Objective To explore the diagnostic value of MRI radiomics model for clinically significant prostate cancer(csPCa)in the transition zone with prostate imaging reporting and data system(PI-RADS)3-4.Methods According to the ratio of prostate cancer(PCa)∶benign prostatic hyperplasia(BPH)=1∶1,the relevant data of patients with transition zone lesions confirmed by pathology were retrospectively collected.Patients were divided into csPCa group and non-csPCa group according to pathological results.Region of interest(ROI)were drawn along the edge of the lesion in T2WI,diffusion weighted imaging(DWI),and apparent diffusion coeffi-cient(ADC)sequences,the radiomics features were extracted,and a single sequence logistic regression model was constructed and validated after feature cleaning and screening.The diagnostic performances of the radiomics model for transition zone csPCa were evaluated and compared.Results A total of 204 patients were included,and there were 94 cases in the csPCa group and 110 cases in the non-csPCa group.There were statistically significant differences in PI-RADS,prostate volume(PV),prostate specific antigen(PSA)and its derivative indicators between csPCa and non-csPCa groups.In the validation set,the area under the cunve(AUC)of the ADC model was 0.766,and the DeLong test showed that the diagnostic performance of the ADC model was significantly higher than that of the DWI and T2WI models.Conclusion When the PI-RADS classification of transition zone lesion is 3-4,the ADC radiomics model has good diag-nostic performance for csPCa,which can reduce unnecessary biopsy in some patients.Radiomics is expected to become a more objec-tive and accurate medical diagnostic tool.