OBJECTIVE: The objective of this study was to reduce metal-induced streak artifact on oral and maxillofacial x-ray computed tomography (CT) images by developing the fast statistical image reconstruction system using iterative reconstruction algorithms. STUDY DESIGN: Adjacent CT images often depict similar anatomical structures in thin slices. So, first, images were reconstructed using the same projection data of an artifact-free image. Second, images were processed by the successive iterative restoration method where projection data were generated from reconstructed image in sequence. Besides the maximum likelihood-expectation maximization algorithm, the ordered subset-expectation maximization algorithm (OS-EM) was examined. Also, small region of interest (ROI) setting and reverse processing were applied for improving performance. RESULTS: Both algorithms reduced artifacts instead of slightly decreasing gray levels. The OS-EM and small ROI reduced the processing duration without apparent detriments. Sequential and reverse processing did not show apparent effects. CONCLUSIONS: Two alternatives in iterative reconstruction methods were effective for artifact reduction. The OS-EM algorithm and small ROI setting improved the performance.
OBJECTIVE: The objective of this study was to reduce metal-induced streak artifact on oral and maxillofacial x-ray computed tomography (CT) images by developing the fast statistical image reconstruction system using iterative reconstruction algorithms. STUDY DESIGN: Adjacent CT images often depict similar anatomical structures in thin slices. So, first, images were reconstructed using the same projection data of an artifact-free image. Second, images were processed by the successive iterative restoration method where projection data were generated from reconstructed image in sequence. Besides the maximum likelihood-expectation maximization algorithm, the ordered subset-expectation maximization algorithm (OS-EM) was examined. Also, small region of interest (ROI) setting and reverse processing were applied for improving performance. RESULTS: Both algorithms reduced artifacts instead of slightly decreasing gray levels. The OS-EM and small ROI reduced the processing duration without apparent detriments. Sequential and reverse processing did not show apparent effects. CONCLUSIONS: Two alternatives in iterative reconstruction methods were effective for artifact reduction. The OS-EM algorithm and small ROI setting improved the performance.
Authors: An De Crop; Jan Casselman; Tom Van Hoof; Melissa Dierens; Elke Vereecke; Nicolas Bossu; Jaime Pamplona; Katharina D'Herde; Hubert Thierens; Klaus Bacher Journal: Neuroradiology Date: 2015-05-01 Impact factor: 2.804