Bone CT Radiomics is Feasible in Identifying CKD5

Introduction: We aimed to investigate the feasibility of bone CT radiomics in identifying CKD5. 120 chronic kidney disease (CKD) patients (60 CKD5 and 60 CKD1) were assessed using the bone CT radiomics method. Radiomics features of the vertebral CT images were obtained by using 3D Slicer software, and then were compared between CKD5 and CKD1 cases. Logic regression model was used for the features with significance. The receiver operating characteristic (ROC) curve was used to determine the performance in identifying CKD5. Mean CT density was not significantly different between the two groups (P=0.12). CKD1 and CKD5 cases differed in the following features: Contrast, Correlation, Dependence Variance, and Dependence Entropy (P<0.05). The AUC of the regression model was 0.854. The sensitivity and specificity of the model in identifying CKD5 were 88% (53/60) and 87% (52/60) respectively.

Conclusion: Bone CT radiomics is feasible in identifying CKD5.