Image quality of different reconstruction algorithms in CBCT

Objectives: To investigate the subjective image quality of different 3D reconstruction approaches in CBCT imaging plus their potential to operate with reduced input information. Methods: An artifact reducing algebraic reconstruction-type-algorithm and a novel-approach algebraic reconstruction algorithm with L1-norm regularization were implemented in software. The Feldkamp algorithm was used as standard. These algorithms were tested with original and reduced numbers of projection images as input data. To test the visual image quality, a full projection-image set was obtained in a CBCT device (Morita Accuitomo FP) of a human dry cervical spine. 3D reconstructions were made with the software and visually compared. Results: L1-Norm regularization with the full number of projection images was subjectively judged to have highest image quality. The potential to reduce the number of projection images (and thus dose) appears also to be highest for this type of regularization when compared to the other algorithms. Conclusions: Algebraic reconstruction in comparison to todays commonly used direct inversion algorithms are capable to deal with less or more corrupted input information (projection images). L1-norm regularization appears to be a promising research direction for future applications. This research is being funded by the German Federal Ministry of Education and Research (FHprofUnt 2009 # 1748X09). 1748X09).