Uterine cervical carcinoma: preoperative staging with 3.0-T MR imaging--comparison with 1.5-T MR imaging.

Purpose: To prospectively evaluate the efficacy of 3.0-T magnetic resonance (MR) imaging in the preoperative staging of cervical carcinoma compared with that at 1.5-T imaging, with surgery and pathologic analysis as the reference standards.

Materials And Methods: Institutional review board approval and informed consent were obtained. Thirty-one consecutive patients (age range, 27-71 years; mean age, 51.1 years) underwent 3.0- and 1.5-T MR imaging. Quantitative and qualitative analyses were performed. Two radiologists independently evaluated images in terms of local-regional staging. MR findings were compared with surgicopathologic findings.

Results: Mean tumor signal-to-noise ratios, mean cervical stroma signal-to-noise ratios, and mean tumor-to-cervical stroma contrast-to-noise ratios at 3.0-T imaging were significantly higher than those at 1.5-T imaging (P = 9.1 x 10(-6), P = 1.8 x 10(-6), and P = .008, respectively). Image homogeneity at 3.0-T imaging was significantly inferior to that at 1.5-T imaging (P = .005). There were no significant differences in terms of the degree of susceptibility artifacts. Interobserver agreement between the two radiologists for local-regional staging was good or excellent (kappa = 0.65-0.89). Sensitivity, specificity, and area under the receiver operating characteristic curve for radiologist 1 in the evaluation of parametrial invasion were (a) 75% for both 3.0- and 1.5-T imaging, (b) 70% for both 3.0- and 1.5-T imaging, and (c) 0.82 for 3.0-T imaging and 0.85 for 1.5-T imaging, respectively. Corresponding values for vaginal invasion were (a) 67% for both 3.0- and 1.5-T imaging, (b) 68% for 3.0-T imaging and 72% for 1.5-T imaging, and (c) 0.62 for 3.0-T imaging and 0.67 for 1.5-T imaging, respectively. Corresponding values for lymph node metastases were (a) 57% for both 3.0- and 1.5-T imaging, (b) 83% for 3.0-T imaging and 88% for 1.5-T imaging, and (c) 0.72 for 3.0-T imaging and 0.78 for 1.5-T imaging, respectively. Neither radiologist noted significant differences between values obtained with 3.0-T imaging and those obtained with 1.5-T imaging (P > .5 for all comparison pairs).

Conclusion: In this study, 3.0-T MR imaging was characterized by high diagnostic accuracy in the presurgical evaluation of patients with cervical carcinoma, although 3.0-T imaging was not significantly superior to 1.5-T imaging.