Evaluation of solid pulmonary nodules in patients with bladder cancer with computed tomography based texture analysis method: Is it possible to differentiate metastatic - non-metastatic nodules?

Abstract

Solid Pulmonary Nodule (SPN) is defined as parenchymal radiopacity smaller than 3 cm in diameter. Evaluating the metastatic nature of the SPNs detected in the thorax computed tomography (TCT) examination for staging purposes in cancer patients becomes a fundamental problem for the physician. Invasive procedures, additional imaging or follow-up imaging, are often used to differentiate metastatic and non-metastatic nodules. In this study, we aimed to distinguish SPNs detected in patients diagnosed with bladder cancer (BC) as metastatic and non-metastatic nodules by texture analysis. TCT images of patients diagnosed with BC in our hospital from January 2007 until December 2017 were retrospectively evaluated. A total of 46 patients with SPN, including metastatic (n= 19) and non-metastatic (n= 27), were included in the study. Short axis diameter, long-axis diameter, nodule volume and volume histogram values of the nodules were obtained. Chisquare test was used to evaluate dependent variables, and the Mann-Whitney U test was used to evaluate independent variables. ROC curves of the obtained data were plotted. Statistically, the significant p-value was determined as less than 0.05. A significant difference was found between SPN long axis, short axis and volume values. In the volumetric histogram analysis, the maximum density value and the mean density value were found to be statistically significant. When the average of the highest densities in the volume histogram data was evaluated, the area under the curve value was 0.702 (95% CI, 519-854). The metastatic nodule could be distinguished with a sensitivity of 88% and a specificity of 70% when the volume histogram has the maximum density threshold of 50 HU. In this study, we concluded that SPN detected on CT images can be distinguished as metastatic and non-metastatic nodules using texture analysis method without invasive procedures.