EEG artifacts detection and removal techniques for brain computer interface applications: a systematic review

Electroencephalogram (EEG) being the measure to record the electrical activity of brain acts as a key factor to many brain computer interface (BCI) applications. These recorded EEG signals often gets interfered with artifacts of different types such as eye blink, muscle movements, cardiac etc. Such artifacts are to be detected and removed for efficient analysis of EEG signals in pre-processing stage. Hence, this systematic review aims to provide an overview on all the available methods to remove the physiological artifacts. In addition, comparison of all the methods and their performance evaluation metrics are discussed. Relevant 159 papers are considered from the databases such as SCOPUS, PUBMED, CROSSREF, WEB OF SCIENCE and GOOGLE SCHOLAR. Several analyses were made based on the collected information and current challenges for BCI applications in handling artifacts are provided. This paper also provides the details of available open source tools for pre-processing EEG data and publicly available artifacts databases. Findings show that: a)independent component analysis (ICA) is the most popular single artifact removal method b)ICA-wavelet is the most popular hybrid artifact removal method c) maximum publications are for removal of ocular artifacts and less on muscle artifact removal d)deep learning methods are to be experimented more to improve the performance. Even though there are many methods to remove the artifacts, there is no specific method to remove all the artifacts completely. This review also shows that there are still many open issues and research opportunities to handle EEG artifacts.