English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Motor brain-computer interfaces (BCIs) are a promising technology that mayenable motor-impaired people to interact with their environment. Designingreal-time and accurate BCI is crucial to make such devices useful, safe, andeasy to use by patients in a real-life environment. Electrocorticography(ECoG)-based BCIs emerge as a good compromise between invasiveness of therecording device and good spatial and temporal resolution of the recordedsignal. However, most ECoG signal decoders used to predict continuous handmovements are linear models. These models have a limited representationalcapacity and may fail to capture the relationship between ECoG signal andcontinuous hand movements. Deep learning (DL) models, which arestate-of-the-art in many problems, could be a solution to better capture thisrelationship. In this study, we tested several DL-based architectures topredict imagined 3D continuous hand translation using time-frequency featuresextracted from ECoG signals. The dataset used in the analysis is a part of along-term clinical trial (ClinicalTrials.gov identifier: NCT02550522) and wasacquired during a closed-loop experiment with a tetraplegic subject. Theproposed architectures include multilayer perceptron (MLP), convolutionalneural networks (CNN), and long short-term memory networks (LSTM). The accuracyof the DL-based and multilinear models was compared offline using cosinesimilarity. Our results show that CNN-based architectures outperform thecurrent state-of-the-art multilinear model. The best architecture exploited thespatial correlation between neighboring electrodes with CNN and benefited fromthe sequential character of the desired hand trajectory by using LSTMs.Overall, DL increased the average cosine similarity, compared to themultilinear model, by up to 60%, from 0.189 to 0.302 and from 0.157 to 0.249for the left and right hand, respectively.

Decoding ECoG signal into 3D hand translation using deep learning