Abstracts

RATIONALE: Epileptic spike detection in simultaneous EEG and continuous fMRI recording is a challenging task due to the huge artifacts induced by fMRI. The present study is to test the performance of Fourier filtering [1] as a post-processing method to remove fMRI-induced artifacts and reveal epileptic activity.
METHODS: 6 patients with active epileptic discharges took part in this study. All the measurements were performed on a 1.5 T Magnetom Vision whole body scanner. EEG was recorded in the scanner using a specially designed EEG amplifier (Schwarzer, Germany) with a sampling rate of 1000 Hz and online filtering of 0.5 to 70 Hz. The digital information was transferred out of the MR cabin by optic fiber to a control computer. Specific measures were taken to immobilize patients, electrode wires, and the head box in the scanner. The EEG recordings were then subject to Fourier filtering to remove the artifacts. Epileptic spikes were visually inspected and marked by EEGers. In addition, Principal Component Analysis (PCA) was tested for the removal of ballistocardiogram artifacts.
RESULTS: After filtering, epileptic spikes were clearly distinguishable in 5 out of 6 patients. In the sixth patient, even the spikes out of the scanner were quite difficult to distinguish due to low amplitude. The signal quality of spikes recovered by filtering heavily depended on the original amplitude of the fMRI induced artifacts. Additional measurements showed that well bandaging electrodes on the head, fixing electrode wires in the scanner, reducing head box vibration by cushion were important measures to reduce the artifact in the first place and therefore improve the quality of the recovered spikes. In addition, PCA is an easy way to remove the ballistocardiogram artifact by eliminating the first principal component.
CONCLUSIONS: fMRI induced artifacts can be reduced significantly by immobilization of the wires and preventing vibration of the amplifier. Ballistocardiogram artifacts are minimal and can be removed by PCA. Epileptic spikes in simultaneous EEG and fMRI recordings are detectable after removing the remaining artifacts by Fourier filtering.
[1] Hoffmann A. Jäger L. Werhahn KJ. Jaschke M. Noachtar S. Reiser M. Electroencephalography during functional echo-planar imaging: detection of epileptic spikes using post-processing methods. Magnetic Resonance in Medicine. 44(5):791-8, 2000 Nov
Support: Supported by Canadian Institutes of Health Research grant MOP-38079