Woman sleeping

Design of Interpretable end-to-end Deep Learning Models for Diagnosis of Sleep Disorders and Sleep Quality Evaluation

Background

Rapid eye movement (REM) sleep behavior disorder (RBD) is a parasomnia, a type of sleep disorder which involves abnormal movement and behavior during sleep. Specifically, RBD is characterized by abnormal behavior during the REM sleep phase, in which individuals physically enact their dreams, especially dreams that are violent and confrontational [1]. The abnormal behavioral release is facilitated by the loss of skeletal muscle paralysis in REM sleep, thus RBD subjects can move and talk in response to the virtual environment of their dream while disregarding their actual surroundings.
A recent study estimated from a population-based sample that the prevalence of RBD is 1.06 % with gender parity [2]. Patients with idiopathic RBD (iRBD) are in a prodromal stage of neurodegeneration, and a precursor of Parkinson’s disease (PD) [3, 4]. The conversion rate to an overt neurodegenerative disease from idiopathic RBD was found to be 6.3 % per year, with 73.5 % converting after a 12-year follow-up [4]. Once reliable neuroprotective agents are available, early diagnosis of RBD and an accurate description of the phenotype that develops overt neurodegenerative disease would enable prevention or slow down the progression of neurodegeneration.
Diagnosis of RBD necessitates sleep analysis with polysomnography (PSG), which includes measurements of several physiological signals that describes sleep patterns and regulation. A PSG recording involves measuring brain activity, eye movements, muscle activity, blood oxygen saturation, and breathing patterns. A technician can now visually analyze sleep patterns [1, 5]; however, this is highly time-consuming and expensive. Therefore, there is a need for efficient methods that are less expensive.

 

Project Objectives

The objectives of the Ph.D. project revolve around the automatic diagnosis of RBD from recorded PSG signals. We will invent an automatic system based on deep learning methods to classify RBD in PSG recordings. An important aim for the project is that the system is interpretable, as the system is intended to be used as an assisting tool in clinical settings. The system should, therefore, be able to highlight the signal regions that attribute to the diagnostic decision.

 

References

  1. American Academy of Sleep Medicine., International classification of sleep disorders, 3 ed., Darien, IL: American Academy of Sleep Medicine, 2014, p. 383.
  2. J. Haba-Rubio, B. Frauscher, P. Marques-Vidal, J. Toriel, N. Tobback, D. Andries, M. Preisig, P. Vollenweider, R. Postuma and R. Heinzer, "Prevalence and determinants of rapid eye movement sleep behavior disorder in the general population," Sleep, vol. 41, no. 2, 1 2 2018.
  3. B. F. Boeve and M. W. Mahowald, "Delayed emergence of a parkinsonian disorder or dementia in 81% of older men initially diagnosed with idiopathic rapid eye movement sleep behavior disorder: a 16-year update on a previously reported series," Sleep Medicine, vol. 14, no. 8, pp. 744-748, 1 8 2013.
  4. R. B. Postuma, A. Iranzo, M. Hu, B. Högl, B. F. Boeve, R. Manni, W. H. Oertel, I. Arnulf, L. Ferini-Strambi, M. Puligheddu, E. Antelmi, V. Cochen De Cock, D. Arnaldi, B. Mollenhauer, A. Videnovic, K. Sonka, K.-Y. Jung, D. Kunz, Y. Dauvilliers, F. Provini, S. J. Lewis, J. Buskova, M. Pavlova, A. Heidbreder, J. Y. Montplaisir, J. Santamaria, T. R. Barber, A. Stefani, E. K. St.Louis, M. Terzaghi, A. Janzen, S. Leu-Semenescu, G. Plazzi, F. Nobili, F. Sixel-Doering, P. Dusek, F. Bes, P. Cortelli, K. Ehgoetz Martens, J.-F. Gagnon, C. Gaig, M. Zucconi, C. Trenkwalder, Z. Gan-Or, C. Lo, M. Rolinski, P. Mahlknecht, E. Holzknecht, A. R. Boeve, L. N. Teigen, G. Toscano, G. Mayer, S. Morbelli, B. Dawson and A. Pelletier, "Risk and predictors of dementia and parkinsonism in idiopathic REM sleep behaviour disorder: a multicentre study," Brain, vol. 142, no. 3, pp. 744-759, 1 3 2019.
  5. R. Berry, C. L. Albertario, S. M. Harding, R. M. Lloyd, D. T. Plante, S. F. Qyan, T. M. M. and B. V. Vaughn, The AASM Manual for the scoring of sleep and associated events : rules, terminology and technical specifications, Version 2. ed., Darien, IL: American Academy of Sleep Medicine: American Academy of Sleep Medicine, 2018.

Photo by Gregory Pappas on Unsplash

Contact

Andreas Brink-Kjær
PhD student
DTU Health Tech

Contact

Helge Bjarup Dissing Sørensen
Groupleader, Associate Professor MSK, PhD
DTU Health Tech
+45 45 25 52 44

Contact

Poul Jørgen Jennum
Professor
Rigshospitalet, Glostrup

Contact

Emmanuel Mignot
Professor of Psychiatry and Behavioral Sciences
Stanford University
https://www.cachet.dk/research/phd-projects/design-of-interpretable-end-to-end-deep-learning-models-for-diagnosis-of-sleep-disorders-and-sleep-q
1 OCTOBER 2020