Classifying Sleep Stages with Heart Rate Variability and Resting State fMRI

MICHAEL MARXEN; MARKO LEHMANN; ENZO TAGLIAZUCCHI; HELMUT LAUFS; HAGEN MALBERG; SEBASTIAN ZAUNSEDER

Hamburgo

Congreso; Organization for Human Brain Mapping; 2014

For decades, different sleep stages have been distinguished based on Polysomnography, primarily electroencephalography (EEG). Following the American Academy of Sleep Medicine (AASM) four sleep stages can be defined: N1-3 & REM. It has been studied previously whether data from other modalities such as fMRI (Tagliazucchi, 2012) or electrocardiography (ECG) (e.g. Bsoul, 2010) may be used to classify sleep according to these stages using manual AASM scores as a gold standard. fMRI is known to provide sleep stage specific functional connectivity measures while ECG is able to detect differences in autonomic regulation. We are presenting here a novel approach of classification based on both fMRI and ECG based features which may benefit from the combination of these functional dimensions and potential interactions. Our hypothesis is that a combined classifier will outperform classifiers based on either modality alone.