Objective: The objective of our study was to compare the diagnostic performance of 16-MDCT with that of MRI in the characterization of kidney lesions.
Subjects and methods: Twenty-eight patients with kidney lesions detected with sonography and requiring further evaluation were examined. MDCT was performed in the unenhanced, arterial, and portal venous phases. MRI was performed at 1.5 T with T2- and T1-weighted and dynamic gadolinium-enhanced sequences. Consensus reading was done by two radiologists. Image quality was rated on a four-point scale. Classification of lesions as surgical or nonsurgical was done with five levels of confidence, and it was required that a definite diagnosis be assigned to each lesion. The 1997 TNM classification was used for staging. Statistical analysis was done by receiver operating characteristic analysis or paired Student's t test. Histologic or follow-up findings at least 12 months after the primary diagnosis served as the standard of reference.
Results: The image quality of MDCT (mean grade, 2.79 on a 0-3 scale) was superior to that of MRI (1.93; p < 0.01). The area under the curve for differentiating surgical from nonsurgical lesions was 0.979 for MDCT and 0.957 for MRI with resulting sensitivity and specificity values of 92.3% and 96.3% for MDCT and 92.3% and 91.3% for MRI. Sensitivity and specificity for definite classification of the lesions were 93.8% and 68.4% for MDCT and 93.8% and 71.4% for MRI.
Conclusion: Both MDCT and MRI are excellent for differentiating surgical from nonsurgical kidney lesions. Both methods have low specificity for the differentiation of benign from malignant lesions.