Reduction of dental filling metallic artifacts in CT-based attenuation correction of PET data using weighted virtual sinograms

The streak artifacts caused by dental fillings are known to generate artifacts in attenuation maps thus leading to overestimation/underestimation of tracer uptake in resulting PET images. Correction of these artifacts is therefore mandatory to achieve accurate attenuation maps to be used by the CT-based attenuation correction procedure. Our group recently proposed a metal artifact reduction method using the virtual sinogram concept. In spite of the satisfactory results obtained on phantom studies, the algorithm was suffering from slight imperfections such as discontinuity of corrected sinogram bins along the second dimension of the sinogram matrix (i.e. the one sampling angular positions) and an evident difference between corrected and non-affected sinogram bins, especially when the number of affected bins in one column of the sinogram matrix is noticeable. The purpose of this study is to present an improved method allowing to overcome the above mentioned limitations. The proposed method enhances the corrected sinogram using weighted values of the influenced bins in the corrected and non-corrected sinograms, and weighted values of the sinogram bins in the neighboring column of the sinogram matrix. The optimum weighting factors associated with the three data sets mentioned above (?, ? and ?) were determined by assessing different combinations of these factors having values falling in the range [0.1-0.9] and statistical analysis of the images of 10 patients with dental fillings. The optimum combination achieved is ?=0.3, ?=0.6 and ?=0.1. The results reveal that the corrected CT images and attenuation maps preserve more detail especially in regions adjacent to metallic objects whereas the original method suffered from eliminating relevant anatomical details. It was concluded that the proposed MAR method using weighted virtual sinograms improves the quality of clinical CT studies thus allowing the achievement of more accurate attenuation correction of PET data.