CT of the chest with model-based, fully iterative reconstruction: comparison with adaptive statistical iterative reconstruction

Background: The recently developed model-based iterative reconstruction (MBIR) enables significant reduction of image noise and artifacts, compared with adaptive statistical iterative reconstruction (ASIR) and filtered back projection (FBP). The purpose of this study was to evaluate lesion detectability of low-dose chest computed tomography (CT) with MBIR in comparison with ASIR and FBP. Methods: Chest CT was acquired with 64-slice CT (Discovery CT750HD) with standard-dose (5.7 +/- 2.3 mSv) and low-dose (1.6 +/- 0.8 mSv) conditions in 55 patients (aged 72 +/- 7 years) who were suspected of lung disease on chest radiograms. Low-dose CT images were reconstructed with MBIR, ASIR 50% and FBP, and standard-dose CT images were reconstructed with FBP, using a reconstructed slice thickness of 0.625 mm. Two observers evaluated the image quality of abnormal lung and mediastinal structures on a 5-point scale (Score 5 = excellent and score 1 = non-diagnostic). The objective image noise was also measured as the standard deviation of CT intensity in the descending aorta. Results: The image quality score of enlarged mediastinal lymph nodes on low-dose MBIR CT (4.7 +/- 0.5) was significantly improved in comparison with low-dose FBP and ASIR CT (3.0 +/- 0.5, p = 0.004; 4.0 +/- 0.5, p = 0.02, respectively), and was nearly identical to the score of standard-dose FBP image (4.8 +/- 0.4, p = 0.66). Concerning decreased lung attenuation (bulla, emphysema, or cyst), the image quality score on low-dose MBIR CT (4.9 +/- 0.2) was slightly better compared to low-dose FBP and ASIR CT (4.5 +/- 0.6, p = 0.01; 4.6 +/- 0.5, p = 0.01, respectively). There were no significant differences in image quality scores of visualization of consolidation or mass, ground-glass attenuation, or reticular opacity among low-and standard-dose CT series. Image noise with low-dose MBIR CT (11.6 +/- 1.0 Hounsfield units (HU)) were significantly lower than with low-dose ASIR (21.1 +/- 2.6 HU, p < 0.0005), low-dose FBP CT (30.9 +/- 3.9 HU, p < 0.0005), and standard-dose FBP CT (16.6 +/- 2.3 HU, p < 0.0005). Conclusion: MBIR shows greater potential than ASIR for providing diagnostically acceptable low-dose CT without compromising image quality. With radiation dose reduction of >70%, MBIR can provide equivalent lesion detectability of standard-dose FBP CT.