Fast Breath-Hold T2-Weighted MR Imaging Reduces Interobserver Variability in the Diagnosis of Adenomyosis

Abstract

OBJECTIVE: We compared two rapid MR imaging T2-weighted pulse sequences with high-resolution turbo spin-echo for the diagnosis of adenomyosis, and we evaluated interobserver variability. SUBJECTS AND METHODS: Fifty-six consecutive patients referred for hysterectomy prospectively underwent MR imaging. Two fast pulse sequences using a breath-hold technique-true fast imaging with steady-state free precession (FISP) and turbo inversion recovery-and turbo spin-echo T2-weighted images of the pelvis were obtained in each patient. The images were analyzed in a blinded manner and independently by three reviewers with different levels of experience for the accuracy of adenomyosis diagnosis, image quality, anatomic visualization, and image artifacts. The accuracy for the diagnosis of adenomyosis on turbo spin-echo T2-weighted imaging combined with one or two fast pulse sequences was evaluated for each reviewer. RESULTS: Twenty-four patients (42.9%) had a histologic diagnosis of adenomyosis. The accuracy for the diagnosis of adenomyosis for reviewers 1, 2, and 3 using turbo spin-echo T2-weighted, true FISP, and turbo inversion recovery sequences was 83.9%, 67.8%, 75%; 83.9%, 67.8, 78.5%; and 87.5%, 73.2%, and 75%, respectively. A difference in the accuracy rate was found among the observers for the three sequences (p < 0.001). Whatever the pulse sequence, the accuracy rate was higher for the reviewer with more experience in gynecologic imaging. The combination of turbo spin-echo T2-weighted imaging with at least one rapid sequence increased the accuracy of observers with little experience in gynecology. With turbo inversion recovery sequences, the image quality score was low for the three reviewers compared with turbo spin-echo T2-weighted and true FISP sequences. The combination of turbo spin-echo T2-weighted and true FISP sequences gave the highest image quality scores. CONCLUSION: Breath-hold T2-weighted sequences optimize the accuracy of MR imaging for the diagnosis of adenomyosis and reduce interobserver variability.