Category: New Technology: Stones
Introduction & Objective :
There is growing interest in analysing CT images of renal tract stones using CT texture analysis (CTTA) software to infer information about the stone characteristics and architecture. This information may help patients and urologists decide between different treatment modalities for the stone. We aimed to explore the meaning and utility of CTTA by correlating CTTA metrics with other known clinical measures that can affect outcome.
Using commercially available CTTA software, we analysed 371 CT images that initially diagnosed solitary renal and ureteric stones, from a single centre. The software populates a region of interest for the largest cross-sectional slice of the stone and produces metrics based on statistical analysis of the Hounsfield unit (HU) values of all the available pixels. These metrics were: mean, standard deviation, entropy, kurtosis and skewness of the HU values, and the total number of pixels.
The total number of pixels measured using the CTTA software was strongly linearly correlated to stone volume measured using the ellipsoid formula (r=0.83), and strongly curvi-linearly correlated to the major axis length (r=0.87). The more linear correlation with stone volume suggests that total number of pixels may more accurately represent stone burden for larger stones than axis length. Entropy, is a CTTA measure of the degree of randomness within the pixel distribution of the stone image. Entropy was also strongly correlated to the total number of pixels (r=0.89) and the major axis length (r=0.84). To investigate the influence of CTTA metrics on ease of fragmentation using shockwave lithotripsy, multivariable analysis was performed using traditional factors and CTTA variables to predict for a stone free outcome after lithotripsy. This revealed that the CTTA variables total number of pixels and entropy were often preferentially selected above other measures of size to be included in the model. Kurtosis, which is a measure of the weight of the tails, relative to the rest of the distribution of HU values, was not intercorrelated with other CTTA variables and also chosen for inclusion in multivariable models, suggesting it is measuring a property not related to HU density or stone size.
Our results show that CTTA variables appear to represent stone size and structure. The utility of using CTTA may lie in the advantage of being able to quantify stone burden in an objective and reproducible way. This can help standardise outcome measures when comparing efficacy of different stone treatments.
Ben Turney– Bernard Senior Clinical Researcher in Urology, Nuffield Department of Surgical Sciences, Oxford, England, United Kingdom
Helen Cui– Clinical Researcher, Urology Department, Oxford University Hospitals NHS Foundation Trust, Oxford, England, United Kingdom
Mafalda Silva– Lisbon, Lisboa, Portugal
Balaji Ganeshan– London, England, United Kingdom
Bernard Senior Clinical Researcher in Urology
Nuffield Department of Surgical Sciences
Oxford, England, United Kingdom
Urology Department, Oxford University Hospitals NHS Foundation Trust
Oxford, England, United Kingdom