Accuracy of iodine removal using dual-energy CT with or without a tin filter: an experimental phantom study.

The effects of a tin filter on virtual non-enhanced (VNE) images created by dual-energy CT have not been well evaluated. To compare the accuracy of VNE images between those with and without a tin filter. Two different types of columnar phantoms made of agarose gel were evaluated. Phantom A contained various concentrations of iodine (4.5-1590 HU at 120 kVp). Phantom B consisted of a central component (0, 10, 25, and 40 mgI/cm(3)) and a surrounding component (0, 50, 100, and 200 mgI/cm(3)) with variable iodine concentration. They were scanned by dual-source CT in conventional single-energy mode and dual-energy mode with and without a tin filter. CT values on each gel at the corresponding points were measured and the accuracy of iodine removal was evaluated. On VNE images, the CT number of the gel of Phantom A fell within the range between -15 and +15 HU under 626 and 881 HU at single-energy 120 kVp with and without a tin filter, respectively. With attenuation over these thresholds, iodine concentration of gels was underestimated with the tin filter but overestimated without it. For Phantom B, the mean CT numbers on VNE images in the central gel component surrounded by the gel with iodine concentrations of 0, 50, 100, and 200 mgI/cm(3) were in the range of -19-+6 HU and 21-100 HU with and without the tin filter, respectively. Both with and without a tin filter, iodine removal was accurate under a threshold of iodine concentration. Although a surrounding structure with higher attenuation decreased the accuracy, a tin filter improved the margin of error.