Deep Learning Approach for Automatic Classification of Ocular and Cardiac Artifacts in MEG Data

Abstract

We propose an artifact classification scheme based on a combined deep and convolutional neural network (DCNN) model, to automatically identify cardiac and ocular artifacts from neuromagnetic data, without the need for additional electrocardiogram (ECG) and electrooculogram (EOG) recordings. From independent components, the model uses both the spatial and temporal information of the decomposed magnetoencephalography (MEG) data. In total, 7122 samples were used after data augmentation, in which task and nontask related MEG recordings from 48 subjects served as the database for this study. Artifact rejection was applied using the combined model, which achieved a sensitivity and specificity of 91.8% and 97.4%, respectively. The overall accuracy of the model was validated using a cross-validation test and revealed a median accuracy of 94.4%, indicating high reliability of the DCNN-based artifact removal in task and nontask related MEG experiments. The major advantages of the proposed method are as follows: (1) it is a fully automated and user independent workflow of artifact classification in MEG data; (2) once the model is trained there is no need for auxiliary signal recordings; (3) the flexibility in the model design and training allows for various modalities (MEG/EEG) and various sensor types.

Decoding Imagined and Spoken Phrases From Non-invasive Neural (MEG) Signals

“…Nevertheless, the efficacy of CNN for neural data analysis can be translated for neural speech decoding which we experimented within this study. Moreover, a few recent studies have shown the efficacy of using CNN to analyze MEG (Hasasneh et al, 2018;Dash et al, 2019a;Huang and Yu, 2019) or EEG data (Cooney et al, 2019a,b), which further strengthens our motivation for this approach. To our knowledge, this is the first study using CNNs to explore neural speech decoding with MEG.…”

Automatic detection of artifacts in EEG by combining deep learning and histogram contour processing

“…The method was shown to be suitable in detecting artifacts from recordings collected with ICU-suitable electrode and device. The obtained results showed an overall accuracy of 0.98, representing high reliability of proposed technique in detecting different types of artifacts and being comparable or outperforming the approaches proposed earlier in the literature reaching the accuracy of 67.59 % for detecting four kinds of artifacts [18], and a median accuracy of 94.4 % for classifying ocular and cardiac artifacts [19].…”

“…In recent years, machine learning methods have been increasingly used for discriminating artifact-free EEG sequences from contaminated ones [12][13][14][15][16][17]. So far, there are only few methods in literature that address a fully automatic removal approach using deep learning on EEG data [18][19][20]. These studies focus on detecting specific types of artifacts which makes them hard to be generalized to cover artifacts resulting from different sources.…”

Automatic Removal of Cardiac Interference (ARCI): A New Approach for EEG Data

“…One issue that should be addressed is the presence of cardiac-related interference in MEG data. Cardiac artifacts contaminate not only EEG, but also present a serious concern in MEG data acquisition and processing (Escudero et al, 2007; Escudero et al, 2011; Breuer et al, 2014a,b; Hasasneh et al, 2018). Indeed, electrical cardiac activity can create greater interference in MEG acquisitions where the insulating properties of the scull do not attenuate the interference (Hämäläinen et al, 1993).…”