Skip to content
/ hear Public

high-variance electrode artifact removal (HEAR) algorithm

License

Notifications You must be signed in to change notification settings

rkobler/hear

Repository files navigation

high-variance electrode artifact removal (HEAR) algorithm

HEAR is a simple, yet efficient algorithm to remove transient, high-variance artifacts from multivariate time-series signals (e.g., electroencephalographic (EEG) signals).

Reference:

Kobler, R. J., Sburlea, A. I., Mondini, V. & Müller-Putz, G. R. HEAR to remove pops and drifts: the high-variance electrode artifact removal (HEAR) algorithm. In Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, 2019 accepted version

HEAR can be applied offline and online. The repository contains a reference implementation in Matlab and a dataset of simulated EEG signals. The demonstration dataset is stored in the eeglab format.

Getting Started

  • Download HEAR and open the downloaded folder.
  • Startup the eeglab toolbox:
  • Open the train_HEAR.m script. The script loads a calibration dataset (demo_simrest.set) that contains simulated artifact-free EEG signals. Then HEAR is fit to the data. The parameter is_causal defines if HEAR should be used online is_causal = true or offline is_causal = false. The calibrated model is stored to the disk 'hear_mdl.mat'.
  • The script apply_HEAR.m uses the calibrated model to correct pop and drift artifacts in a second dataset (demo_simreach.set).

Contact

Feel free to contact me at reinmar.kobler@tugraz.at.

Acknowledgements

This work was supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (Consolidator Grant 681231 'Feel Your Reach').