A CNN based graphical user interface controlled by imagined movements

Kumar, Sandeep; Verma, Poonam Rani; Bharti, Manisha; Agarwal, Prabhakar

doi:10.1007/s13198-021-01096-w

Cited by 7 publications

(8 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Nguyen et al 10 visual cue was shown for 7 s during which the subjects had to imagine the vowels and a beep sound was given regularly after 1 s. Hundred such trials were recorded individually by 15 subjects and accuracy was calculated in a subject dependent framework. Kumar et al 23 6,7,10,11,24,25,48 The proposed work is subject-independent and presents a more generalized BCI. The imagined trials of all the subjects were mixed and then the accuracy was calculated.…”

Section: Resultsmentioning

confidence: 99%

“…They calculated powers in various EEG frequency bands using Morlet wavelet transform and then used CNN for classification. The aforementioned work was subject dependent and the authors recorded 1346 samples for one alphabet with an F s of 128 Hz, which means a time duration of approximately 10.51 s. The existing BCI models in the literature are based on individual subjects and hence the performance metrics have been calculated subject wise 6,7,10,11,24,25,48 . The proposed work is subject‐independent and presents a more generalized BCI.…”

Section: Resultsmentioning

confidence: 99%

“…The subject had to perform five MI tasks for the display of each alphabet which may be uncomfortable for ALS patients and also this method is time consuming. In a similar work on the same dataset, Kumar et al 25 has evaluated the display time on GUI of all the alphabets employing MI using CNN and XG Boost classifier. The display time of the alphabets varied between 10.9 and 36.82 s which is a considerable amount of time.…”

Section: Resultsmentioning

confidence: 99%

“…Due to these repetitive actions, the time consumed for the display of each alphabet is large. 25 BCI with SI can overcome the above challenges. Currently, alphabet imagery has been largely confined to vowels and few characters only.…”

Section: Research Gapsmentioning

confidence: 99%

“…So, the subject has to repeat certain imagined actions to complete 26 alphabets. Due to these repetitive actions, the time consumed for the display of each alphabet is large 25 . BCI with SI can overcome the above challenges.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Electroencephalography based imagined alphabets classification using spatial and time‐domain features

Agarwal

Kumar

2021

Int J Imaging Syst Tech

Self Cite

View full text Add to dashboard Cite

Imagined speech is a neuro-paradigm that can provide an alternative communication channel for patients in a locked-in syndrome state. We have performed an experiment in which a 32 channel industry-standard electroencephalography (EEG) device was used to record 26 imagined English alphabets from 13 subjects. We denoised the imagined signals by discrete wavelet transform and extracted the spatial filters by common spatial pattern method, and time-domain features. Spatial features when classified with linear support vector machine, and time-domain features classified by random forest gave the best results. Alpha, beta, and theta bands could classify imagined alphabets better than other bands and had average classification accuracies of 88.59%, 87.39%, and 88.97%, respectively by using spatial features and 81.88%, 76.72%, and 79.25%, respectively, by time-domain features. The grand average accuracies of all the 26 alphabets in six EEG frequency bands was found to be 77.97% in a subject independent binary classification framework.

show abstract