Deep Learning Based Inter-subject Continuous Decoding of Motor Imagery for Practical Brain-Computer Interfaces

Roy, Sujit; Chowdhury, Anirban; McCreadie, Karl; Prasad, Girijesh

doi:10.3389/fnins.2020.00918

Cited by 40 publications

(28 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Li et al [59] employed the continuous wavelet transform (CWT) and a simplified convolutional neural network (SCNN), achieving ACC of 83.2% for a subject-specific BCI. Roy et al [60] transformed the EEG epochs into the time-frequency representation by applying short-time Fourier transform (STFT), which were applied in sequence as images to CNN architectures ACC of 77.46% for an intra-subject crosssession validation, and ACC of 70.94% for a cross-subject transfer learning were obtained. Sun et al [61] extracted dominant spectral EEG features by applying the spectrotemporal decomposition method (SSD-SE-CNN), achieving ACC of 79.3%.…”

Section: Sakhavi and Guanmentioning

confidence: 99%

“…Experiment #4. We evaluated an inter-subject transfer learning strategy [60]. In contrast to Experiments #2 and #3, the accuracy for a specific subject in the dataset 2a (or dataset 2b) was obtained here by applying the pretrained model over his/her full dataset.…”

Section: Evaluation and Statistical Analysismentioning

confidence: 99%

“…Table VII shows the results obtained on the dataset 2b. Here, we only compared our results with previous works that used all training sessions, such as CNN-SAE [57], CWT-SCNN [59] and CAgross [60]. For this reason, the system proposed in [38] was excluded for this evaluation.…”

Section: B Experimentsmentioning

confidence: 99%

“…For this reason, the system proposed in [38] was excluded for this evaluation. Also notice that the best results in [60] were not considered in Table VII for comparison, as they were obtained in other conditions. For this reason, we show the results obtained in [60] by using CAgross.…”

Section: B Experimentsmentioning

confidence: 99%

“…Also notice that the best results in [60] were not considered in Table VII for comparison, as they were obtained in other conditions. For this reason, we show the results obtained in [60] by using CAgross. We observed that our method outperformed the results obtained by applying CNN-SAE and CAgross, but CWT-SCNN performed better.…”

Section: B Experimentsmentioning

confidence: 99%

See 4 more Smart Citations

Shallow Convolutional Network Excel for Classifying Motor Imagery EEG in BCI Applications

Hermosilla

Codorniú

Baracaldo³

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Many studies applying Brain-Computer Interfaces (BCIs) based on Motor Imagery (MI) tasks for rehabilitation have demonstrated the important role of detecting the Event-Related Desynchronization (ERD) to recognize the user's motor intention. Nowadays, the development of MI-based BCI approaches without or very few calibration stages session-by-session for different days or weeks is still an open and emergent scope. In this work, a new scheme is proposed by applying Convolutional Neural Networks (CNN) for MI classification, using an end-to-end Shallow architecture that contains two convolutional layers for temporal and spatial feature extraction. We hypothesize that a BCI designed for capturing event-related desynchronization/synchronization (ERD/ERS) at the CNN input, with an adequate network design, may enhance the MI classification with less calibration stage. The proposed system using the same architecture was tested on three public datasets through multiple experiments, including both subject-specific and nonsubject-specific training. Comparable and also superior results with respect to the state-of-the-art were obtained. On subjects whose EEG data were never used in the training process, our scheme also achieved promising results with respect to existing non-subject-specific BCIs, which is other progress to facilitate the clinical application.

show abstract

Section: Sakhavi and Guanmentioning

confidence: 99%

Section: Evaluation and Statistical Analysismentioning

confidence: 99%

Section: B Experimentsmentioning

confidence: 99%

Section: B Experimentsmentioning

confidence: 99%

Section: B Experimentsmentioning

confidence: 99%

See 3 more Smart Citations

Shallow Convolutional Network Excel for Classifying Motor Imagery EEG in BCI Applications

Hermosilla

Codorniú

Baracaldo³

et al. 2021

IEEE Access

View full text Add to dashboard Cite

show abstract

Use of covariance matrix images for electroencephalography signal classification for multiclass motor imagery‐based brain computer interface

Thrikkannoor Kolathod,

Sanjay

2023

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Electroencephalogram (EEG) signals can be used to control devices using brain computer interface (BCI) systems, which offer assistance to patients with neuromuscular diseases. This article aims to determine the feasibility of using spatial covariance matrices with deep convolutional neural networks for multiclass motor imagery (MI) task classification for EEG‐based BCIs. Covariance matrices derived from multichannel EEG data are converted into color‐scaled images, which are used to train the AlexNet neural network using transfer learning. Samples of 2‐s duration are extracted every 200 ms intervals from each MI activity EEG eventually increasing the training data size sixfold and reducing the latency in real‐time implementations. Observed the performance using two popular publicly available BCI datasets for two‐class and four‐class MI classification. BCI competition IV Dataset IIa has shown average accuracy of 93.99% (k = 0.88) for two‐class and 86.70% (k = 0.82) for four‐class MI classification. BCI competition III Dataset IVa exhibited average accuracy of 97.02% (k = 0.94) for two‐class classification. Good performance was observed for training with cross‐subject data for two‐class and four‐class classification. The proposed method has shown improved accuracy for multiclass MI classification compared with existing works. The CNN is able to extract features from the covariance matrix and classify the MI data with reasonable accuracy. Since this method can handle large‐dimension data, dimension reduction, and source separation techniques are not required.

show abstract

Deep learning techniques for classification of electroencephalogram (EEG) motor imagery (MI) signals: a review

Altaheri

Muhammad

Alsulaiman

et al. 2021

Neural Comput & Applic

205

View full text Add to dashboard Cite

Deep Learning Based Inter-subject Continuous Decoding of Motor Imagery for Practical Brain-Computer Interfaces

Cited by 40 publications

References 63 publications

Shallow Convolutional Network Excel for Classifying Motor Imagery EEG in BCI Applications

Shallow Convolutional Network Excel for Classifying Motor Imagery EEG in BCI Applications

Use of covariance matrix images for electroencephalography signal classification for multiclass motor imagery‐based brain computer interface

Deep learning techniques for classification of electroencephalogram (EEG) motor imagery (MI) signals: a review

Contact Info

Product

Resources

About