Annals of Biomedical Engineering, Vol. 19, pp. 1-14, 1991 0090-6964/91 $3.00 + .00 Printed in the USA. All rights reserved. Copyright 9 1991 Pergamon Press plc The Application of Delta Modulation to EEG Waveforms for Database Reduction and Real-Time Signal Processing Denise Wawrzynski Nicoletti and Banu Onaral Department of Electrical and Computer Engineering Drexel University Philadelphia, PA (Received 6/23/89; Revised 8/1/90) The large volume of digital data and the demanding processing task involved in electroencephalogram (EEG) analysis place stringent requirements on computer re- sources in terms of data transfer, computation speed, and temporary or permanent storage. The reduction of the database to a manageable size is therefore necessary for economical use of transmission channels and the storage media. The two criteria, waveform reproducibility and processing applications, must be analyzed and opti- mized in terms of signal-to-noise ratio (SNR) using the various factors affecting the coding. This analysis and optimization can become cumbersome, and a specialized workstation has been developed specifically for analysis of digital coding. Our inter- est in data compression stems from the study of the feasibility of predicting pilots" acceleration (Gz) tolerance during flight by processing both their uncoded and coded EEG. INTRODUCTION We have investigated the coding of electroencephalograms (EEG), and have ana- lyzed the waveform reproducibility and processing applications. The large volume of digital data and the demanding processing task involved in quantified multichannel EEG analysis place stringent requirements on computer resources in terms of data transfer, computation speed, and temporary or permanent storage. These constraints are more severe in applications requiring multielectrode processing such as brain ac- tivity mapping (5,6) and real-time operations, such as automatic control of orthopedic prostheses (4,22). The reduction of the database to a manageable size is therefore nec- essary for economical use of transmission channels and the storage media. The two criteria, waveform reproducibility and processing applications, must be analyzed and optimized, i.e., minimize the signal-to-noise ratio (SNR), using the various factors af- fecting the coding: the structure of the coder, pre- or postfiltering, sampling fre- quency, and other factors, depending on the type and complexity of the coding implemented. This analysis can become cumbersome, and a specialized workstation, Address correspondence to ProfessorDeniseW. Nicoletti, Departmentof Electricaland Computer En- gineering, Drexel University, Philadelphia, PA 19104. 1