Development of a 3D printed anthropomorphic lung phantom for image quality assessment in CT

Purpose: To design a 3D printed anthropomorphic lung vessel phantom for CT image quality assessment and to evaluate the phantom image and dose characteristics. Methods: An in-house algorithm generated a vessel tree model, based on human lungs anatomy, which was 3D printed using a multi jet modeling printer (0.25 mm <= vessel diameters <= 8.25 mm) and inserted in an elliptical holder (thorax surrogate). The phantom was scanned (Toshiba Aquilion Genesis CT) and compared in terms of attenuation (Hounsfield units, HU) and dose characteristics with studies of five patients (normal BMI) and a commercial torso phantom, performed with the same thorax protocol. The pixel value distribution in the lung area was assessed with histograms. To investigate the adjustment of tube current modulation, tube load and CTDI were compared. Results: The histogram peaks for respectively vessels and surrounding tissue were at 105 HU and - 985 HU (3D printed phantom), at - 25 HU and -1000 HU (torso phantom) and at 25 HU and -875 HU (average patient). The contrast between vessels and surrounding was -1090 HU (3D printed), - 975 HU (torso phantom), and - 900 HU (average patient). The measured HU values (soft tissue and vertebra) were (32 +/- 15) HU and (210 +/- 71) HU (average patient); (4 +/- 4) HU, (390 +/- 39) HU (torso phantom) and (119 +/- 5) HU, (951 +/- 31) HU (3D printed phantom and holder). CTDIvol was (1.9 +/- 4.7 mGy) for patients, 1.9 mGy for the torso phantom and 2.1 mGy for the 3D printed lung phantom. Conclusions: An anthropomorphic 3D printed lung phantom was developed and its CT image and dose characteristics evaluated. The phantom has the potential to provide clinically relevant and reproducible measures of CT image quality.