Deep Learning to EEG and EEG Connectivity Analysis
Presented by: Arnaud Delorme, PhD
Deep Learning EEG & EEG Connectivity
〰️
Deep Learning EEG & EEG Connectivity 〰️
Deep learning applied to EEG and EEG connectivity analysis are two recent computational methods applied to EEG analysis. Deep Learning has achieved impressive performance on many tasks, such as object detection, image recognition, image retrieval, etc. These achievements benefit from the Deep Learning outstanding capability to learning discriminative features. This has inspired the EEG research community to adopt DL. However, DL learned features are not immediately interpretable. Understating DL features could lead to new biomarkers for EEG and qEEG. This presentation will explain in plain language where the field of DL applied to EEG is and how it may benefit in the future the field of EEG and qEEG.
The second part of this presentation will present a method for EEG connectivity analysis. These methods have the potential to unravel the brain dynamics associated with specific mental states or clinical conditions. Understanding change in cortical dynamics may inform treatment strategies, but also potentially create new targets for neurofeedback training. For example, one may target the upregulation of connectivity between brain areas in a specific frequency band. Because EEG is a noisy signal, and because of the difficulty in extracting brain sources from EEG, this is a challenging field. This presentation will review challenges and potential future development for EEG and qEEG.
Using these tools, EEG researchers using Deep Learning can better identify the learned EEG features, possibly identifying new class relevant biomarkers.