Image quality assessment of standard- and low-dose chest CT using filtered back projection, adaptive statistical iterative reconstruction, and novel model-based iterative reconstruction algorithms.

OBJECTIVE: The purpose of this article is to compare image quality between filtered back projection (FBP), adaptive statistical iterative reconstruction (ASIR), and model-based iterative reconstruction (MBIR) at standard dose and two preselected low-dose scans. SUBJECTS AND METHODS: Thirty patients (16 men and 14 women; mean age, 67 years) were prospectively recruited. Patients underwent three scans (one standard-dose scan and two low-dose scans at noise indexes [NIs] of 33, 60, and 70, respectively). All three scans were reconstructed with FBP, ASIR, and MBIR. Objective and subjective image qualities were compared. Dose-length products and effective doses for each scans were recorded. Mean image noise and attenuation values were compared between different reconstruction algorithms using repeated-measures analysis of variance and paired Student t tests. The interobserver variation between the two radiologists for subjective image quality and lesion assessment was estimated by using weighted kappa statistics.
RESULTS: Objective image analysis supports significant noise reduction with low-dose scans using the MBIR technique (p < 0.05). There was no significant change in mean CT numbers between different reconstructions (p > 0.05). Subjective analysis reveals no significant difference between image quality and diagnostic confidence between low-dose MBIR scans compared with standard-dose scans reconstructed using ASIR (p > 0.05). Average effective doses were 3.7, 1.2, and 0.9 mSv for standard scans at NIs of 33, 60, and 70, respectively.
CONCLUSION: MBIR shows superior noise reduction and improved image quality. Substantial dose reduction can be achieved by increasing the NI parameters as tested in this study without affecting image quality and diagnostic confidence.