[Zurück]

P. Anderer, H. Semlitsch, M. Eberstaller, B. Saletu, P. Pfundner:
"A Robust and Valid Artifact Minimization & Identification Procedure for 'Cognitive' Event Related Brain Potentials";
Poster: EMBEC 1999 European Medical and Biological Engineering Conference, Vienna; 04.11.1999 - 07.11.1999; in: "Proceedings of the European Medical and Biological Engineering Conference", P. Rehak, H. Hutten (Hrg.); Journal of Medical and Biological Engineering & Computing, 37 (1999), Suppl. 2; S. 486 - 487.

Late positive components of event-related brain potentials (ERPs) are closely related to cognitive information processing. For clinical applications automatic artifact processing is essential. Based on ERP-recordings obtained in 80 normal healthy subjects and 80 cognitively impaired patients, an artifact minimization and identification procedure was developed and validated. It consists of the following steps: (1) Automatic detection of epochs with exclusively ocular artifacts (OA). (2) Estimation of the OA propagation factors. (3) Calculation of 'corrected' EEG. (4) Determination of major artifacts based on EEG-amplitude measures and on a discriminant score derived from log-transformed spectra! density power values. (5) Sensitive artifact identification based on a discriminant score derived from individually scaled z-transformed data. Cut-off points determined for a specificity of 95% in the training set were applied to the validation set, which resulted in a specificity of 95% and a sensitivity of 78% corresponding to an overall diagnostic power of 94% with a prevalence of 11%. The developed automatic artifact processing method makes the clinical application of 'cognitive' ERPs inexpensive and readily available.