Performance Analysis of a Source-Space Low-Density EEG-Based Motor Imagery BCI

Gurgen Soghoyan, Nikolai Smetanin, Mikhail Lebedev, Alexei Ossadtchi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


Brain-computer interfaces are considered as the next level of human-machine interaction. A bunch of approaches in decoding human states aims to achieve sufficient precision and accommodate a growing number of distinct states to decode. The following study investigates the capabilities of the EEG-based inverse modelling to improve the classification accuracy and provides a comparison between different inverse models. The computational pipeline of represented BCI includes clustering and dimension reduction of the forward model. The obtained results show the advantages of minimal norm estimate (MNE) inverse operator in comparison to the Beamformer, sLORETA. We have also observed that a motor imagery BCI based on the fully blown individual inverse model outperformed that based on Riemann geometry-based approaches, while the latter demonstrated performance superior to the approaches using band specific sensor space power distribution. The performance analysis was done using a 32 channel EEG data recorded during motor imagery of the four limbs.

Original languageEnglish
Title of host publicationAdvances in Cognitive Research, Artificial Intelligence and Neuroinformatics - Proceedings of the 9th International Conference on Cognitive Sciences, Intercognsci-2020
EditorsBoris M. Velichkovsky, Pavel M. Balaban, Vadim L. Ushakov
PublisherSpringer Science and Business Media Deutschland GmbH
Number of pages5
ISBN (Print)9783030716363
Publication statusPublished - 2021
Externally publishedYes
Event9th International Conference on Cognitive Sciences, Intercognsci 2020 - Moscow, Russian Federation
Duration: 10 Oct 202016 Oct 2020

Publication series

NameAdvances in Intelligent Systems and Computing
Volume1358 AIST
ISSN (Print)2194-5357
ISSN (Electronic)2194-5365


Conference9th International Conference on Cognitive Sciences, Intercognsci 2020
Country/TerritoryRussian Federation


  • BCI
  • Beamformer
  • Brain-computer interface
  • Brain-machine interface
  • EEG inverse problem
  • EEG Riemannian approach
  • Low-density EEG
  • Minimal norm estimate
  • Motor imagery BCI


Dive into the research topics of 'Performance Analysis of a Source-Space Low-Density EEG-Based Motor Imagery BCI'. Together they form a unique fingerprint.

Cite this