“Interactive Analysis and Display of The Electroencephalogram (EEG) In Real Time” by Gevins, Yeager and Diamond

  • ©Alan S. Gevins, Charles L. Yeager, and Stephen L. Diamond




    Interactive Analysis and Display of The Electroencephalogram (EEG) In Real Time



    The development of fast, relatively inexpensive graphics terminals permits the integration of interactive display with on-line, real time signal analysis in dedicated minicomputer systems. Interactive display in the context of electroencephalogram (EEG) analysis includes: 1. control over form and parameters of an isometric plot using hidden line suppression and floating level; 2. selection of function and transform to graph; and 3. expansion or contraction of coordinate axes and time scale, from “wide angle” views of the course of several hours of recording of the entire scalp, to “telescopic” blow-ups of the details of two or three seconds of activity in a single channel.The analysis paradigms function on-line and in real time, and incorporate parallel time domain and spectral analysis on data from eight or fewer electrode placements. Spectral estimates of power and simple coherences are typically formed using ensemble averages of non-overlapping one-half to two second periodograms obtained by Fast Fourier Transform. Transient analysis uses heuristic strategies to isolate clinically significant patterns, such as the sharp paroxysmal wave forms associated with the epilepsies. Instrumental and extra-cerebral artifacts are partially eliminated by algorithms based upon heuristic criteria. Parameters of spectral and transient analysis are interactively alterable to facilitate varying experimental paradigms.To implement such systems, which can be both compute and I/0 bound, it is necessary to optimize systems software. A simple and multitasking executive coordinates program task modules. Memory space and computing time are conserved through extensive use of dynamic buffer allocation and double precision integer arithmetic.

ACM Digital Library Publication:

Overview Page: