Decoding sleep stages from dynamic functional connectivity reveals signatures of REM & NREM sleep
Poster No:
1392
Submission Type:
Abstract Submission
Authors:
Maria Giulia Preti1, Célia Lacaux2, Nicolas Francio1, Isabelle ARNULF3, Stéphane Lehéricy3, Sophie Schwartz2, Delphine Oudiette3, Dimitri Van De Ville1
Institutions:
1Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland, 2University of Geneva, Geneva, Switzerland, 3Paris Brain Institute (ICM), Paris, France
First Author:
Co-Author(s):
Introduction:
Fluctuations of brain hemodynamic activity measured through functional magnetic resonance imaging (fMRI) recently showed to provide key signatures of consciousness alterations and sleep [1-3]. However, how well fMRI-derived functional connectivity (FC) can decode specific sleep stages remains to be explored. In addition, rapid eye movement (REM) sleep has been rarely recorded within an MRI scanner [4], due to its occurrence after roughly 90 minutes of uninterrupted sleep. In this work, we aim at using fMRI dynamic FC (dFC) features to classify sleep stages, including REM. This will provide insightful information at high spatial resolution on brain network patterns involved in different phases of sleep. In the future, a successful classification could also ameliorate conventional electroencephalography (EEG)-based sleep scoring, now time-consuming and extremely difficult when EEG is recorded in the MRI setting.
Methods:
18 narcoleptic patients were included in this study. This specific population was chosen for: i) propensity to easily fall asleep under challenging conditions, due to excessive daytime sleepiness; ii) facility to rapidly reach REM sleep. Each participant had different 30-min nap sessions during the day (typically 3), undergoing combined EEG-fMRI. A total of 42 nap sessions were analysed in this study. Wake-sleep stages were scored by 3 independent experts based on 30-sec EEG epochs, according to standard sleep scoring guidelines (American Academy of Sleep Medicine [5]). Each window was assigned to either wake, N1, N2, N3, REM or unknown class. fMRI timecourses were preprocessed with a standard pipeline [6] and averaged in 360 cortical [7] and 19 subcortical [8] atlas regions. dFC was computed as pairwise Pearson correlations in the same 30-sec windows scored for sleep. The upper triangular part of FC matrices was then vectorized and concatenated across windows and subjects, yielding the feature matrix used as input for classification. This was implemented with a support vector machine (SVM) classifier and 5-fold cross validation, adopting a Synthetic Minority Oversampling Technique (SMOTE) as data augmentation approach to account for imbalancedness. Classification performance was assessed in terms of F1-score, and classification weights were used to evaluate key brain regions for decoding different sleep stages.
Results:
Samples with no agreement between scorers, or classified as "unknown", were excluded from the classification analysis, yielding a dataset of N=1985 data points. Satisfactory classification results were obtained for all stages but N1 (see Tab. 1), with a macro F1-score of .72. Being a transitory period between wake and sleep, N1 also presents with the lowest inter-scorer agreement based on EEG features [9]. Our results indicate N1 as a mixture of dFC brain features, which makes its decoding harder. Classification weights suggest visual, temporal as well as prefrontal and thalamic areas to be key for distinguishing NREM states from wake (Fig. 1A-B), in accordance with previous literature reporting a decrease in brain activity in those areas during NREM phases, interpreted as homeostatic need for brain energy recovery [10]. More complex patterns of regions distinguish REM sleep from wake and NREM sleep (Fig. 1C-D), including as well language areas, middle temporal gyrus, anterior cingulate and inferior frontal junction. The role of these regions during REM sleep appears in line with previous EEG findings and with theories of REM sleep generation and dreaming properties [10].
Conclusions:
We investigated for the first time the capability of dFC to decode sleep stages. Our results shed light on functional networks characterizing different sleep stages, in particular REM phase, hard to record with fMRI data. Future studies considering temporal sequencing at the level of the extracted FC features or within more complex classification strategies will expand this analysis, to help optimizing the classification performance for all stages.
Modeling and Analysis Methods:
fMRI Connectivity and Network Modeling 1
Perception, Attention and Motor Behavior:
Sleep and Wakefulness 2
Keywords:
Electroencephaolography (EEG)
FUNCTIONAL MRI
Sleep
Other - Functional Connectivity
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
[2] A.B.A. Stevner, D. Vidaurre, J. Cabral, K. Rapuano, S. F. V. Nielsen, E. Tagliazucchi, H. Laufs, P. Vuust, G. Deco, M. W. Woolrich, E. Van Someren, M. L. Kringelbach, “Discovery of key whole-brain transitions and dynamics during human wakefulness and non-REM sleep,” Nat. Commun. 10, 1035, 2019.
[3] E. Tagliazucchi and H. Laufs, “Decoding Wakefulness Levels from Typical fMRI Resting-State Data Reveals Reliable Drifts between Wakefulness and Sleep,” Neuron, vol. 82, no. 3, pp. 695–708, 2014.
[4] R. Wehrle, C. Kaufmann, T. C. Wetter, F. Holsboer, D. P. Auer, T. Pollmächer, M. Czisch, “Functional microstates within human REM sleep: first evidence from fMRI of a thalamocortical network specific for phasic REM periods,” Eur J Neurosci, 25(3):863-71, 2007.
[5] M. Troester, S. Quan, R. Berry, D. T. Plante, A. R. Abreu, M. Alzoubaidi, “The AASM Manual for the Scoring of Sleep and Associated Events,” Rules, Terminology and Technical Specifications Version 3, 2023.
[6] M. G. Preti and D. Van De Ville, “Decoupling of brain function from structure reveals regional behavioral specialization in humans,” Nature Communications, vol. 10, no. 1, pp. 4747, dec 2019.
[7] M. F. Glasser, T. S. Coalson, E. C. Robinson, C. D. Hacker, J. Harwell, E. Yacoub, K. Ugurbil, J. Andersson, C. F. Beckmann, M. Jenkinson, S. M. Smith, D. C. Van Essen, “A multi-modal parcellation of human cerebral cortex,” Nature, vol. 536, no. 7615, pp. 171–178, 2016.
[8] M. F. Glasser, S. N. Sotiropoulos, J. A. Wilson, T. S. Coalson, B. Fischl, J. L. Andersson, J. Xu, S. Jbabdi, M. Webster, J. R. Polimeni, D. C. Van Essen, and M. Jenkinson, “The minimal preprocessing pipelines for the Human Connectome Project,” NeuroImage, vol. 80, pp. 105–124, oct 2013.
[9] R. S. Rosenberg, and S. Van Hout, “The American Academy of Sleep Medicine inter-scorer reliability program: sleep stage scoring,” J. Clin. Sleep Med. 9, 81–87, 2013.
[10] T. T. Dang-Vu, M. Schabus, M. Desseilles, V. Sterpenich, M. Bonjean, P. Maquet, “Functional neuroimaging insights into the physiology of human sleep,” Sleep, 33(12):1589-603, 2010.