School of Medicine

Purpose : To optimize acquisition parameters for three-dimensional fast spin-echo (FSE) imaging of the knee using subjective and objective metrics of image quality. Materials and methods : This prospective, HIPAA-compliant, institutional review board-approved study was performed with informed consent of the subjects. The knees of 8 healthy male and female volunteers were imaged in a 3 T MRI scanner using an 8-channel knee coil. A total of 146 intermediate-weighted isotropic resolution 3D FSE (3D-FSE-Cube) images with varied acquisition parameter settings were acquired with an additional 3D-FSE-Cube scan acquired as a reference for subjective image quality assessment. Images were randomized and graded for overall quality, parallel imaging artifact severity and blurring with respect to the reference. Cartilage, muscle and fluid signal-to-noise ratios (SNR) and fluid-cartilage contrast-to-noise ratios (CNR) were quantified by acquiring scans with and without RF excitation and custom-reconstructing the k-space data to produce signal and noise-only images. Mixed effects regression modeling was used to determine statistically significant effects of the different parameters on image quality. Results : All quality metrics were significantly affected by varying receiver bandwidth, repetition time and echo train length (p < .05). Lowering bandwidth improved image quality for all metrics except blurring. Reader agreement was slight to fair for subjective metrics, but overall trends in quality ratings were apparent. Conclusion : We used a systematic approach to optimize parameters for 3D-FSE-Cube for knee imaging. Image quality was found to be overall improved using a receiver bandwidth of ±31.25 kHz, and blurring increased with lower bandwidth and longer echo trains.