Inter-reader agreement of multi-parametric MR imaging for the detection of prostate cancer: evaluation of a scoring system.

PURPOSE: Functional prostate MR is performed in varying combinations of T2-weighted images with diffusion-weighted imaging (DWI), dynamic contrast-enhanced MRI (DCE-MRI), and spectroscopic imaging (MRSI). Recently, a European consensus meeting proposed the use of a simple 5-point scale for estimating the probability of a lesion being malignant. The aim of the present study was to determine the inter-reader agreement of MR imaging using a scoring system based on the recommendations of the consensus.
MATERIALS AND METHODS: The appearance of 108 predefined lesions in three different MR sequences (T2-weighted images, DWI, and DCE-MRI) in 50 functional prostate MR examinations were retrospectively scored by three blinded radiologists using a 5-point scale for each MR sequence. After scoring T2/DWI and T2/DWI/DCE-MRI, every lesion was graded based on its probability for malignancy. The inter-observer reliability was evaluated using Kappa statistics (ĸ).
RESULTS: With respect to T2-weighted images, DWI and DCE-MRI ĸ was 0.49, 0.97, and 0.77, respectively. Combined scoring of T2-weighted images and DWI demonstrated correct tumor diagnosis (true positive) in 71-88% (depending on reader) of cases (ĸ=0.78). The accuracy was further improved to 88-96% after scoring all three MR sequences including DCE-MRI (ĸ=0.90).
CONCLUSION: The use of a simple 5-point scoring system for T2-weighted images, DWI, and DCE-MRI is feasible in functional prostate MRI and has high inter-observer reliability. © Georg Thieme Verlag KG Stuttgart · New York.