Assessment of salivary gland tumors using MRI and CT: impact of experience on diagnostic accuracy

Rationale and objectives: To analyze the accuracy of radiological diagnosis in MRI and CT studies of salivary gland tumors depending on the radiologist’s experience. Materials and methods: Three radiologists with differing experience (R1 > 20, R2 > 11, and R3 > 7 years, respectively) retrospectively reviewed 128 cases (116 MRI, 12 CT studies) with suspected salivary gland tumors regarding dignity and classification using histopathology as a reference standard. Sensitivity, specificity, positive/negative predictive value and inter-observer agreement (using Cohen’s κ) were calculated to compare diagnostic performance. Results: Lesions were benign in 87 and in 23 cases malignant. Neoplasia was absent in 18 cases (15 cases without neoplasia and 3 cases without disease). The highest inter-observer agreement for determining dignity using CT was found between R1 and R2 (κ = 0.74, p < 0.001), and the lowest between R2 and R3 (κ = 0.28, p < 0.001). MRI sensitivity/specificity for classifying pleomorphic adenomas was as follows: R1 (100%/100%), R2 (76.92%/87.01%), R3 (43.53%/67.53%), and for CT: R1 (100%/100%), R2 (100%/88.89%), R3 (66.67%/88.89%; for Warthin’s tumor using MRI: R1 (100%/97.44%), R2 (68.42%/83.33%), R3 (50.00%/67.95%), and using CT: R1 (100%/100%), R2 (50.00%/100%), R3 (100%/100%; for squamous cell carcinomas using MRI: R1 (100%/100%), R2 (75.00%/97.12%), R3 (75.00%/99.04%), and using CT: R1 (100%/100%), R2 (66.67%/88.89%), R3 (66.67%/66.67%). The highest agreement was found between R1 and R2 for MRI (κ = 0.62, p < 0.001), and the lowest between R1 and R3 at MRI (κ = 0.28, p < 0.001). Conclusion: Diagnostic accuracy in the assessment of salivary gland tumors strongly depends on the observer’s expertise and increases with higher experience.