Abstract
Objectives:
Three-dimensional magnetic resonance imaging (3D-MRI) system has already performed at cardiovascular surgery area, and it’s useful for improving the reliability of surgical planning and informed consent for patients. However, it is difficult to construct 3D-MRI of rotator cuff tendon because of existing similar intensity area for other tendon and muscular tissue. The purpose of this study is to construct 3D-MRI of rotator cuff tendon from 2D-MRI and compare the accuracy of rotator cuff tear characters between 2D-MRI and 3D-MRI. We hypothesized that 3D-MRI might be more accurate than 2D-MRI for classifying tear shape and size.
Methods:
An MRI scan was carried out on 28 rotator cuff tear patients 2 days before the surgery. Humerus, rotator cuff tear, and rotator cuff tendon were manually traced. 3D-MRI data was acquired using a medical image processing, analysis, and visualization (MIPAV). Four orthopedic surgeons, 2 residents (resident G) and 2 shoulder surgeons, more than 100 case shoulder surgery experiences (shoulder surgeon G) reviewed the MRI imaging for each patient independently. The five rotator cuff tear shapes used in this study: crescent, reverse L, L shaped, trapezoidal, and massive tear (Ellman and Gartsman Classification). The four rotator cuff tear sizes used in this study: small < 1cm, medium < 3cm, large < 5cm, and massive > 5cm (Cofield Classification). All images from the 2D and 3D-MRI were separated. In the first, the readers characterized the shape and size of the rotator cuff tendon tears by reviewing the 2D-MRI and blind to the intraoperative findings. Secondly, the 3D-MRI of each rotator cuff tear was reviewed, and shape and size of the tear were documented by each surgeon independently and blind to the intraoperative findings. These results were compared with the 2D-MRI imaging evaluation and intraoperative findings. (Figure 1)
Results:
The accuracy for tear shape using measurements on 2D-MRI was 64.3% for resident group and 78.6% for shoulder surgeon group. The accuracy for tear shape characterization using 3D-MRI was 85.7% for resident group and 92.9% for shoulder surgeon group. Moreover, the accuracy for tear size using measurements on 2D-MRI was 71.4% for resident group and 92.9% for shoulder surgeon group. The accuracy for tear size using 3D-MRI was 92.8% for resident group and 100% for shoulder surgeon group. The overall tear shape accuracy of 3D-MRI was 89.3%, significantly different from 2D-MRI accuracy (71.4%) (P=0.02). Furthermore, the overall tear size accuracy of 3D MRI was 96.4%, not significantly different from 2-D MRI accuracy (82.1%) (P=0.058). (Figure 2)
Conclusion:
In this study, 3D-MRI and 2D-MRI were significantly different in accuracy for diagnosing the shape of rotator cuff tears. Almost readers could easily review the shape and size of rotator cuff tear by 3D-MRI. This study has demonstrated that 3D-MRI of the rotator cuff tear improved the accuracy of characterizing the shape of a rotator cuff tear compare with the current 2D MRI-based techniques. The findings in this study suggest that 3D-MRI of the rotator cuff can work an important instrument for better categorizing and understanding rotator cuff tears, and those providing information that can be beneficial for pre-surgical planning. Further study and these data programing could automatically develop three-dimensional rotator cuff tendon magnetic resonance imaging system.
