Dr Denis A. Engemann | Large-scale analysis of electrophysiology data in cognitive neurology

Guest Lecture

  • Date: Nov 4, 2019
  • Time: 03:00 PM - 04:00 PM (Local Time Germany)
  • Speaker: Dr Denis A. Engemann
  • French National Institute of Computer Science (INRIA), Saclay, France
  • Location: MPI for Human Cognitive and Brain Sciences
  • Room: Wilhelm Wundt Room (A400)
Magnetoencephalography and electroencephalography (M/EEG) access population-level neuronal dynamics across diverse temporal scales from seconds to milliseconds. This makes M/EEG a promising technology for the study of brain function, cognition and neurological conditions. Yet, processing M/EEG data is inherently challenging due to its high-dimensional nature and low signal-to-noise ratio. Unfortunately, many common elements of M/EEG analysis have not been fully automated. This incurs high costs in terms of human processing time, makes research findings less repeatable and potentially less applicable in the clinic. In this lecture, I will present our research focusing on machine learning as a tool to enhance analysis of M/EEG at different levels.

My talk will introduce common methodological issues with machine learning in clinical neuroscience [background a-c]. I will then present two clinical applications in which we tackle the challenge of EEG-based diagnosis in severely brain injured patients [1] and explore between-subjects variability in MEG signals from early blind individuals [2] with machine learning. Finally, I will present machine learning solutions for three recurrent problems in large-scale analysis of M/EEG: i) cleaning data [3], ii) hyper-parameter selection in source modeling [4], iii) cross-person regression for biomarker discovery when anatomical source modeling is not available [5]. I will conclude with some remarks on how these issues drive MNE software [https://mne-tools.github.io/stable/index.html] development and highlight related tools and resources [background d-f].

References

[1] *Engemann*, Raimondo, King, Rohaut, Louppe, Faugeras, Annen, Cassol, Gosseries, Fernandez-Slezak, Laureys, Naccache, Dehaene and Sitt. Robust EEG-based cross-site and cross-protocol classification of states of consciousness (2018). Brain 141 (11), 3179–3192, https://doi.org/10.1093/brain/awy251

[2] Abboud, *Engemann*, Cohen (biorXiv). Semantic coding in the occipital cortex of early blind individuals, https://doi.org/10.1101/539437

[3] Jas, *Engemann*, Bekhti, Raimondo, and Gramfort. "Autoreject: Automated artifact rejection for MEG and EEG data." NeuroImage 159 (2017), 417-429.

[4] *Engemann* and Gramfort. "Automated model selection in covariance estimation and spatial whitening of MEG and EEG signals." NeuroImage 108 (2015): 328-342.

[5] Sabbagh, Ablin, Varoquaux, Gramfort, *Engemann*. Manifold-regression to predict from MEG/EEG brain signals without source modeling. Preprint (accepted for NeurIPS 2019) arXiv:1906.02687v2,https://arxiv.org/abs/1906.02687

Background

[a] Varoquaux, Raamana, *Engemann*, Hoyos-Idroboa, Schwartz, Thirion (2016). Assessing and tuning brain decoders: Cross-validation, caveats, and guidelines. Neuroimage (http://www.sciencedirect.com/science/article/pii/S105381191630595X)

[b] Bzdok, *Engemann*, Grisel, Varoquaux, Thirion (biorXiv). Prediction and inference diverge in biomedicine: Simulations and real-world data (https://www.biorxiv.org/content/early/2018/05/21/327437)

[c] Woo, Chang, Lindquist & Wager (2017) Building better biomarkers: brain models in translational neuroimaging. Nature Neurosciene (http://www.nature.com/neuro/journal/v20/n3/full/nn.4478.html)

[d] Gramfort, Luessi, Larson, *Engemann*, Strohmeier, Brodbeck, Parkkonen and Hämäläinen. "MNE software for processing MEG and EEG data." Neuroimage 86 (2014): 446-460.

[e] Jas, Larson, *Engemann*, Leppakangas, Taulu, Hämäläinen and Gramfort. A Reproducible MEG/EEG Group Study With the MNE Software: Recommendations, Quality Assessments, and Good Practices. Frontiers in neuroscience (2018) 12.

[f] Esch, Dinh, Larson, *Engemann*, Jas, Khan, Gramfort, Hämäläinen. MNE: Software for Acquiring, Processing, and Visualizing MEG/EEG Data. https://rd.springer.com/referenceworkentry/10.1007/978-3-319-62657-4_59-1
Go to Editor View