Iterative Reconstructions versus Filtered Back-Projection for Urinary Stone Detection in Low-Dose CT

Abstract

To evaluate prospectively, in patients with suspected or known urinary stone disease, the image quality anddiagnostic confidence of nonenhanced abdominal low-dose computed tomography (CT) with iterative reconstruction (IR) compared to filtered back-projection (FBP). Fifty consecutive patients with suspected (n=33) or known (n=17) urinary stone disease underwent nonenhanced abdominal low-dose CT (120 kVp, 30 effective mAs, 1.6±0.5 mSv). Reconstructions were performed with sinogram-affirmed IR and with FBP. Attenuation (in Hounsfield units) was measured in subcutaneous fat and urinary bladder; image noise was determined. Two readers assessed image quality, number and location of urinary calculi were recorded, and diagnostic confidence was assessed. Statistical analyses included Mann-Whitney, Friedman's two-way, Wilcoxon signed rank, Pearson's, and Spearman's rank order correction tests. Attenuation of urinary bladder (P=.208, reader 1; P=.123, reader 2) and fat (P=.568, reader 1; P=.834, reader 2) was similar among FBP and IR datasets. Image noise was reduced in IR datasets by 40.1% (P<.001). IR improved image quality (P<.01), and obesity as factor impairing image quality was noted in FBP but not in IR images (P<.05). There was no significant difference in number of calculi in datasets reconstructed with IR and FBP (P=.102, reader 1; P=.059, reader 2). Diagnostic confidence regarding identification of urinary calculi improved with IR (P<.05, reader 1; P<.01, reader 2). IR improves image quality and confidence for diagnosing urinary stone disease in abdominal low-dose CT.