Combining automated attenuation-based tube voltage selection and iterative reconstruction: a liver phantom study

To determine the value of combined automated attenuation-based tube-potential selection and iterative reconstructions (IRs) for optimising computed tomography (CT) imaging of hypodense liver lesions.

A liver phantom containing hypodense lesions was imaged by CT with and without automated attenuation-based tube-potential selection (80, 100 and 120 kVp). Acquisitions were reconstructed with filtered back projection (FBP) and sinogram-affirmed IR. Image noise and contrast-to-noise ratio (CNR) were measured. Two readers marked lesion localisation and rated confidence, sharpness, noise and image quality on a five-point scale (1 = worst, 5 = best).

Image noise was lower (31-52%) and CNR higher (43-102%) on IR than on FBP images at all tube voltages. On 100-kVp and 80-kVp IR images, confidence and sharpness were higher than on 120-kVp FBP images. Scores for image quality score and noise as well as sensitivity for 100-kVp IR were similar or higher than for 120-kVp FBP and lower for 80-kVp IR. Radiation dose was reduced by 26% at 100 kVp and 56% at 80 kVp.

Compared with 120-kVp FBP images, the combination of automated attenuation-based tube-potential selection at 100 kVp and IR provides higher image quality and improved sensitivity for detecting hypodense liver lesions in vitro at a dose reduced by 26%.

• Combining automated tube voltage selection/iterative CT reconstruction improves image quality. • Attenuation values remain stable on IR compared with FBP images. • Lesion detection was highest on 100-kVp IR images.