The Clinical Application Of CT Radiomics In Bladder Cancer's Preoperative Diagnosis

Objective The purpose of this study is to explore the relationship between CT radiomics features and the histological grade,pathological stage and prognosis based on urthalial carcinoma of the bladder(UCB),so as to provide relevant basis for the treatment selection for patients,monitoring tumor progress and recurrence,and provide more valuable reference for predicting the histological grading,pathological staging and prognosis.Methods The study subjects selected 99 patients with bladder tumor admitted to the Department of Urology of Baotou Central Hospital from June 2017 to June 2019.According to the histological grade of bladder cancer,the study subjects were divided into 63 cases of high-grade urothelial carcinoma(HGUC)and 36 cases of low-grade urothelial carcinoma(LGUC)(WHO2004 Grading Method).There were 8 cases of highly differentiated Grade I,47 cases of moderately differentiated Grade II and 44 cases of poorly differentiated Grade III(WHO1973 Grading Method).According to the pathological stages of bladder cancer,72 cases were classified as muscle-invasive bladder cancer(MUSCLE-INVASIVE Bladder Cancer MIBC)and 27 cases as non-muscle-invasive bladder cancer NMIBC.These patients were treated by transurethral resection of bladder tumor(TURBT),and the postoperative pathological results confirmed bladder urothelial carcinoma.The patients in the group were scanned on the same CT machine(Siemens Sensation Open)with the same scanning parameters before operation.The image with the largest cross-sectional area of tumor is selected for region of interest(ROI)delineation and extraction,and the selected image is imported into matlab2017 a in DICOM format for texture analysis.The mean,standard deviation,entropy,kurtosis and skewness of the image's texture features are analyzed.SPSS 20.0(SPSS Inc.,Chicago,USA)software was used for statisticalanalysis.Statistical analysis took bilateral tests,and significance test level ?=0.05.Results Compared with HGUC,the mean value(P=0.012)and entropy(P=0.019)of LGUC texture parameters are lower.The area under the curve of entropy in distinguishing HGUC from LGUC is the largest,AUC = 0.737(95% CI: 0.645-0.839),sensitivity 71.43%,specificity 61.11%,and Jordan index 0.33.Compared with MIBC,the entropy(P=0.004)and mean value(P=0.006)of the texture parameters of NMIBC are lower,and the area under the curve of entropy is the largest in distinguishing MIBC from NMIBC,with AUC=0.723(95% CI: 0.611-0.836),sensitivity 61.11%,specificity 71.07%,and Jordan index 0.35.In Grade I,Grade II and Grade III groups,the average value(P=0.002)and entropy(p<0.001)of texture parameters of Grade II and Grade III are higher than those of Grade I,and the maximum AUC under the curve identifying entropy of texture parameters in Grade I,Grade II and Grade III is0.843(95% CI: 0.707-0.979),the sensitivity is 95.60%,the specificity is 62.50%,and the index is about 0.58.Conclusion The texture analysis method of imaging group based on CT radiomics is helpful to distinguish the stage and grade of bladder urothelial carcinoma before operation,and the entropy of texture parameter may be the best diagnostic parameter.