[Dose reduction and image quality in MDCT of the upper abdomen: potential of an adaptive post-processing filter]

To evaluate the effects of a 2D non-linear adaptive post-processing filter (2D-NLAF) on image quality in dose-reduced multi-detector CT (MDCT) of the upper abdomen.

MDCT of the upper abdomen was simulated on a 64-slice scanner using a multi-modal anthropomorphic phantom (CIRS, Norfolk, USA). While keeping the collimation (64 x 0.6 mm) and pitch (p = 1) unchanged, the tube current (100 - 500 mAs) and tube potential (80 - 140 kVp) were varied to perform MDCT as high dose (CTDI > 20), middle dose (CTDI 10 - 20) and low dose (CTDI < 10) level protocols. Four independent blinded radiologists evaluated axial images with a thickness of 7 and 3 mm with respect to the presentation of "mesenteric low contrast lesions", "liver veins", "liver cysts", "renal cysts" and "big vessels". The subjective image quality of original data and post-processed images using a 2D-NLAF (SharpViewCT, Linköping, Sweden) was graded on a 5-point scale (from "1" not visible to "5" excellent) and statistically analyzed. The effective dose (E) was estimated using commercial software (CT-EXPO).

For all protocol groups, 2D-NLAF led to a significant improvement in subjective image quality for all examined lesions (p < 0.01), particularly at the protocols of middle dose (E: 5 - 8 mSv) and low dose level (E: 1 - 5 mSv). A maximum effect was seen in middle dose protocols for "low contrast lesions" (score "3.3" with filter versus "2.5" without) and "liver veins" ("4.5" versus "3.9").

The phantom study indicates a potential dose reduction of up to 50 % in MDCT of the upper abdomen by use of a 2D-NLAF, which should be further examined in clinical trails.