THALAMO-CORTICAL SYNCHRONIZATION DURING ACOUSTIC STIMULATION MAY BE SPECIFIC TO PHASIC REM SLEEP M. Czisch, R. Wehrle, C. Kaufmann, T. C. Wetter, T. Pollmächer, D. P. Auer Max Planck Institute of Psychiatry, Kraepelinstr. 10, 80804 Munich czisch@mpipsykl.mpg.de INTRODUCTION An apparently reduced reactivity to external stimuli is a basic feature of sleep, and the changes of information processing during sleep, especially in stage REM with intensified mental activity, are still to be disclosed. Making use of the novel approach of combined EEG and fMRI investigations in sleep 1-3 , the objective of the present study was to explore regional activation patterns and functional connectivities upon different kinds of acoustic stimulation during wakefulness, nonREM sleep and REM sleep. METHODS To enhance REM-sleep probability, the volunteers underwent partial sleep deprivation two days prior to the study night. After positioning of the MR compatible EEG (frontal, central, parietal and occipital), EOG, EMG and ECG electrodes, two fMRI sessions of 2-3 hours at 23:00 and 03:00 were performed at 1.5T (GE Signa Echospeed) while continuously presenting scanner noise via headphones. For fMRI a multi-slice EPI sequence (7 slices, 5 mm thick, 96x96 matrix, resolution 2 mm, TE=60 ms, TR=3 s) parallel to the AC-PC line was used with additional acoustic stimuli presented in a block design fashion (3 on, 4 off periods of 30 sec each). During postprocessing, EEG artefacts due to the imaging gradients of the fMRI experiments were removed by Fourier filtering allowing unambigous classification of all sleep stages. Once unambiguous sleep stages were established in the online polysomnography for at least one minute, scans were started applying one of the following stimuli: a narrative text, the same text in reverse mode, a 1 kHz sinus beep (0.5 s length, 1 s repetition) and piano music. Results of 8 subjects during nonREM sleep and one subject during REM sleep are presented. fMRI experiments were analysed using SPM99. Functional connectivity maps were calculated using the thalamic response as regressor. RESULTS After EEG post-processing and sleep staging, scans with apparent arousals or changes in sleep stages during the 3.5 min block duration were discarded. Dataset with movements larger than 2mm were excluded. During wakefulness acoustic stimulation leads to a stable activation of the auditory cortex. In nonREM sleep, this response is drastically reduced or even inverted. In addition, in some experiments a negative BOLD signal can be observed in the visual cortex, precuneus and cingulum with both extend and amplitude being correlated to the number of K-complexes and the delta power in the EEG. The thalamus does not show a strong response. During stage REM, activation maps were highly variable even in repeated