- July 30, 2017
- Category: Scientific Publications
Clément Chevillard1,2, Jean-Luc Dumas2, Alejandro Mazal2, François Husson1
1 R&D Medical physics, DOSIsoft S.A., Cachan, France
2 Radiotherapy Department, Institut Curie, Paris, France
Presented at AAPM 2017
Purpose: The scope of this study was to review and compare the main issues of kV-CBCT imaging highlighted from different generation systems and their respective image quality in regards with fan-beam CT (FBCT). Moreover, kV-CBCT is potentially the reference data for the calculation of the daily actual delivered dose. However, the main limitation of kV-CBCT in clinical routine is the quality of the images. Indeed kV-CBCT beam geometry is different from FBCT, introducing much more artefacts and leading to Hounsfield Units (HU) inaccuracies.
Methods: Different kV-CBCT acquisition modes with different tube voltage (kV) and exposure (mAs) were studied: “pelvis” and “pelvis obese” with a Varian Truebeam™ (unit 1), “pelvis” and “low dose thorax” with a Varian Clinac 2100CD (unit 2). Using the TomoTherapy® Cheese phantom (Gammex RMI, Middleton, WI) for HU to electron density (de) calibration curve, and the NEMA body water phantom for HU uniformity, noise and mean water number analysis, were investigated. From the FBCT and kV-CBCT HU differential histograms, frequency peaks representing water, air, lung, soft tissue and bone were compared.
Results: For unit 2, streaking and ring artifacts are much greater than for unit 1 as well as HU uniformity is deteriorated. Noise values are comparable between units but standard deviation is twice at unit 2. Peaks observed in the differential histograms are broader in kV-CBCT than those of FBCT especially for high tissue density. kV-CBCT uniformity is more degraded than in FBCT because CBCT is more prone to artefacts in images such as: streaking, cupping, or ring artifacts. Variations in calibration protocols were observed between FBCT and kV-CBCT that could explain differences for mean water number.
Conclusion: The two commercial kV-CBCT systems demonstrate different performances. We found that HU differential histograms are an efficient way for comparing and controlling image quality in addition to existing quality metrics.