RESUMO
Objective To explore the value of readout segmentation of long variable echo-trains (RESOLVE) in the differentiation of prostate cancer from benign prostatic hyperplasia (BPH).Methods Seventy two consecutive patients with suspected prostate cancer were evaluated by 3.0 T MR examination (RESOLVE sequence included,b values=0 and 800 s/mm2) were included in our retrospective study.All the patients had ultrasound guided systemic biopsy with histopathological diagnosis.The patients were divided into group A (23 prostate cancer cases with total 43 malignant lesions) and group B (49 BPH cases with total 64 benign lesions).Two radiologists who were blinded to the clinical data quantitatively analyzed the ADC values of suspicious lesions independently.Inter-reader agreement for ADC values was assessed with Bland and Altman test,and the intra-class correlation coefficient (ICC).Difference of ADC values in two groups was assessed by student's t test.Receiver operating characteristic curve (ROC) was used to determine the best predictor and cutoff value.Results A total of 107 lesions (43 malignant and 64 benign) were identified in 72 patients.ICC was 0.976,P＜0.01.The mean ADC value of prostate cancer is lower than BPH (t=19.223,P＜0.01),(0.74±0.12) × 10 3 and (1.21±0.12) × 10-3mm2/s respectively.Diagnostic cut-off point was 0.946× 10-3mm2/s,diagnostic sensitivity 95.3 ％ (41/43),specificity 98.4％ (63/64),accuracy 97.2％ (104/107).Conclusion RESOLVE ADC value is valuable in the differential diagnosis of prostate cancer and BPH.