Near-metal MRI Using PETRA With Extended Phase Encoding and Compressed Sensing

OBJECTIVES: To apply compressed sensing (CS) PETRA sequence to MRI near metallic implants, and to investigate the use of an extended phase encoding in csPETRA without significantly increasing the total acquisition time. MATERIALS AND METHODS: Using a 3D realistic model of an orthopedic implant, csPETRA with regular and extended phase encoding was simulated. csPETRA protocols with phase-encoded samples covering from 3.2% to 6.5% of k-space were prepared with a total acquisition time of up to 7±1 minutes and compared to a TSE sequence with VAT and SEMAC in grid phantoms with 3 different total knee arthroplasty implants. In vivo measurements on a patient with a wrist implant and a patient with a knee prosthesis were performed. Quantitative analysis based on the artifact size and qualitative comparison of the tested sequences are presented. RESULTS: In the csPETRA images, geometric distortions were reduced by 40% compared to those in the SEMAC-VAT images. A ratio of 6.5% for the central single-point imaging (SPI) coverage yielded a 20% improvement in artifact size compared to 3.2% coverage, and a 16-fold acceleration reduced the total scan time to 8 minutes. CONCLUSIONS: csPETRA with a higher SPI/radial ratio achieves a 20% reduction in artifact size relative to the minimum SPI/radial ratio, that is, conventional PETRA, without a time penalty. In addition, it reduces SAR by a factor of 2.5 compared to SE-based techniques. csPETRA should currently be regarded as a complementary research tool rather than a standalone clinical sequence, with potential niche applications such as high-resolution (0.5 mm isotropic) field mapping and implant assessment, or imaging patients with strict SAR or acoustic-noise constraints, while its clinical value requires further validation together with the development and evaluation of suitable soft-tissue contrast mechanisms.