[Reliability and interobserver concordance in detection of prostatic carcinoma with magnetic resonance tomography].
PURPOSE: To determine the reliability and the measuring agreement between two radiologists in the identification of relevant structures of prostate anatomy in patients with prostate carcinoma. PATIENTS AND METHOD: 15 patients with prostate carcinoma proved by histology underwent MR Imaging on 1.5 Tesla MR-system (Siemens Erlangen). T1- and T2-weighted sequences were used in the depiction of important anatomic structures for diagnosing prostate cancer. 5 sequences of every patient were evaluated for 5 times by each reader. Reliability was proved by the intraclass correlation coefficient and the measuring agreement was proved by Cohen's cappa test. RESULTS: Only the T2-weighted sequence was of sufficient sensitivity to reach a moderate reliability (r1 = 0.74) and a moderate measurement agreement between the two readers (kappa = 0.5). CONCLUSION: For the evaluation of MR images in staging prostate carcinoma, besides the validity and the measuring agreement, the reliability of the radiologist is the important factor.