Resource Centre

549

Validation of automated kidney stone volumetry in low dose computed tomography

By:

Wilhelm K.1, Hein S.1, Schlager D.1, Adams F.1, Miernik A.1, Schoenthaler M.1, Hesse A.2, Neubauer J.3

Institutes:

1Faculty of Medicine and Medical Center - University of Freiburg, Center For Surgery Department of Urology, Freiburg, Germany, 2University of Bonn, Department of Urology, Division of Experimental Urology, Bonn, Germany, 3Faculty of Medicine and Medical Center - University of Freiburg, Department of Radiology, Freiburg, Germany

Introduction & Objectives

Stone desintegration and fragment extraction time depend on the stone volume. However, in the vast majority of clinical trials on urolithiasis stone size is assessed as maximum diameter. This neglects the fact that most stones are not spherical, resulting in potential bias and misleading results. Nowadays software solutions can provide an automated assessment of maximum diameter and volume by just clicking on the stone of interest on an existing CT scan. The objective of the study was a validation of the software at our institution.

Material & Methods

109 human and animal kidney and bladder stones (kindly provided by The German Stone Center Bonn) with a diameter of at least 5mm were measured by three urologists independently, assessing the maximum diameter using a digital sliding caliper and the volume using a water displacement/overflow method. We calculated the interrater correlation with a Single Score Intraclass Correlation Model and used the mean values as reference.
The same stones were then positioned in a water bath with human vertebrae simulating a human body and CT scans were acquired at low dose settings with 100KVp, 55mAs, CTDIvol 2.3mGy for the axial and coronal reconstructions with 3mm slice thickness and slice sparing (kernel: I31f) and applying dual energy technique with 100KVp 172mAs/ Sn140KVp 122mAs, CTDIvol12.94 mGy for axial reconstructions with 0.75mm slice thickness and slice sparing (kernel: Q30f). Three radiologists measured the maximum diameter and volume using the software for automated measurement with 0.75mm reconstructions. 4 weeks later the radiologists measured the diameter in the 3mm reconstructions by hand. In order to assess the value of the automated stone volumetry we calculated the expected volume using the radiological maximum diameter and the formula: V = 4/3 x Pi x r3 and compared the calculated volume to the measured volume.

Results

13 stones had to be excluded from further analysis (crumbling= 9/ the automated assessment did not work= 4). The interrater correlation for the reference measures was 0.99 for the diameter and volume. The mean reference diameter was 13.3mm (5.2-32.1). The correlation of the automated radiologic diameter with the reference value was 0.91. The radiologic assessment by hand correlated significantly better (0.98, p=<0.001). The mean reference volume was 1.18ml (0.01-9). The mean automated radiologic volume was 1.06ml (0.01-7.69) and the mean calculated volume based on the radiologic diameter was 2.25ml (0.02-18.81). The correlation of the reference value with the automated radiologic volume assessment was 0.99 and significantly better compared to the calculated values based on hand measurements (0.95, p=<0.001).

Conclusions

Automated measurement of stone volume based on CT scans is possible and the accuracy is significantly higher as compared to volumetric calculations based on the diameter. Such calculations depend on the shape of the stone. In order to avoid bias in clinical trials, size should be measured as volume or a combination of diameter and volume. However, automated diameter measurement is not as accurate as the measurements by hand. Thus, measurement of the diameter by hand still seems mandatory for the evaluation of urolithiasis.

  • Type: Abstract
  • Date: 26-03-2017
  • Rating: 0,0
  • Views: 326
  • Event: 32nd Annual EAU Congress London
  • Nr: 549
  • Session: Percutaneous nephrolithotomy
  • Location: Sunday, 26 March 2017, 14:00 - 15:30, Room Paris, North Hall (Level 1)