Jan Fritz

Musculoskeletal magnetic resonance imaging has evolved substantially, driven by advances in hardware, image acquisition, and reconstruction techniques. Improvements in gradient performance and dedicated radiofrequency coils have enhanced spatial resolution and scan efficiency across field strengths. Image acceleration strategies, including parallel imaging, simultaneous multislice acquisition, and compressed sensing, now enable high-quality two-dimensional and three-dimensional magnetic resonance imaging with markedly reduced examination times and facilitate the time-neutral incorporation of advanced metal artifact reduction techniques into clinical magnetic resonance imaging protocols. Deep learning–based reconstruction and super-resolution augmentation methods have further expanded achievable acceleration and image quality. Emerging techniques such as synthetic magnetic resonance imaging, magnetic resonance neurography, kinematic magnetic resonance imaging, and zero echo time magnetic resonance imaging expand the capabilities of musculoskeletal magnetic resonance imaging. At the same time, renewed interest in low-field magnetic resonance imaging provides intriguing opportunities to improve accessibility and sustainability. Ultra-high field magnetic resonance imaging provides unprecedented spatial resolution and quantitative insights in selected applications. These developments are redefining musculoskeletal magnetic resonance imaging practice and broadening its clinical value.

Knee MRI plays a central role in musculoskeletal diagnostics but has traditionally been associated with relatively long acquisition times. Recent technological advances have fundamentally changed this paradigm. Parallel imaging (PI), simultaneous multi-slice acquisition (SMS), compressed sensing (CS), and combinations thereof have substantially reduced scan times without compromising diagnostic image quality. The introduction of deep learning (DL)-based reconstruction further elevates this transformative breakthrough, as it can reconstruct high-quality diagnostic MR images at higher acceleration factors, where conventional image reconstruction methods have traditionally struggled to succeed. Sixfold PIxSMS-accelerated DL protocols have demonstrated excellent diagnostic performance and image quality, allowing comprehensive knee MRI examinations to be completed in under five minutes. Accelerated three-dimensional (3D) TSE techniques, such as CAIPIRINHA-accelerated SPACE sequences, further expand the potential of knee MRI by enabling high-resolution isotropic 3D imaging at acquisition times that are increasingly practical for routine clinical use. Ongoing improvements in DL-based reconstruction and denoising may soon bridge the remaining gap, promising to enable the acquisition of isotropic 3D datasets with multiple contrasts within minutes. Beyond technical acceleration, the successful implementation of fast MRI requires careful workflow optimization and consideration of architectural and economic factors. This review outlines the technical principles underlying modern acceleration strategies, summarizes evidence from validation studies, discusses practical aspects of clinical implementation and protocol optimization, and highlights future opportunities and challenges.

Knee injuries are one of the most common complaints in sports medicine. Magnetic resonance imaging is an essential adjunct to clinical evaluation for many traumatic injuries and overuse conditions. Given the heavy use of knee magnetic resonance imaging, developing faster magnetic resonance imaging acquisition methods and deployment in clinical practice would be valuable. In this article, we illustrate a spectrum of knee abnormalities from our clinical practice, utilizing a recently developed, publicly available sub-5-minute knee magnetic resonance imaging protocol with super-resolution image reconstruction based on deep learning. We review common traumatic injuries and overuse conditions of the knee and illustrate cases with this novel fast knee magnetic resonance imaging protocol.

Modern image acceleration techniques and deep learning–based image reconstruction facilitate substantially faster and higher-quality musculoskeletal MRI examinations than do conventional techniques.

Objectives: Metal artifact reduction MRI can exceed specific absorption rate (SAR) limits due to high-bandwidth radiofrequency pulses, causing scan interruptions and prolonged acquisition times. The aim of the current study is to reduce SAR and potentially scan time in metal artifact reduction MRI using an optimized variable refocusing flip angle (VRFA) scheme compared with the standard constant refocusing flip angle (CRFA). Materials and Methods: Three VRFA variants (VRFA1 to VRFA3) were developed to maximize tissue signal and contrast while minimizing SAR and image blur. The optimal variant was selected through quantification of metal artifacts and image blur in phantoms and tissue signal in a volunteer. Patients with hip arthroplasty underwent CRFA and optimal VRFA imaging using high-bandwidth turbo-spin-echo (HBW-TSE) and compressed-sensing slice-encoding-for-metal-artifact-correction (CS-SEMAC) sequences. Three readers ranked paired CRFA and VRFA scans for quality. Analyses included repeated measures ANOVA, noninferiority testing, and paired t /Wilcoxon signed-rank tests. Results: CRFA and VRFA1 to VRFA3 showed no significant differences in image blur (full-width-at-half-maximum, mean ± SD, 1.9 ± 0.2 vs 1.9 ± 0.2 vs 1.9 ± 0.3 vs 1.9 ± 0.3 pixels, P = 0.06) or metal artifacts (8.2 ± 2.8 vs 8.4 ± 2.7 vs 8.4 ± 2.6 vs 8.4 ± 2.7 pixels, P = 0.57). The optimal VRFA variant (VRFA3) preserved 81% of CRFA fat-muscle contrast at 77% SAR and 70% scan time on proton-density (PD), and 94% of fluid-muscle contrast at 80% SAR and 67% scan time on short-tau-inversion-recovery (STIR). In 23 patients [mean age, 67.3 y ± 12.2 (SD); 14 females], the optimal VRFA was noninferior to CRFA in all quality metrics (all 95% CI < noninferiority margin = 0.1) and significantly reduced SAR (mean, PD-HBW-TSE/STIR-HBW-TSE/PD-CS-SEMAC/STIR-CS-SEMAC: 1.11/1.35/1.17/1.18 vs 1.85/1.83/1.49/1.46 W/kg, all P ≤ 0.001). In HBW-TSE, reduced SAR allowed longer echo trains and 15% to 32% shorter scan times. Conclusion: Metal artifact reduction MRI with VRFA reduces SAR without compromising image quality. It allows shorter acquisitions in HBW-TSE.

Objective This study aims to determine the diagnostic performance of sub-5-min combined sixfold parallel imaging (PIx3)–simultaneous multislice (SMSx2)–accelerated deep learning (DL) super-resolution 3T knee MRI in children and adolescents. Materials and methods Children with painful knee conditions who underwent PIx3–SMSx2–accelerated DL super-resolution 3T knee MRI and arthroscopy between October 2022 and December 2023 were retrospectively included. Nine fellowship-trained musculoskeletal radiologists independently scored the MRI studies for image quality and the presence of artifacts (Likert scales, range: 1 = very bad/severe, 5 = very good/absent), as well as structural abnormalities. Interreader agreements and diagnostic performance testing was performed. Results Forty-four children (mean age: 15 ± 2 years; range: 9–17 years; 24 boys) who underwent knee MRI and arthroscopic surgery within 22 days (range, 2–133) were evaluated. Overall image quality was very good (median rating: 5 [IQR: 4–5]). Motion artifacts (5 [5–5]) and image noise (5 [4–5]) were absent. Arthroscopy-verified abnormalities were detected with good or better interreader agreement (κ ≥ 0.74). Sensitivity, specificity, accuracy, and AUC values were 100%, 84%, 93%, and 0.92, respectively, for anterior cruciate ligament tears; 71%, 97%, 93%, and 0.84 for medial meniscus tears; 65%, 100%, 86%, and 0.82 for lateral meniscus tears; 100%, 100%, 100%, and 1.00 for discoid lateral menisci; 100%, 95%, 96%, and 0.98 for medial patellofemoral ligament tears; and 55%, 100%, 98%, and 0.77 for articular cartilage defects. Conclusion Clinical sub-5-min PIx3–SMSx2–accelerated DL super-resolution 3T knee MRI provides excellent image quality and high diagnostic performance for diagnosing internal derangement in children and adolescents.

Seven-minute threefold parallel imaging–accelerated deep learning super-resolution 3-T shoulder MRI showed robust diagnostic performance, yielding images of high quality and depicting shoulder abnormalities with good to excellent accuracy.

Deep learning superresolution simultaneous multislice parallel imaging–accelerated knee MRI provided excellent image quality and improved diagnostic performance for articular cartilage defects while maintaining performance for meniscus and ligament tears.