Effect of iterative reconstruction techniques on image quality in low radiation dose chest CT: a phantom study.

PURPOSE To evaluate the quality of chest CT images obtained with low-dose computed tomography (LDCT) using three iterative reconstruction (IR) algorithms. METHODS Two 64-detector spiral CT scanners (HDCT, iCT) were used to scan a chest phantom containing 6 ground-glass nodules (GGNs) at 11 radiation dose levels. CT images were reconstructed by FBP or three IR algorithms. Reconstructed images were analyzed for CT values, average noise, contrast-to-noise ratio (CNR) values, subjective image noise, and diagnostic acceptability of the GGNs. Repeated-measures analysis of variance was used for statistical analyses. RESULTS Average noise decreased and CNR increased with increasing radiation dose when the same reconstruction algorithm was applied. Image average noise reconstructed with MBIR was significantly lower than reconstructed with iDOSE4 at the same low radiation doses. The two radiologists showed good interobserver consistency in image quality with kappa 0.83. A significant relationship was found between image noise and diagnostic acceptability of the GGNs. CONCLUSION Three IR algorithms are able to reduce the image noise and improve the image quality in LDCT. In the same radiation dose, the LDCT images quality reconstructed with MBIR algorithms is better than that of other IR algorithms.