Contextual MEG and EEG Source Estimates Using Spatiotemporal LSTM Networks

Christoph Dinh; John G Samuelsson; Alexander Hunold; Matti S Hämäläinen; Sheraz Khan

doi:10.3389/fnins.2021.552666

Contextual MEG and EEG Source Estimates Using Spatiotemporal LSTM Networks

Front Neurosci. 2021 Mar 9:15:552666. doi: 10.3389/fnins.2021.552666. eCollection 2021.

Authors

Christoph Dinh^{1

2

3}, John G Samuelsson^{1

4

5}, Alexander Hunold⁶, Matti S Hämäläinen^{1

2

4}, Sheraz Khan^{1

2

4}

Affiliations

¹ Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States.
² Department of Radiology, Massachusetts General Hospital (MGH), Charlestown, MA, United States.
³ Institute for Medical Engineering, Research Campus STIMULATE, Otto-von-Guericke University, Magdeburg, Germany.
⁴ Harvard Medical School, Boston, MA, United States.
⁵ Harvard-MIT Division of Health Sciences and Technology (HST), Massachusetts Institute of Technology, Cambridge, MA, United States.
⁶ Institute of Biomedical Engineering and Informatics, Technische Universität Ilmenau, Ilmenau, Germany.

Abstract

Most magneto- and electroencephalography (M/EEG) based source estimation techniques derive their estimates sample wise, independently across time. However, neuronal assemblies are intricately interconnected, constraining the temporal evolution of neural activity that is detected by MEG and EEG; the observed neural currents must thus be highly context dependent. Here, we use a network of Long Short-Term Memory (LSTM) cells where the input is a sequence of past source estimates and the output is a prediction of the following estimate. This prediction is then used to correct the estimate. In this study, we applied this technique on noise-normalized minimum norm estimates (MNE). Because the correction is found by using past activity (context), we call this implementation Contextual MNE (CMNE), although this technique can be used in conjunction with any source estimation method. We test CMNE on simulated epileptiform activity and recorded auditory steady state response (ASSR) data, showing that the CMNE estimates exhibit a higher degree of spatial fidelity than the unfiltered estimates in the tested cases.

Keywords: EEG; LSTM; MEG; deep learning; grid-based Markov localization; source estimation; spatial filtering; spatiotemporal source estimation.

Abstract

Grants and funding