RESUMO
PURPOSE: To evaluate the added value of diffusion-weighted imaging (DWI) to T2-weighted imaging (T2WI) for improved identification of pelvic lymph nodes (LN) by radiation oncologists. METHODS/PATIENTS: This retrospective study included 20 patients with histopathologically proven node-negative prostate cancer. All patients underwent 3T-MRI of the prostate; matched axial T2WI and DWI sequences were assessed by an experienced uro-radiologist as the reference standard. Consultant and specialist registrar radiation oncologists were asked to identify all LN first on T2WI alone (read 1) and then on T2WI and DWI combined (read 2); LN were measured in size and divided into true positives (TP), false positives (FP) and false negatives (FN). Sensitivity, positive predictive value (PPV) and false negative rate (FNR) were then calculated and compared using Pearson's Chi square test. RESULTS: A total of 177 LN comprised the reference standard. 16 TP, 16 FP and 161 FN LN (sensitivity 9.0%, PPV 50.0%, FNR 91.0%) and 124, 15 and 53 LN (70.1%, 89.2%, 30%) were identified by reader 1 on reads 1 and 2, respectively; χ2 (2, N = 385) = 137.8, p < 0.0001. 27, 21 and 150 LN (15.3%, 56.3%, 84.8%) and 120, 13 and 57 LN (67.8%, 90.2%, 32.2%) were identified by reader 2 on the two reads; χ2 (2, N = 388) = 102.4, p < 0.0001. CONCLUSIONS: Adding DWI to T2WI significantly improved identification of pelvic LN by radiation oncologists and can therefore be regarded as a useful LN contouring technique for RT planning in pelvic malignancies.