3D-FSE-cube imaging without fat suppression of the knee joint : optimization of scan parameters and diagnostic accuracy comparison with fat suppressed imaging at 1.5T MRI

Abstract

Background and Purpose: Three-dimensional fast spin-echo imaging sequence with variable flip angle (3D-FSE-Cube) images have been developed recently and established as an essential sequence in the routine knee magnetic resonance imaging (MRI) protocol. However, no study has compared 3D-FSE-Cube imaging of the knee joints “with” and “without” fat suppression, and the diagnostic performance of 3D-FSE-Cube imaging without fat suppression (3D-FSE-Cube-NFS) has not yet been investigated. The aims of this study were as follows: (1) to optimize scan parameters for 3D-FSE-Cube imaging with fat suppression (3D-FSE-Cube-FS) and 3D-FSE-Cube-NFS in 1.5T knee MRI and (2) to compare 3D-FSE-Cube-FS and 3D-FSE-Cube-NFS to evaluate the diagnostic performance of 3D-FSE-Cube-NFS for detecting lesions of the meniscus, ligaments, bone marrow, or cartilage in knee imaging. Materials and Methods: A porcine knee phantom was used to optimize scan parameters for 3D-FSE-Cube imaging both with and without fat suppression in a 1.5T MRI system (Signa Horizon; GE Healthcare, Waukesha, WI, USA). Sagittal images with varied settings of repetition time (TR) from 1000 to 1300 ms and echo train length (ETL) from 30 to 60 were acquired. The image acquired with TR=1300 ms and ETL=30 served as a reference scan. Two musculoskeletal radiologists graded all images on a scale from -8 to 8 on the basis of image blurring and overall image quality relative to the reference scan. Subsequently, the same survey was performed on a healthy human volunteer by using parameter settings that received scores of -2 or above in the phantom study. Images with a score of -1 or above were regarded as acceptable. Signal-to-noise ratio (SNR) and SNR per unit time were measured in the patellar cartilage and femoral bone marrow on each image. After consideration of both subjective image evaluation and SNR per unit time, optimized scan parameters were determined. Using these optimized parameters, knee MRI scans were performed on 124 patients with knee pain by using both 3D-FSE-Cube-FS and 3D-FSE-Cube-NFS between September 2015 and December 2015. Among these patients, 25 subsequently underwent arthroscopic surgeries. Using the arthroscopic results and 2D images as reference standards, the two radiologists compared the diagnostic performance of 3D-FSE-Cube-NFS and 3D-FSE-Cube-FS images to determine which sequence was more diagnostically useful in a particular clinical situation, such as ligament tear, meniscus tear, subchondral bone marrow edema (BME) lesions, or cartilage defect. McNemar’s test was performed to compare the diagnostic performance of the two sequences at a significance level of p < 0.05. Results: Image quality and SNR increased with longer TR and shorter ETL. Among phantom images with a score of -1 or above, the highest SNR per unit time was acquired with scan parameters of TR=1300 ms and ETL=45 in both images with fat suppression (FS) and without fat suppression (NFS). In subsequent volunteer imaging, the same parameters were found to be the best with and without FS. In the imaging study of the 124 patients, there were no significant differences between the two sequences for the detection of meniscus tears and cartilage defect except for medial collateral ligament (MCL) tears and subchondral BME lesions. Compared to 3D-FSE-Cube-FS, 3D-FSE-Cube-NFS had lower sensitivity for the detection of MCL tears, and lower sensitivity and specificity for the detection of BME lesions. Nevertheless, 3D-FSE-Cube-NFS images showed advantages, such as reduced susceptibility artifact, ability for replacing inappropriate 3D-FSE-Cube-FS images, and detection of subtle BME lesions. Conclusion: Considering both acceptable image quality and short scan time, the optimized scan parameters for both 3D-FSE-Cube FS and 3D-FSE-Cube NFS were found to be TR=1300 ms and ETL=45. 3D-FSE-Cube-FS and 3D-FSE-Cube-NFS also showed similar sensitivity and specificity for the detection of meniscus tears or carti...