Hong Zhang1, Vic Kong1, Ke Huang1, Jian-Yue Jin1,2. 1. 1 Department of Radiation Oncology, Georgia Regents University, Augusta, GA, USA. 2. 2 Department of Radology, Georgia Regents University, Augusta, GA, USA.
Abstract
PURPOSE: To present our experiences in understanding and minimizing bowtie-filter crescent artifacts and bowtie-filter normalization artifacts in a clinical cone beam computed tomography system. METHODS: Bowtie-filter position and profile variations during gantry rotation were studied. Two previously proposed strategies (A and B) were applied to the clinical cone beam computed tomography system to correct bowtie-filter crescent artifacts. Physical calibration and analytical approaches were used to minimize the norm phantom misalignment and to correct for bowtie-filter normalization artifacts. A combined procedure to reduce bowtie-filter crescent artifacts and bowtie-filter normalization artifacts was proposed and tested on a norm phantom, CatPhan, and a patient and evaluated using standard deviation of Hounsfield unit along a sampling line. RESULTS: The bowtie-filter exhibited not only a translational shift but also an amplitude variation in its projection profile during gantry rotation. Strategy B was better than strategy A slightly in minimizing bowtie-filter crescent artifacts, possibly because it corrected the amplitude variation, suggesting that the amplitude variation plays a role in bowtie-filter crescent artifacts. The physical calibration largely reduced the misalignment-induced bowtie-filter normalization artifacts, and the analytical approach further reduced bowtie-filter normalization artifacts. The combined procedure minimized both bowtie-filter crescent artifacts and bowtie-filter normalization artifacts, with Hounsfield unit standard deviation being 63.2, 45.0, 35.0, and 18.8 Hounsfield unit for the best correction approaches of none, bowtie-filter crescent artifacts, bowtie-filter normalization artifacts, and bowtie-filter normalization artifacts + bowtie-filter crescent artifacts, respectively. The combined procedure also demonstrated reduction of bowtie-filter crescent artifacts and bowtie-filter normalization artifacts in a CatPhan and a patient. CONCLUSION: We have developed a step-by-step procedure that can be directly used in clinical cone beam computed tomography systems to minimize both bowtie-filter crescent artifacts and bowtie-filter normalization artifacts.
PURPOSE: To present our experiences in understanding and minimizing bowtie-filter crescent artifacts and bowtie-filter normalization artifacts in a clinical cone beam computed tomography system. METHODS: Bowtie-filter position and profile variations during gantry rotation were studied. Two previously proposed strategies (A and B) were applied to the clinical cone beam computed tomography system to correct bowtie-filter crescent artifacts. Physical calibration and analytical approaches were used to minimize the norm phantom misalignment and to correct for bowtie-filter normalization artifacts. A combined procedure to reduce bowtie-filter crescent artifacts and bowtie-filter normalization artifacts was proposed and tested on a norm phantom, CatPhan, and a patient and evaluated using standard deviation of Hounsfield unit along a sampling line. RESULTS: The bowtie-filter exhibited not only a translational shift but also an amplitude variation in its projection profile during gantry rotation. Strategy B was better than strategy A slightly in minimizing bowtie-filter crescent artifacts, possibly because it corrected the amplitude variation, suggesting that the amplitude variation plays a role in bowtie-filter crescent artifacts. The physical calibration largely reduced the misalignment-induced bowtie-filter normalization artifacts, and the analytical approach further reduced bowtie-filter normalization artifacts. The combined procedure minimized both bowtie-filter crescent artifacts and bowtie-filter normalization artifacts, with Hounsfield unit standard deviation being 63.2, 45.0, 35.0, and 18.8 Hounsfield unit for the best correction approaches of none, bowtie-filter crescent artifacts, bowtie-filter normalization artifacts, and bowtie-filter normalization artifacts + bowtie-filter crescent artifacts, respectively. The combined procedure also demonstrated reduction of bowtie-filter crescent artifacts and bowtie-filter normalization artifacts in a CatPhan and a patient. CONCLUSION: We have developed a step-by-step procedure that can be directly used in clinical cone beam computed tomography systems to minimize both bowtie-filter crescent artifacts and bowtie-filter normalization artifacts.
Authors: Dandan Zheng; John C Ford; Jun Lu; Dimitrios Lazos; Geoffrey D Hugo; Damodar Pokhrel; Lisha Zhang; Jeffrey F Williamson Journal: Med Phys Date: 2011-05 Impact factor: 4.071