The evaluation of agreement between the measurement methods used in the diagnosis of prostate cancer with the statistical methods

Abstract

The aim of this study is to evaluate the compatibility of the "Computed Tomography (CT)" and "Magnetic Resonance (MR)" methods which are imaging techniques used to describe prostate cancer with the pathology accepted as the reference method. In this study, the concordance between CT and MR results and pathology results of 37 prostate cancer patients was evaluated using the Bland-Altman, Interclass Correlation Coefficient, Concordance Correlation Coefficient, Deming Regression and Passing-Bablok method comparison methods. Inter-class correlation coefficient and Concordance Correlation Coefficient values for CT-Pathology results were 0.62 and 0.62, respectively. Inter-class correlation coefficient and Concordance Correlation Coefficient values for MR-Pathology results were 0.74 and 0.75, respectively. The regression equation for the Deming regression method is y=-6.21+1.03x for CT-Pathology, whereas for the MR-Pathology y=-6.86+1.11x. When the CT-Pathology measurement values are evaluated by the Bland-Altman statistical method, the mean values of the measurement differences are -2.42 and the standard deviation is 33.50. When the agreement between MR-Pathology measurement values is examined by the Bland-Altman method, the mean values for the difference between the measurement values are 4.14 and the standard deviation is 26.78. Among the applied methods, the Bland-Altman method is the most suitable method for data structure. According to the results of the Bland-Altman method, the tumor size obtained by the pathology method can be found to be 2.4 smaller than the mean values obtained from the CT results and an average of 4.1 cc from the measurement values obtained by the MR imaging technique.