Application of pelvis multi-parameter MRI diagnostic model for the diagnosis of prostate cancer in early stage

Abstract

Objective To explore the effectiveness and clinical value of pelvic multi-parameter MRI diagnostic model in the diagnosis of prostate cancer in early stage. Methods the clinical data of 260 patients whose PSA<20μg/L were Collected from October 2013 to November 2014 in Beijing hospital. All the patients were first evaluated by MRI diagnostic model which consists of the imaging presentation from T2 weighted imaging (T2WI), Diffusion weighted imaging (DWI), and the both together in order to acquire the likelihood of prostate cancer, and followed by systemic prostate biopsy in order to verify the diagnosis above. Results Among 260 patients, a total of 107 cases of patients were diagnosed prostate cancer via trans-rectal fine needle biopsy, the positive rate was 41.15%. the scores of T2WI, DWI and T2WI + DWI significantly correlated with the rate of positive biopsy, from 6.7% to 65.6%, 14.4% to 60.7% and 13.5% to 55.9% respectively, P<0.05. Receiver Operating Characteristic (ROC) curves were made based on the T2WI, DWI and T2WI + DWI diagnostic model, whose Area Under Curve(AUC)s were 0.651 ± 0.034, 0.744 ± 0.031 and 0.728 ± 0.031 respectively. To take the 1 score in the T2WI model as the baseline, from 2 score to 5 score, the risk of prostate cancer incidence was 2.933, 3.882, 5.005and 7.944 times respectively. To take the 1 score in the DWI model as the as the baseline, from 2 score to 5 score, the risk of prostate cancer incidence was 2.233, 8.136, 8.614 and 18.313 times respectively. In the T2WI + DWI group, the risk of incidence was 1.333, 6.228, 8.305 and 9.996 times respectively based on the 1 score of the model itself. Conclusion This model has the capacity for enhancing the accuracy of diagnosis the prostate cancer in the early stage, which could reduce the incidence of misdiagnosis and over diagnosis at same time. Key words: Prostate cancer; Early diagnosis; Magnetic resonance (MRI); Diagnostic model