A Detailed Analysis & Parametric Comparison Of Eeg Processing Models

Abstract

Electroencephalograms (EEGs) are capable of representing brain signals in terms of numerical vector sets. These signals are used to
estimate a wide variety of neurological disorders including Dementias, Epilepsy, Parkinson, Stroke, Transient Ischemic Attack, etc. A
wide variety of machine learning based methods are proposed for processing these signals, and each of these are variant in terms of the
qualitative nuances, function advantages, application-specific characteristics, qualitative limitations, and deployment-specific future
scopes. Thus, it is difficult for researchers to identify optimal EEG processing models for their clinical use cases. These models vary
in terms of their performance metrics, which further complicates the process of model selection for clinical use cases. To overcome
these issues, this paper initially provides a detailed discussion about existing EEG processing techniques, in terms of their functional
details. This will allow readers to identify optimal machine learning techniques, which are suited for the functional use cases.
Continuing this discussion, a comparative analysis of these models is done on the basis of their clinical accuracy, precision,
computational complexity, delay and scalability metrics, which will assist readers to identify models for their performance-specific use
cases. Thus, this text will allow readers to identify optimally performing models for different EEG processing scenarios.