Impact of metal artifacts due to EEG electrodes in brain PET imaging

Neurological PET/CT examinations are often performed with electroencephalogram (EEG) monitoring for epilepsy patients. However, the EEG electrodes cause metal artifacts in the reconstructed CT images which might propagate through the attenuation map into the PET reconstructions. In this study, the impact of the electrodes on the visual quality and quantification of brain PET images was investigated. Twenty epilepsy patients with EEG monitoring referred for a PET/CT examination were used. The CT data were reconstructed using both FBP yielding an artifactual image and a metal artifact reduction (MAR) algorithm yielding a CT image with reduced artifacts. Both data sets were used for CT-based attenuation correction (AC) of the PET data. In addition, a calculated attenuation correction (CALC) technique using a three-compartment model (brain tissue, skull and scalp) was also considered. This resulted in three PET reconstructions for each patient which were assessed visually and analyzed quantitatively. For the quantitative analysis, the PET images were spatially normalized to a 3D ¹⁸F-FDG stereotactic template. The mean activity concentration of 63 VOIs was calculated for each PET image and correlations between the different AC methods were assessed on a VOI by VOI basis using MAR-based AC as the gold standard. The visual assessment showed local hot spots corresponding to the locations of the electrodes when using FBP-AC. These artifacts disappeared when MAR-AC or CALC-AC were used. However, the tracer uptake was overestimated in the neighborhood of the nasal cavities when using with the CALC-AC. The quantitative analysis showed a very good correlation (R²=0.99) between FBP- AC and MAR-AC, however a bias ranging from 4.1% to 5.4% was noticed. There was also a good correlation between CALC- AC and MAR-AC (R²=0.98) with a bias ranging from -5.1% to 4.3%. The EEG electrodes give rise to local hot spots and quantification bias. The use of the - MAR algorithm combined with an optimized attenuation coefficient for the electrodes significantly reduces the artifacts and bias. When a CT scan is not available, the use of a three-compartment calculated attenuation map is a good alternative; however care has to be taken in the neighborhood of air and nasal cavities.