The Value Of Transrectal Ultrasound Texture Analysis In Differentiating Benign And Malignant Prostate Diseases

Objective: To investigate the diagnostic value of texture analysis derived from transrectal ultrasound in differentiating benign and malignant prostate diseases.Methods: A retrospective analysis of 125 clinically suspected prostate cancer(PCa)patients admitted to the First Affiliated Hospital of Guangxi Medical University from February 2018 to June 2020.All patients underwent transrectal gray-scale ultrasound of prostate and contrast-enhanced ultrasound,followed by transrectal prostate biopsy.Pathologically confirmed 84 patients with PCa and 41 patients with benign prostate lesions.Compared with the range of suspicious lesions in the outer glands found by contrast-enhanced ultrasound,the region of interest(ROI)of the lesions were manually delineated on the gray-scale and contrast-enhanced ultrasound images of the transrectal prostate at the peak time of the lesion enhancement,and extract the gray-scale and contrast-enhanced ultrasound images texture features respectively.Use mutual information(MI),Fisher coefficient(Fisher),Classification Error Probability Combined Average Correlation Coefficients(POE+ACC)and 3 methods combined(FPM)screening texture features.Then,Raw Data Analysis(RDA),Principal Component Analysis(PCA),Linear Discriminant Analysis(LDA)and Nonlinear Discriminant Analysis(NDA)in B11 software package were used for texture classification,and the diagnostic efficiency is expressed by the misdiagnosis rate.The texture characteristics screened by FPM were analyzed for differences between groups.Receiver operating characteristic curve(ROC)analysis was performed on the features with significant differences between groups.The best texture feature of contrast-enhanced ultrasound and transrectal gray-scale ultrasound were selected for single and combined diagnosis,and the diagnostic efficiency of the best texture feature combined diagnosis and the diagnosis of PCa alone was compared.Results: The lowest misdiagnosis rates of transrectal gray-scale ultrasound and contrast-enhanced ultrasound texture analysis were 10.4% and 6.4%,respectively,using FPM combined with NDA.Gray-level co-occurrence Matrix(GLCM)is the main source of texture features screened by FPM method.By testing the differences in texture features between the two groups and comparing the statistically significant differences between the two groups,the area under the curve(AUC)showed that the texture features with AUC>0.75 all came from GLCM.The optimal texture feature of transrectal grey-scale ultrasound images of prostate was S(0,1)Entropy,with AUC,sensitivity and specificity of 0.773,65.48% and 85.37%,respectively.The optimal texture features of transrectal contrast-enhanced ultrasound images were S(5,-5)Dif Entrp,and AUC,sensitivity and specificity were 0.817,91.67% and 60.98%,respectively.S(5,-5)Dif Entrp and S(0,1)Entropy alone showed no significant difference in AUC in the diagnosis of PCa(P>0.05),and the combination of the best texture features of the two had the highest diagnostic efficiency for PCa(P<0.05).Conclusion: 1.Transrectal ultrasound image texture analysis of the prostate can be used as an auxiliary tool for clinical diagnosis of PCa.2.The texture features of the gray-level co-occurrence matrix extracted by transrectal ultrasound image texture analysis can help distinguish benign and malignant prostate lesions.3.The combination of optimal texture features,derived from transrectal gray-scale ultrasound of prostate and contrast-enhanced ultrasound,has high diagnostic efficacy for PCa and has potential clinical application value.