EEG artifact detection and removal
In collaboration with Andrea Mognon, Jorge Jovicich (CIMeC, University of Trento) and Lorenzo Bruzzone (Dipartimento di Ingegneria e Scienza dell'Informazione, University of Trento) , I have developed ADJUST, a fully automatic algorithm to detect and remove artifacted components from EEG data. The algorithm has been implemented as a plugin of EEGLAB, the most popular software for EEG data analysis, and is available on the NIH Neuroimaging Tools and Resources Platform. The associated article (Mognon et al., 2011) has been cited more than 1000 times.
ADJUST classification of artefacted ICs (Mognon et al., 2011)
Examples of typical ICs from each artifact class and of a typical neural IC. Top row: IC topography. Middle row: ERP image illustrating the color-coded amplitude fluctuations of the IC in 100 contiguous epochs. Bottom row: histograms of feature values normalized by the corresponding automatically calculated threshold value. Bars of features belonging to the same artifact class are grouped together, and are marked in red color if they all cross the threshold, indicating that ADJUST classifies the IC as a component of that artifact class.
In collaboration with Velu Kumaravel (PhD at CIMeC and FBK), Elisabetta Farella (FBK) and Eugenio Parise (CIMeC), we have developed NEAR, a novel pipeline for artifact removal in newborn and infant EEG data. See the related articles (Kumaravel et al., 2022a,b) and NEAR open source software, including a sample of newborn EEG data to test it.
NEAR performance in cleaning EEG artifacts (Kumaravel et al., 2022a)
Comparison of the performance of artifact removal between the pipelines NEAR, MADE and HAPPE on simulated data of a frequency-tagging paradigm.
We also developed a similar version of the same software, based on ASR, to devise an efficient artifact removal process for low-density, portable EEG systems (Kumaravel et al., 2021a). ASR parameter calibration on publicly available EEG dataset also provided positive results (Kumaravel et al., 2021b).
References:
Mognon A, Jovicich J, Bruzzone L, Buiatti M,
ADJUST: An Automatic EEG artifact Detector based on the Joint Use of Spatial and Temporal features,
Psychophysiology 48 (2), 229-240 (2011).
Kumaravel VP, Farella E, Parise E, Buiatti M,
NEAR: An artifact removal pipeline for human newborn EEG data.
Developmental Cognitive Neuroscience, 101068 (2022)
Kumaravel VP, Buiatti M, Parise E, Farella E,
Adaptable and Robust EEG Bad Channel Detection Using Local Outlier Factor (LOF).
Sensors 22 (19), 7314 (2022)
Kumaravel VP, Kartsch V, Benatti S, Vallortigara G, Farella E, Buiatti M,
Efficient Artifact Removal from Low-Density Wearable EEG using Artifacts Subspace Reconstruction,
43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 333-336 (2021).
Kumaravel VP, Buiatti M, Farella E,
Hyperparameter selection for reliable EEG denoising using ASR: a benchmarking study.
IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 3638-3641 (2021).