Deep Learning Reconstruction Enables Prospectively Accelerated Clinical Knee MRI.

Background MRI is a powerful diagnostic tool with a long acquisition time. Recently, deep learning (DL) methods have provided accelerated high-quality image reconstructions from undersampled data, but it is unclear if DL image reconstruction can be reliably translated to everyday clinical practice. Purpose To determine the diagnostic equivalence of prospectively accelerated DL-reconstructed knee MRI compared with conventional accelerated MRI for evaluating internal derangement of the knee in a clinical setting. Materials and Methods A DL reconstruction model was trained with images from 298 clinical 3-T knee examinations. In a prospective analysis, patients clinically referred for knee MRI underwent a conventional accelerated knee MRI protocol at 3 T followed by an accelerated DL protocol between January 2020 and February 2021. The equivalence of the DL reconstruction of the images relative to the conventional images for the detection of an abnormality was assessed in terms of interchangeability. Each examination was reviewed by six musculoskeletal radiologists. Analyses pertaining to the detection of meniscal or ligament tears and bone marrow or cartilage abnormalities were based on four-point ordinal scores for the likelihood of an abnormality. Additionally, the protocols were compared with use of four-point ordinal scores for each aspect of image quality: overall image quality, presence of artifacts, sharpness, and signal-to-noise ratio. Results A total of 170 participants (mean age ± SD, 45 years ± 16; 76 men) were evaluated. The DL-reconstructed images were determined to be of diagnostic equivalence with the conventional images for detection of abnormalities. The overall image quality score, averaged over six readers, was significantly better (P < .001) for the DL than for the conventional images. Conclusion In a clinical setting, deep learning reconstruction enabled a nearly twofold reduction in scan time for a knee MRI and was diagnostically equivalent with the conventional protocol. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Roemer in this issue.

Using Deep Learning to Accelerate Knee MRI at 3 T: Results of an Interchangeability Study.

OBJECTIVE. Deep learning (DL) image reconstruction has the potential to disrupt the current state of MRI by significantly decreasing the time required for MRI examinations. Our goal was to use DL to accelerate MRI to allow a 5-minute comprehensive examination of the knee without compromising image quality or diagnostic accuracy. MATERIALS AND METHODS. A DL model for image reconstruction using a variational network was optimized. The model was trained using dedicated multisequence training, in which a single reconstruction model was trained with data from multiple sequences with different contrast and orientations. After training, data from 108 patients were retrospectively undersampled in a manner that would correspond with a net 3.49-fold acceleration of fully sampled data acquisition and a 1.88-fold acceleration compared with our standard twofold accelerated parallel acquisition. An interchangeability study was performed, in which the ability of six readers to detect internal derangement of the knee was compared for clinical and DL-accelerated images. RESULTS. We found a high degree of interchangeability between standard and DL-accelerated images. In particular, results showed that interchanging the sequences would produce discordant clinical opinions no more than 4% of the time for any feature evaluated. Moreover, the accelerated sequence was judged by all six readers to have better quality than the clinical sequence. CONCLUSION. An optimized DL model allowed acceleration of knee images that performed interchangeably with standard images for detection of internal derangement of the knee. Importantly, readers preferred the quality of accelerated images to that of standard clinical images.

Comparison of a Fast 5-Minute Shoulder MRI Protocol With a Standard Shoulder MRI Protocol: A Multiinstitutional Multireader Study.

OBJECTIVE: The purpose of this study was to compare the diagnostic performance of a 5-minute shoulder MRI protocol consisting of multiplanar 2D fast spin-echo (FSE) sequences with parallel imaging to that of a standard shoulder MRI protocol. MATERIALS AND METHODS: A retrospective review of 151 3-T MRI examinations of shoulders of 147 patients (mean age, 46.95 years) and 50 1.5-T MRI examinations of shoulders of 50 patients (mean age, 53.74 years) with four fast and five standard sequences from two academic centers between January 2014 and April 2015 was performed by three musculoskeletal radiologists. Interchangeability of fast and standard MRI was tested by comparing interprotocol (fast vs standard) interreader agreement with standard MRI interreader agreement. Interreader agreement was also compared using kappa statistics. The frequency of major findings was compared using an adjusted McNemar test. Sensitivity and specificity of MRI were measured for 51 patients who underwent surgery. RESULTS: Interprotocol reader agreement was essentially equal to reader agreement on standard MRI (mean difference ≤ 1%; 95% CI, -3.8% to 3.9%; 61-96% across structures). Interprotocol kappa values (0.373-0.645) were similar to standard MRI kappa values (0.320-0.726). Frequencies of major findings on fast and standard MRI were similar (0.7-19.6% across structures; p ≥ 0.08). Sensitivities of fast MRI for tendon and labral tears (33-92%) were equivalent or higher than those of standard MRI with similar specificities (77-98%). CONCLUSION: Fast 5-minute shoulder MRI with multiplanar 2D FSE sequences using parallel imaging is interchangeable, with similar interreader agreement and accuracy, with standard shoulder MRI for evaluating shoulder injuries.

Adventures and Misadventures in Plastic Surgery and Soft-Tissue Implants.

Soft-tissue augmentation and implants are increasingly seen by the radiologist as more techniques emerge for a variety of indications and locations. Some surgical and implant procedures are performed for purely cosmetic reasons in otherwise healthy patients seeking to improve their body image, and some are performed for reconstruction after cancer or other chronic illnesses. Abdominoplasty, liposuction, and autologous fat grafting can be performed for abdominal and gluteal contouring. Injection of liquid injectable silicone has historically been fraught with legal issues, although it continues to be used for augmentation in a variety of anatomic locations. Newer solid silicone implants have revolutionized cosmetic and reconstructive muscular contouring. Subdermal implants placed by nonmedical professionals are relatively new and unrecognized within the medical establishment, although such implants have been described in the popular culture. Perhaps the most rapidly increasing segment of cosmetic procedures, however, is minimally invasive cosmesis in the form of soft-tissue fillers in the hands and face. Finally, the major principles of breast augmentation and penile implants are also reviewed. Regardless of the location and the type of implant, complications of plastic surgery and soft-tissue implants can generally be classified into the following categories: seroma, hematoma, infection, migration, vascular or nerve compression, fibrosis, foreign-body reaction, and rupture or breakdown. Key concepts include knowing the appropriate anatomic location and the normal postoperative appearance so that complications can be properly detected. A broad range of approved, off-label, and illicit plastic surgical and implant procedures are described and their complications illustrated with cases with classic imaging findings. ©RSNA, 2017.

Imaging features of iBalance, a new high tibial osteotomy: what the radiologist needs to know.

OBJECTIVE: To describe the post-surgical imaging appearance and complications of high tibial osteotomy in patients with the iBalance implant system (iHTO; Arthrex, Naples, FL, USA). MATERIALS AND METHODS: Retrospective, institutional review board-approved, Health Insurance Portability and Accountability Act-compliant review of imaging after 24 iBalance procedures was performed with attention to: correction of varus malalignment, healing at the osteotomy site, resorption of the osteoinductive compound, and complications. RESULTS: Immediate correction of the varus deformity was present in all cases. Lobular radiolucency was present in all cases, more pronounced on the lateral knee radiograph, simulating infection or erosive disease. Four radiographic signs of healing were observed: blurring at the opposing osteotomy bony margins and at the osteoinductive compound and the adjacent bone interface, callus formation, and resorption of the osteoinductive compound. Complications were present in 33 % of cases, including fracture through the lateral tibial cortex (21 %), genu varum recurrence (8 %), painful exuberant bone formation (4 %), persistent pain, requiring total knee arthroplasty (4 %), and non-union (after >6 months' follow-up), with suspected infection (4 %). CONCLUSION: Radiologists should be aware of the normal radiographic appearance following iBalance high tibial osteotomy, which may be confused with infection. Radiologists should also be aware of potential post-operative complications and compare all post-operative radiographs with the immediate post-operative examination to detect collapse of the osteotomy site and recurrence of varus angulation.

Subluxation of the peroneus long tendon in the cuboid tunnel: is it normal or pathologic? An ultrasound and magnetic resonance imaging study.

OBJECTIVE: To evaluate the position of the peroneus longus (PL) tendon relative to the cuboid tuberosity and cuboid tunnel during ankle dorsiflexion and plantarflexion using ultrasound and MRI. MATERIALS AND METHODS: The study population included two groups: 20 feet of 10 asymptomatic volunteers who underwent prospective dynamic ultrasound and 55 ankles found through retrospective review of routine ankle MRI examinations. The location of the PL tendon at the cuboid tuberosity and cuboid tunnel was designated as completely within the tunnel, indeterminate, or subluxed with respect to ankle dorsiflexion and plantarflexion. RESULTS: On dynamic ultrasound, the PL tendon was perched plantar to the cuboid tuberosity in dorsiflexion, and glided to enter the cuboid tunnel distal to the tuberosity in plantarflexion in all 20 feet. On the MRI evaluation, there was a statistically significant difference (p = 0.0006) in the location of the PL tendon between the ankles scanned in dorsiflexion and plantarflexion. CONCLUSION: Based on our findings on ultrasound and MRI, the PL tendon can glide in and out of the cuboid tunnel along the cuboid tuberosity depending on ankle position. Thus, "subluxation" of the tendon as it curves to enter the cuboid tunnel, which to the best of our knowledge has not yet been described, should be recognized as a normal, position-dependent phenomenon and not be reported as pathology.

Tarsal tunnel disease and talocalcaneal coalition: MRI features.

OBJECTIVE: To assess, utilizing MRI, tarsal tunnel disease in patients with talocalcaneal coalitions. To the best of our knowledge, this has only anecdotally been described before. MATERIALS AND METHODS: Sixty-seven ankle MRIs with talocalcaneal coalition were retrospectively reviewed for disease of tendons and nerves of the tarsal tunnel. Interobserver variability in diagnosing tendon disease was performed in 30 of the 67 cases. Tarsal tunnel nerves were also evaluated in a control group of 20 consecutive ankle MRIs. RESULTS: Entrapment of the flexor hallucis longus tendon (FHL) by osseous excrescences was seen in 14 of 67 cases (21 %). Attenuation, split tearing, tenosynovitis, or tendinosis of the FHL was present in 26 cases (39 %). Attenuation or tenosynovitis was seen in the flexor digitorum longus tendon (FDL) in 18 cases (27 %). Tenosynovitis or split tearing of the posterior tibial tendon (PT) was present in nine cases (13 %). Interobserver variability ranged from 100 % to slight depending on the tendon and type of disease. Intense increased signal and caliber of the medial plantar nerve (MPN), indicative of neuritis, was seen in 6 of the 67 cases (9 %). Mildly increased T2 signal of the MPN was seen in 15 (22 %) and in 14 (70 %) of the control group. CONCLUSIONS: Talocalcaneal coalitions may be associated with tarsal tunnel soft tissue abnormalities affecting, in decreasing order, the FHL, FDL, and PT tendons, as well as the MPN. This information should be provided to the referring physician in order to guide treatment and improve post-surgical outcome.

Plantar tendons of the foot: MR imaging and US.

Tendon disorders along the plantar aspect of the foot may lead to significant symptoms but are often clinically misdiagnosed. Familiarity with the normal anatomy of the plantar tendons and its appearance at magnetic resonance (MR) imaging and ultrasonography (US) is essential for recognizing plantar tendon disorders. At MR imaging, the course of the plantar tendons is optimally visualized with dedicated imaging of the midfoot and forefoot. This imaging should include short-axis images obtained perpendicular to the long axis of the metatarsal shafts, which allows true cross-sectional evaluation of the plantar tendons. Normal plantar tendons appear as low-signal-intensity structures with all MR sequences. At US, accurate evaluation of the tendons requires that the ultrasound beam be perpendicular to the tendon. The normal tendon appears as a compact linear band of echogenic tissue that contains a fine, mixed hypoechoic and hyperechoic internal fibrillar pattern. Tendon injuries can be grouped into six major categories: tendinosis, peritendinosis, tenosynovitis, entrapment, rupture, and instability (subluxation or dislocation) and can be well assessed with both MR imaging and US. The radiologist plays an important role in the diagnosis of plantar tendon disorders, and recognizing their imaging appearances at MR imaging and US is essential.

US and MR imaging of the extensor compartment of the ankle.

Injuries to the extensor compartment of the ankle are uncommon and often are overlooked or misinterpreted at clinical presentation. Ultrasonography (US) and magnetic resonance (MR) imaging play a critical role in the diagnosis and evaluation of these injuries. US is a dynamic, quick, cost-effective imaging method for assessing ankle extensor compartment injuries as an alternative or adjunct to MR imaging. MR imaging provides multiplanar cross-sectional delineation of regional anatomic structures and also can be used to assess the predisposing cause, estimate the extent of injury, and aid in preoperative planning for requisite surgical repair. The spectrum of pathologic conditions affecting the ankle extensor compartment ranges from tendinosis secondary to degenerative, inflammatory, or depositional disease to traumatic tendon or retinacular rupture and entrapment neuropathy. Major components of the ankle extensor compartment at risk for injury include the anterior tibial, extensor hallucis longus, and extensor digitorum longus tendons; the extensor retinacular mechanism; and the anterior tarsal tunnel. Familiarity with the normal anatomic appearance and pathologic features of the ankle extensor compartment at US and MR imaging as well as potential imaging pitfalls is critical for accurate injury evaluation.

Automated Radiology-Operative Note Communication Tool; Closing the Loop in Musculoskeletal Imaging.

RATIONALE AND OBJECTIVES: Correlation of imaging studies and reference standard outcomes is a significant challenge in radiology. This study evaluates the effectiveness of a new communication tool by assessing the ability of this system to correctly match the imaging studies to arthroscopy reports and qualitatively assessing radiologist behavior before and after the implementation of this system. MATERIALS AND METHODS: Using a commercially available communication or educational tool and applying a novel matching rule algorithm, radiology and arthroscopy reports were matched from January 17, 2017 to March 1, 2017 based on anatomy. The interpreting radiologist was presented with email notifications containing the impression of the imaging report and the entire arthroscopy report. Total correlation rate of appropriate report pairings, modality-specific correlation rate, and the anatomy-specific correlation rate were calculated. Radiologists using the system were given a survey. RESULTS: Overall correlation rate for all musculoskeletal imaging was 83.1% (433 or 508). Low correlation was found in fluoroscopic procedures at 74.4%, and the highest correlation was found with ultrasound at 88.4%. Anatomic location varied from 51.6% for spine to 98.8% for hips and pelvis studies. Survey results revealed 87.5% of the respondents reporting being either satisfied or very satisfied with the new communication tool. The survey also revealed that some radiologists reviewed more cases than before. CONCLUSIONS: Matching of radiology and arthroscopy reports by anatomy allows for excellent report correlation (83.1%). Automated correlation improves the quality and efficiency of feedback to radiologists, providing important opportunities for learning and improved accuracy.

MRI Screening in Operative Scheuermann Kyphosis: Is it Necessary?

STUDY DESIGN: Patients with preoperative spine magnetic resonance imaging (MRI) studies from a prospective multicenter study of operative adolescent Scheuermann kyphosis (SK). OBJECTIVES: To investigate the usefulness of MRI screening in operative planning for SK surgeries. SUMMARY OF BACKGROUND DATA: Neural axis abnormalities in operative SK have not been previously studied with MRI screening, despite its use. METHODS: One orthopedic surgeon and two radiologists evaluated all images retrospectively. Radiographs were evaluated for kyphosis apex and magnitude. MRIs were evaluated for spinal cord abnormalities, epidural lipomatosis, location and number of vertebral wedging, Schmorl nodes and posterior disc herniations, frequency of spondylolysis, etc. The relationship of these pathologies to the kyphosis apex was explored. This group was compared to a surgical SK group without preoperative MRIs. RESULTS: Eighty-six patients with MRIs, mean age 16.3 years, 64% male, and a mean preoperative kyphosis of 75.9° were evaluated. There were 17 spinal cord abnormalities. Low-lying conus was found in 2 patients, and syrinx in 15 (no Chiari malformations). Epidural lipomatosis was found in 49 patients, average of 5.7 levels. Anterior vertebral wedging occurred in all (mean 4.7 levels). Posterior disc herniations averaged 5.2 levels/patient and 1.8 levels caudad to the apex. Spondylolysis was reported in 8.1%. Four cases (4.7%) had the operative plan changed as a result of the preoperative MRI: two due to neural compression, one due to disc herniation and one due to a spinal cord draped over the apex. Thirty-one patients did not receive an MRI; there were no significant differences between the two groups. The rate of postoperative neurologic change was 3.5% in the MRI group and 3.2% in the no-MRI group. CONCLUSIONS: Based on 4.7% of cases requiring a change in the operative plan as a result of preoperative MRI, the authors recommend considering performing screening MRI in operative SK patients.

Postchemoembolization liver abscess: findings on hepatobiliary scintigraphy.

We describe a 41-year-old man with a 1-week history of nausea and vomiting 1 month after chemoembolization of a liver metastasis. The patient subsequently became febrile and developed right upper quadrant abdominal and midback pain. Findings of initial laboratory and imaging studies (a noncontrast computed tomographic [CT] scan and ultrasound) were not remarkable. Hepatobiliary scintigraphy, performed to rule out cholecystitis, revealed an abnormal area in the right lobe of the liver consistent with a focal bile leak into an abscess cavity. The patient was subsequently treated for liver abscess. In conclusion, hepatobiliary scintigraphy should be considered as a first-line test in the work-up of patients whenever a postchemoembolization complication is considered likely.

Correlation between Rotator Cuff Tears and Systemic Atherosclerotic Disease.

The purpose of this study was to investigate the association of aortic arch calcification, a surrogate marker of atherosclerosis, with rotator cuff tendinosis and tears given the hypothesis that decreased tendon vascularity is a contributing factor in the etiology of tendon degeneration. A retrospective review was performed to identify patients ages 50 to 90 years who had a shoulder MRI and a chest radiograph performed within 6 months of each other. Chest radiographs and shoulder MRIs from 120 patients were reviewed by two sets of observers blinded to the others' conclusions. Rotator cuff disease was classified as tendinosis, partial thickness tear, and full thickness tear. The presence or absence of aortic arch calcification was graded and compared with the MRI appearance of the rotator cuff. The tendon tear grading was positively correlated with patient age. However, the tendon tear grading on MRI was not significantly correlated with the aorta calcification scores on chest radiographs. Furthermore, there was no significant correlation between aorta calcification severity and tendon tear grading. In conclusion, rotator cuff tears did not significantly correlate with aortic calcification severity. This suggests that tendon ischemia may not be associated with the degree of macrovascular disease.

Magnetic resonance imaging of adhesive capsulitis: correlation with clinical staging.

The purpose of this study was to evaluate non-contrast magnetic resonance imaging (MRI) findings of adhesive capsulitis and correlate them with clinical stages of adhesive capsulitis. This will hopefully define a role for shoulder MR imaging in the diagnosis of adhesive capsulitis as well as in potentially directing appropriate treatment. Forty-seven consecutive non-contrast magnetic resonance imaging examinations of 46 patients with a clinical diagnosis of adhesive capsulitis were retrospectively reviewed and correlated with clinical staging. Specific MRI criteria correlated with the clinical stage of adhesive capsulitis, including the thickness and signal intensity of the joint capsule and synovium as well as the presence and severity of scarring in the rotator interval. Routine MRI of the shoulder without intraarticular administration of gadolinium can be used to diagnose all stages of adhesive capsulitis, including stage 1, where findings may be subtle on clinical examination. We believe that future studies assessing the role of MRI in guiding the initiation of appropriate treatment should be undertaken.

Ultrasound-guided interdigital neuroma injections: short-term clinical outcomes after a single percutaneous injection--preliminary results.

PURPOSE: To describe the procedure of ultrasound-guided Morton's neuroma and recurrent stump neuroma injections and early clinical outcomes after a single injection. MATERIALS AND METHODS: Retrospective review of 44 percutaneous ultrasound-guided neuroma injections in 24 patients who had completed clinical outcomes questionnaires. A 10-point pain scale [scale of 1 (no pain) to 10 (severe pain)] in a 7-day pain log format was distributed to patients at the time percutaneous neuroma injection was performed. RESULTS: Neuromas were clearly visualized with sonography as hypoechoic nodules and were distinguishable from other causes of forefoot pain, such as metatarsophalangeal joint synovitis and intermetatarsal bursae. The sizes of the neuromas injected ranged between 4 and 19 mm. Postinjection, all neuromas displayed increased echogenicity and/or the appearance of fluid surrounding it, confirming localization of the therapeutic mixture. We arbitrarily subdivided the pain ratings into symptomatic (greater than 4) and asymptomatic (less than or equal to 4) for statistical analysis. Average pain level pre injection was 5.2 and average pain level was 3.7 at 7 days post single injection, with 62% of the initially symptomatic patients asymptomatic on day 7 (p < 0.000001). Overall, 76% of the total number of neuromas injected once were asymptomatic on day 7. CONCLUSION: Ultrasound can be used to accurately target Morton's neuromas and, therefore, appropriately direct therapeutic interventions, with good short-term clinical results.