Attenuation-based automatic kilovoltage selection and sinogram-affirmed iterative reconstruction: effects on radiation exposure and image quality of portal-phase liver CT

Abstract

Objective: To compare the radiation dose and image quality between standard-dose CT and a low-dose CT obtained with the combined use of an attenuation-based automatic kilovoltage (kV) selection tool (CARE kV) and sinogram-affirmed iterative reconstruction (SAFIRE) for contrast-enhanced CT examination of the liver.

Materials and methods: We retrospectively reviewed 67 patients with chronic liver disease in whom both, standard-dose CT with 64-slice multidetector-row CT (MDCT) (protocol A), and low-dose CT with 128-slice MDCT using CARE kV and SAFIRE (protocol B) were performed. Images from protocol B during the portal phase were reconstructed using either filtered back projection or SAFIRE with 5 different iterative reconstruction (IR) strengths. We performed qualitative and quantitative analyses to select the appropriate IR strength. Reconstructed images were then qualitatively and quantitatively compared with protocol A images.

Results: Qualitative and quantitative analysis of protocol B demonstrated that SAFIRE level 2 (S2) was most appropriate in our study. Qualitative and quantitative analysis comparing S2 images from protocol B with images from protocol A, showed overall good diagnostic confidence of S2 images despite a significant radiation dose reduction (47% dose reduction, p < 0.001).

Conclusion: Combined use of CARE kV and SAFIRE allowed significant reduction in radiation exposure while maintaining image quality in contrast-enhanced liver CT.

Keywords: Computed tomography; Image quality; Iterative reconstruction; Radiation dose reduction; Tube potential.

Fig. 1

Transverse contrast-enhanced liver CT images of 47-year-old female (body mass index, 22.4 kg/m²) with chronic hepatitis B. All images were obtained with tube voltage of 100 kV and 110 effective mAs (protocol B). Image noise decreased as SAFIRE level increased. However, as level increased, pixelated image appearance also increased.

Fig. 2

Transverse contrast-enhanced liver CT images of 67-year-old female (body mass index, 26.7 kg/m²) with liver cirrhosis. Previous CT (A) was scanned at 120 kV (136 eff. mAs) with volume CT dose index (CTDI_vol) of 10.4 mGy and follow-up CT (B) was performed at 100 kV (132 eff. mAs) with CTDI_vol of 5.2 mGy. Both readers selected (B) as preferred image.

Fig. 3

Transverse contrast-enhanced liver CT images of 52-year-old female (body mass index, 33.2 kg/m²) with liver cirrhosis and history of radiofrequency ablation due to hepatocellular carcinoma. Previous CT (A) was scanned at 120 kV (249 eff. mAs) with volume CT dose index (CTDI_vol) of 16.48 mGy and follow-up CT (B-D) was scanned at 100 kV (244 eff. mAs) with CTDI_vol of 9.6 mGy. Although S4 (D) and S5 (not shown) images are almost free of image noise, overall diagnostic confidence was rated average (score 3) by both readers due to pixelated and artificial appearance. For comparison of protocol A and B, both readers selected 2 images (A, C) with same preference.