector Machine (SVM) as arousal start or nostart segments. The features extracted for each segment are mainly frequency based according to the definition of an arousal of the AASM American Academy of Sleep Medicine. In addition, newly investigated feature sets, based on an AR-models and statistical tests, are implemented. Secondly, the exact position and length of the detections are computed and further arousal criteria are checked. Moreover, an analysis on the relation between arousals and leg movements was performed. It could be shown that the events are dependent from each other, but that the relation follows a complex mechanism rather than a simple causality. In addition, the relation between the intensity of a leg movement and the occurrence of an associated arousal was investigated. As a measure for intensity, the duration of a leg movement and a value computed from 3D detection that could be interpreted as the size of a leg movement, were tested. An interesting finding was that with both intensity values it could be shown that more intense leg movements are more likely to occur with arousals.

Schlagworte:
Polysomnographie / Schlafstörungen / Zeitreihenanalyse / Statistik

Erstellt aus der Publikationsdatenbank der Technischen Universitšt Wien.