Dose reduction in CT urography and vasculature phantom studies using model-based iterative reconstruction

To evaluate the feasibility of radiation dose reduction using model-based iterative reconstruction (MBIR) for evaluating the ureters and vasculature in a phantom, a tissue-equivalent CT dose phantom was scanned using a 64-channel CT scanner. Tubes of varying diameters filled with different dilutions of a contrast agent, simulating ureters or vessels, were inserted into the center of the phantom. Each combination was scanned using an existing renal protocol at 140 kVp or 120 kVp, yielding a display volumetric CT dose index (CTDI_vol) of 24 mGy. The scans were repeated using reduced scan techniques to achieve lower radiation doses down to 0.8 mGy. The images were reconstructed using filtered back-projection (FBP) and model-based iterative reconstruction (MBIR). The noise and contrast-to-noise ratio (CNR) was measured for each contrast object. Comparisons between the two reconstruction methods at different dose levels were evaluated using a factorial design. At each CTDI_vol the measured image noise was lower using MBIR compared to FBP (p < 0.0001). At low doses, the percent change in measured image noise between FBP and MBIR was larger. For the 12 mm object simulating a ureter or large vessel with an HU of 600, the measured CNR using MBIR at a CTDI_vol of 1.7 mGy was greater than the CNR of FBP at a CTID_vol of 24 mGy (p < 0.0001). For the 5 mm object simulating a medium-sized vessel with a HU of 250, the measured CNR using MBIR at a CTDI_vol of 1.7 mGy was equivalent to that of FBP at a CTDI_vol of 24 mGy. For the 2 mm, 100 HU object simulating a small vessel, the measured CNR using MBIR at a CTDI_vol of 1.7 mGy was equivalent to that of FBP at a CTDI_vol of 24 mGy. Low-dose (3.6 mGy) CT imaging of vasculature and ureter phantoms using MBIR results in similar noise and CNR compared to FBP at approximately one-sixth the dose. This suggests that, using MBIR, a one milliSievert exam of the ureters and vasculature may be clinically possible whilst still maintaining adequate image quality