Low-dose computed tomography of urolithiasis in obese patients: a feasibility study to evaluate image reconstruction algorithms

De-Hua Chang,¹ Karin Slebocki,² Ekaterina Khristenko,¹ Jan Herden,³ Johannes Salem,³ Nils Große Hokamp,² Kamal Mammadov,² Martin Hellmich,⁴ Christoph Kabbasch<sup>2

1</sup>Department of Diagnostic and Interventional Radiology, University Medical Center Heidelberg, Heidelberg, Germany; ²Department of Diagnostic and Interventional Radiology, University Hospital of Cologne, Cologne, Germany; ³Department of Urology, University Hospital of Cologne, Cologne, Germany; ⁴Institute of Medical Statistics and Computational Biology (IMSB), Medical Faculty, University of Cologne, Cologne, Germany

Purpose: Retrospective evaluation and comparison of image quality generated by low-dose computed tomography (LDCT) from obese patients with urolithiasis using alternative reconstruction algorithms.
Materials and methods: Twenty-five obese patients (body mass index [BMI]>25 kg/m^2,) underwent LDCT scans for suspected urolithiasis. The scans were recompiled using filtered-back projection (FBP), statistical iterative reconstruction (iDose) and iterative model-based reconstruction (IMR). Dose-length product (DLP) and patient details were obtained from the CT dose report and clinical charts, respectively. Objective image noise was assessed by measuring the SD of Hounsfield units (HUs) in defined locations. Additionally, subjective image evaluation was independently performed by two radiologists using a 3-point Likert scale. The inter-reviewer agreement of image quality was calculated.
Results: Ureteral concretions were observed in all CT scans, two of which revealed bilateral stones. The assessed patients’ mean BMI was 29.29±3.74 kg/m,² and the DLP of the CT scans was 100.04±10.00 mGy*cm. All scans were rated diagnostic with the iDose and iterative model-based reconstructions, whereas 41% of the scans performed with FBP reconstruction were nondiagnostic. With respect to image quality, IMR was superior to iDose and FBP, both in the objective (P<0.001) and overall subjective (P≤0.008) evaluation of the respective data sets. The inter-reviewer agreement for overall image quality was “almost perfect” for IMR, “substantial” for iDose and “moderate” for FBP (κ values of 1.0, 0.6 and 0.46, respectively).
Conclusion: Using iterative image reconstruction algorithms, LDCT of urolithiasis is feasible in overweight patients with a BMI between 25 and 35 kg/m^2,. Due to higher image quality, IMR is the preferred algorithm for scan reconstruction as it may help to avoid repeated examinations due to initial nondiagnostic scans.

Keywords: low-dose computed tomography, model-based iterative reconstruction, statistical iterative, urolithiasis, obesity