Recent Trends in EEG-Based Motor Imagery Signal Analysis and Recognition: A Comprehensive Review

Abstract

The electroencephalogram (EEG) motor imagery (MI) signals are the widespread paradigms in the brain-computer interface (BCI). Its significant applications in the gaming, robotics, and medical fields drew our attention to perform a detailed analysis. However, the problem is ill-posed as these signals are highly nonlinear, unpredictable, and noisy, hence making it exceedingly hard to be analyzed adequately. This paper provides a first-of-its-kind comprehensive review of conventional signal processing and deep learning techniques for BCI MI signal analysis. The review comprises extensive works carried out in the domain in the recent past, highlighting the current challenges of the problem. A new categorization of the existing approaches has been presented for better clarification. An all-inclusive description of the signal processing techniques has been corroborated by relevant works in the area. Moreover, architectures of various standard deep learning algorithms along with their merits and demerits are also explicated to assist the readers. The tabular representations of the numerical results are also readily provided. This work also presents the open research problems and future directions.

Keywords

• Brain-computer interface (BCI)
• convolutional neural network (CNN)
• electroencephalogram (EEG)
• motor imagery (MI)
• variational autoencoders (VAE)