Reduction in dose from CT examinations of liver lesions with a new postprocessing filter: a ROC phantom study.

BACKGROUND: It is desirable to lower the dose from computed tomography (CT) examinations as much as possible without reducing diagnostic performance. Mathematical postprocessing filters are one tool to achieve dose reduction. PURPOSE: To evaluate the possibilities of reducing CT doses from liver examinations using a new postprocessing filter. MATERIAL AND METHODS: An anthropomorphic upper-abdomen phantom was used in receiver operating characteristic (ROC) studies of the detectability of liver lesions. A standard abdominal CT protocol was used. Only mA settings were changed; all other scan parameters were constant. The postprocessing filter used was SharpView CT, which provides context-controlled restoration of digital images using adaptive filters. Six readers were given a set of 10 images obtained at five different dose levels, each image with 32 predefined areas to be evaluated on a five-point scale. In total, 1920 areas were evaluated. At each dose level, the readers evaluated five images without enhancement and five images based on postprocessing filters. All images were randomized with respect to dose level. RESULTS: The postprocessing filter improved the diagnostic performance significantly compared to the unenhanced images at all dose levels. Radiation dose for abdominal CT examinations of liver lesions in the range 2-7 mm was reduced by 30% using postprocessing filters, while diagnostic performance of the examination was maintained or even improved. CONCLUSION: This study indicates great potential for lowering doses for CT examinations of liver lesions using the new postprocessing filter. The software must be fully tested clinically to reliably assess the benefits of this filtration.