OBJECTIVE: The objective of our study was to evaluate the method for detection and removal of artifacts in evoked potential monitoring described earlier by Cluitmans and colleagues in a clinical setting. METHODS: The method for detection and removal of artifacts by Cluitmans and colleagues is based on the assumption that a sweep of the recorded electroencephalogram (EEG) signal contains artifacts if one or more variables derived from the signal deviates strongly from the normal range of values. Once these normal ranges are defined, all future EEG recordings that are recorded under comparable circumstances can be automatically evaluated for artifacts by tracking when one or more signal variables falls outside the normal range. To assess the performance of this method in a clinical setting, recordings from a learning set were visually evaluated for artifacts. From the empirical distribution functions of the signal variables, the thresholds for automatic detection of artifacts were determined. The auditory evoked potential (AEP) waveforms resulting after automatic screening were compared with the waveforms obtained after visual evaluation of the raw signal combined with manual exclusion of signal periods containing artifacts. RESULTS: The quality of the resulting waveform was improved by our method of automatic detection and removal of artifacts in 97% of partly contaminated recordings. In only 2% of the recordings, automatic screening slightly degraded the resulting waveform. CONCLUSIONS: We conclude that the described method of automatic detection and removal of artifacts in AEP recordings effectively improves the quality of the resulting AEP waveform, without excessive rejection of artifact-free signal periods. The signal variables used in this method seem appropriate for distinguishing artifact-free signal periods from periods containing artifacts for the types of artifact that were studied.
OBJECTIVE: The objective of our study was to evaluate the method for detection and removal of artifacts in evoked potential monitoring described earlier by Cluitmans and colleagues in a clinical setting. METHODS: The method for detection and removal of artifacts by Cluitmans and colleagues is based on the assumption that a sweep of the recorded electroencephalogram (EEG) signal contains artifacts if one or more variables derived from the signal deviates strongly from the normal range of values. Once these normal ranges are defined, all future EEG recordings that are recorded under comparable circumstances can be automatically evaluated for artifacts by tracking when one or more signal variables falls outside the normal range. To assess the performance of this method in a clinical setting, recordings from a learning set were visually evaluated for artifacts. From the empirical distribution functions of the signal variables, the thresholds for automatic detection of artifacts were determined. The auditory evoked potential (AEP) waveforms resulting after automatic screening were compared with the waveforms obtained after visual evaluation of the raw signal combined with manual exclusion of signal periods containing artifacts. RESULTS: The quality of the resulting waveform was improved by our method of automatic detection and removal of artifacts in 97% of partly contaminated recordings. In only 2% of the recordings, automatic screening slightly degraded the resulting waveform. CONCLUSIONS: We conclude that the described method of automatic detection and removal of artifacts in AEP recordings effectively improves the quality of the resulting AEP waveform, without excessive rejection of artifact-free signal periods. The signal variables used in this method seem appropriate for distinguishing artifact-free signal periods from periods containing artifacts for the types of artifact that were studied.
Authors: M R Goldstein; D T Plante; B K Hulse; S Sarasso; E C Landsness; G Tononi; R M Benca Journal: Acta Psychiatr Scand Date: 2011-11-19 Impact factor: 6.392