Pitfalls in urinary stone identification using CT attenuation values: are we getting the same information on different scanner models?

Abstract

Introduction: Evaluate the capability of different Computed Tomography scanners to determine urinary stone compositions based on CT attenuation values and to evaluate potential differences between each model.

Methods: 241 human urinary stones were obtained and their biochemical composition determined. Four different CT scanners (Siemens, Philips, GEMS and Toshiba) were evaluated. Mean CT-attenuation values and the standard deviation were recorded separately and compared with a t-paired test.

Results: For all tested CT scanners, when the classification of the various types of stones was arranged according to the mean CT-attenuation values and to the confidence interval, large overlappings between stone types were highlighted. The t-paired test showed that most stone types could not be identified. Some types of stones presented mean CT attenuation values significantly different from one CT scanner to another. At 80kV, the mean CT attenuation values obtained with the Toshiba Aquilion were significantly different from those obtained with the Siemens Sensation. On the other hand, mean values obtained with the Philips Brilliance were all significantly equal to those obtained with the Siemens Sensation and with the Toshiba Aquilion. At 120kV mean CT attenuation values of uric acid, cystine and struvite stones obtained with the Philips model are significantly different from those obtained with the Siemens and the Toshiba but equal to those obtained with the GE 64.

Conclusions: According to our study, there is a great variability when different brands and models of scanners are compared directly. Furthermore, the CT scan analysis and HU evaluation appears to gather insufficient information in order to characterize and identify the composition of renal stones.