RESUMO
OBJECTIVE: The aim of the study was to evaluate dynamic contrast-enhanced magnetic resonance (MR) imaging in the characterization of parotid gland tumors. METHODS: Fifty-five parotid lesions in 55 patients were retrospectively included. Two observers interpreted 2 reading protocols derived from all MR imaging in 2 distinct sessions, independently and blinded. Benign versus malignant distinction was carried out for protocol 1 (without contrast administration) and protocol 2 (with dynamic contrast-enhanced sequence). Histopathological results after surgical resection were used as the criterion standard. Diagnostic accuracy was compared between protocols using McNemar test. A P values of less than 0.05 indicated significant difference. RESULTS: There was no intraobserver statistical discordance between protocols for both observers (P = 0.27 and P = 1). Interobserver reliability showed moderate agreement for protocol 1 (κ = 0.591; 95% confidence interval [CI], 0.376-0.806) and 2 (κ = 0.463, 95% CI, 0.226-0.701). Intraobserver reliability showed moderate agreement for observer 1 (κ = 0.507; 95% CI, 0.279-0.736) and 2 (κ = 0.477; 95% CI, 0.241-0.712). CONCLUSIONS: Magnetic resonance imaging protocol including dynamic sequence for the characterization of parotid gland lesion yielded nonsignificant increases in sensitivity, specificity, or positive predictive values, and negative predictive values over noninjected protocol.