Automating Angle Measurements on Foot Radiographs in Young Children: Feasibility and Performance of a Convolutional Neural Network Model.

Measurement of angles on foot radiographs is an important step in the evaluation of malalignment. The objective is to develop a CNN model to measure angles on radiographs, using radiologists' measurements as the reference standard. This IRB-approved retrospective study included 450 radiographs from 216 patients (< 3 years of age). Angles were automatically measured by means of image segmentation followed by angle calculation, according to Simon's approach for measuring pediatric foot angles. A multiclass U-Net model with a ResNet-34 backbone was used for segmentation. Two pediatric radiologists independently measured anteroposterior and lateral talocalcaneal and talo-1st metatarsal angles using the test dataset and recorded the time used for each study. Intraclass correlation coefficients (ICC) were used to compare angle and paired Wilcoxon signed-rank test to compare time between radiologists and the CNN model. There was high spatial overlap between manual and CNN-based automatic segmentations with dice coefficients ranging between 0.81 (lateral 1st metatarsal) and 0.94 (lateral calcaneus). Agreement was higher for angles on the lateral view when compared to the AP view, between radiologists (ICC: 0.93-0.95, 0.85-0.92, respectively) and between radiologists' mean and CNN calculated (ICC: 0.71-0.73, 0.41-0.52, respectively). Automated angle calculation was significantly faster when compared to radiologists' manual measurements (3 ± 2 vs 114 ± 24 s, respectively; P < 0.001). A CNN model can selectively segment immature ossification centers and automatically calculate angles with a high spatial overlap and moderate to substantial agreement when compared to manual methods, and 39 times faster.

A Patch-Based Deep Learning Approach for Detecting Rib Fractures on Frontal Radiographs in Young Children.

Chest radiography is the modality of choice for the identification of rib fractures in young children and there is value for the development of computer-aided rib fracture detection in this age group. However, the automated identification of rib fractures on chest radiographs can be challenging due to the need for high spatial resolution in deep learning frameworks. A patch-based deep learning algorithm was developed to automatically detect rib fractures on frontal chest radiographs in children under 2 years old. A total of 845 chest radiographs of children 0-2 years old (median: 4 months old) were manually segmented for rib fractures by radiologists and served as the ground-truth labels. Image analysis utilized a patch-based sliding-window technique, to meet the high-resolution requirements for fracture detection. Standard transfer learning techniques used ResNet-50 and ResNet-18 architectures. Area-under-curve for precision-recall (AUC-PR) and receiver-operating-characteristic (AUC-ROC), along with patch and whole-image classification metrics, were reported. On the test patches, the ResNet-50 model showed AUC-PR and AUC-ROC of 0.25 and 0.77, respectively, and the ResNet-18 showed an AUC-PR of 0.32 and AUC-ROC of 0.76. On the whole-radiograph level, the ResNet-50 had an AUC-ROC of 0.74 with 88% sensitivity and 43% specificity in identifying rib fractures, and the ResNet-18 had an AUC-ROC of 0.75 with 75% sensitivity and 60% specificity in identifying rib fractures. This work demonstrates the utility of patch-based analysis for detection of rib fractures in children under 2 years old. Future work with large cohorts of multi-institutional data will improve the generalizability of these findings to patients with suspicion of child abuse.

Artificial Intelligence Algorithm Improves Radiologist Performance in Skeletal Age Assessment: A Prospective Multicenter Randomized Controlled Trial.

Background Previous studies suggest that use of artificial intelligence (AI) algorithms as diagnostic aids may improve the quality of skeletal age assessment, though these studies lack evidence from clinical practice. Purpose To compare the accuracy and interpretation time of skeletal age assessment on hand radiograph examinations with and without the use of an AI algorithm as a diagnostic aid. Materials and Methods In this prospective randomized controlled trial, the accuracy of skeletal age assessment on hand radiograph examinations was performed with (n = 792) and without (n = 739) the AI algorithm as a diagnostic aid. For examinations with the AI algorithm, the radiologist was shown the AI interpretation as part of their routine clinical work and was permitted to accept or modify it. Hand radiographs were interpreted by 93 radiologists from six centers. The primary efficacy outcome was the mean absolute difference between the skeletal age dictated into the radiologists' signed report and the average interpretation of a panel of four radiologists not using a diagnostic aid. The secondary outcome was the interpretation time. A linear mixed-effects regression model with random center- and radiologist-level effects was used to compare the two experimental groups. Results Overall mean absolute difference was lower when radiologists used the AI algorithm compared with when they did not (5.36 months vs 5.95 months; P = .04). The proportions at which the absolute difference exceeded 12 months (9.3% vs 13.0%, P = .02) and 24 months (0.5% vs 1.8%, P = .02) were lower with the AI algorithm than without it. Median radiologist interpretation time was lower with the AI algorithm than without it (102 seconds vs 142 seconds, P = .001). Conclusion Use of an artificial intelligence algorithm improved skeletal age assessment accuracy and reduced interpretation times for radiologists, although differences were observed between centers. Clinical trial registration no. NCT03530098 © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Rubin in this issue.

RADIAL: leveraging a learning management system to support radiology education.

Increasing volume and complexity of cases in academic radiology and the drive toward pediatric sub-specialization have threatened knowledge assimilation for radiologists. There is a clear need for a system that retrieves vetted information from the excess available on the internet. Partnered with an interdisciplinary team from the Graduate School of Education, the authors created the first comprehensive learning management system (LMS) for radiology, implemented in the reading room to augment image interpretation and point-of-care education. The LMS supports quantitative analysis using a robust analytics platform to evaluate user statistics, facilitating improved quality of patient care by revolutionizing the way radiologists assimilate knowledge. This integration promises to enhance workflow and point-of-care teaching and to support the highest quality of care.

Specific Absorption Rate and Specific Energy Dose: Comparison of 1.5-T versus 3.0-T Fetal MRI.

Background MRI performed at 3.0 T offers greater signal-to-noise ratio and better spatial resolution than does MRI performed at 1.5 T; however, for fetal MRI, there are concerns about the potential for greater radiofrequency energy administered to the fetus at 3.0-T MRI. Purpose To compare the specific absorption rate (SAR) and specific energy dose (SED) of fetal MRI at 1.5 and 3.0 T. Materials and Methods In this retrospective study, all fetal MRI examinations performed with 1.5- and 3.0-T scanners at one institution between July 2012 and October 2016 were evaluated. Two-dimensional (2D) and three-dimensional (3D) steady-state free precession (SSFP), single-shot fast spin-echo, 2D and 3D T1-weighted spoiled gradient-echo (SPGR), and echo-planar imaging sequences were performed. SAR, SED, accumulated SED, and acquisition time were retrieved from the Digital Imaging and Communications in Medicine header. Data are presented as mean ± standard deviation. Two one-sided tests with equivalence bounds of 0.5 (Cohen d effect size) were performed, with statistical equivalence considered at P < .05. Results A total of 2952 pregnant women were evaluated. Mean maternal age was 30 years ± 6 (age range, 12-49 years), mean gestational age was 24 weeks ± 6 (range, 17-40 weeks). A total of 3247 fetal MRI scans were included, with 2784 (86%) obtained at 1.5 T and 463 (14%) obtained at 3.0 T. In total, 93 764 sequences were performed, with 81 535 (87%) performed at 1.5 T and 12 229 (13%) performed at 3.0 T. When comparing 1.5- with 3.0-T MRI sequences, mean SAR (1.09 W/kg ± 0.69 vs 1.14 W/kg ± 0.61), mean SED (33 J/kg ± 27 vs 38 J/kg ± 26), and mean accumulated SED (965 J/kg ± 408 vs 996 J/kg ± 366, P < .001) were equivalent. Conclusion Fetal 1.5- and 3.0-T MRI examinations were found to have equivalent energy metrics in most cases. The 3.0-T sequences, such as two-dimensional T1-weighted spoiled gradient-echo and three-dimensional steady-state free precession, may require modification to keep the energy delivered to the patient as low as possible. © RSNA, 2020 Online supplemental material is available for this article.

Three-dimensional classification of the Lenke 1 adolescent idiopathic scoliosis using coronal and lateral spinal radiographs.

BACKGROUND: Classification of the spinal deformity in adolescent idiopathic scoliosis (AIS) remains two-dimensional (2D) as the spinal radiographs remain the mainstay in clinical evaluation of the disease. 3D classification systems are proposed, however are time consuming. We here aim to evaluate the clinical application of a 3D classification system by the use of only posterior-anterior and lateral radiographs in Lenke 1 adolescent idiopathic scoliosis (AIS). METHODS: Forty Lenke 1 AIS were classified by five observers following a three-step flowchart, developed based on our previous 3D classification system. This 3D classification characterizes the curve in the frontal and sagittal views and infers the third dimension with rules based on prior data to determine the 3D subtypes of the curve. Repeated rating was performed for 20 randomly selected patients in the same cohort. In addition to the classification by the raters, the 3D model of the spines were generated to determine the actual curve subtype based on the algorithm that was originally used to develop the 3D classification system. The interobserver and intraobserver reliability and the classification accuracy were determined for both 3D and axial classifications of the cohort. RESULTS: The interobserver reliability was moderate to strong with a kappa value between 0.61-0.89 for 3D and axial classifications. Comparing the mathematical classification and the raters' classification, the classification accuracy among all raters ranged between 56 and 89%. CONCLUSION: We evaluated the reliability of a previously developed 3D classification system for Lenke 1 AIS patients when only two-view spinal radiographs are available. Radiologists and orthopedic surgeons were able to identify the 3D subtypes of Lenke 1 AIS from the patients' radiographs with moderate to strong reliability. The new 3D classification has the potential to identify the subtypes of the Lenke 1 AIS without a need for quantitative 3D image post-processing.

Osteochondral lesion of the talus in children: Are there MRI findings of instability?

PURPOSE: The purpose of our study was to investigate the performance of MRI findings to predict instability of osteochondral lesion of the talus (OLT) in children and the association between skeletal maturity and lesion stability. MATERIALS AND METHOD: This retrospective IRB-approved and HIPPA-compliant study included children with OLT, who underwent an ankle MRI examination between March 1, 2011, and May 31, 2018. Blinded to the clinical outcome, 2 radiologists retrospectively assessed each MRI study for the presence or absence of various features on the articular side, along the interface, and on the subchondral side of each lesion. Regional skeletal maturity was recorded. Lesion stability was classified using clinical and surgical findings. Mann-Whitney U, Chi-square, Fisher's exact, and Cochran-Armitage tests were used to compare demographic and MRI findings between children with stable and unstable lesions. RESULTS: Of the 48 ankles identified, 36 were stable (12.7 + 3.9 years) and 12 were unstable (14.2 + 1.6 years) lesions. None of the lesions presented as a detached fragment. Skeletal immaturity (p = 0.01) was significantly more common in stable than unstable lesions. No other MRI features were found to be significantly different between stable and unstable lesions, which included the presence of an effusion (p = 0.27), intra-articular body (p = 0.25), cartilage changes (p = 0.19), subchondral disruption (p = 0.51), T2-weighted signal intensity rim (p = 0.16), cysts (p = 0.48), marginal sclerosis (p = 0.70), and perilesional marrow edema (p = 0.17). CONCLUSION: Results from our study suggest that previously published OCD criteria using conventional MRI are not sufficient for predicting stability of OLT in children. Regional skeletal maturity and older age were more predictive of unstable lesions.

Imaging Properties of Additive Manufactured (3D Printed) Materials for Potential Use for Phantom Models.

Over the last few decades, there has been growing interest in the application of additive manufacturing (AM) or 3D printing for medical research and clinical application. Imaging phantoms offer clear benefits in the way of training, planning, and quality assurance, but the model's availability per catalog tend to be suited for general testing purposes only. AM, on the contrary, offers flexibility to clinicians by enabling custom-built phantoms based on specific interests or even individual patient needs. This study aims to quantify the radiographic properties (ultrasound, magnetic resonance imaging, and computed tomography) of common additive manufacturing technologies and to discuss potential opportunities to fabricate imaging phantoms. Test phantoms were composed of samples from the three most common AM styles, namely PolyJet, fused deposition modeling (FDM), and stereolithography (SLA). Test imaging of the phantoms was performed on ultrasound, MRI, and CT and reviewed and evaluated with radiology software. The ultrasound images showed clearly defined upper and lower edges of the material but did not demonstrate distinct differences in internal echogenicity between materials. The MR scans revealed a distinct signal intensity difference between the model (17 grayscale value) and the printer support (778 grayscale value). Finally, the CT images showed a slight variation between the plastic (82 HU) and rubber (145 HU) materials. The radiographic properties of AM offer a clear opportunity to create basic two- or three-material phantoms. These would be high-accuracy and cost-effective models. Although the materials currently available are not suitable for complex multi-material applications as realistic as true human anatomy, one can easily foresee the development of new materials with broader density in the near future.

MRI of the Sacroiliac Joint in Healthy Children.

To listen to the podcast associated with this article, please select one of the following: iTunes, Google Play, or direct download. OBJECTIVE. Increasing reliance on MRI for the evaluation of sacroiliitis requires that radiologists be familiar with the normal appearance of the developing sacroiliac joint. We describe age-related MRI features of the sacroiliac joints in healthy children. SUBJECTS AND METHODS. Seventy healthy children from three age groups-prepubertal (8-10 years), peripubertal (11-13 years), and approaching skeletal maturity (14-17 years)-completed questionnaires and underwent sacroiliac joint MRI. Imaging studies were evaluated by three experienced pediatric radiologists. Metaphyseal-equivalent signal intensity, nonperiarticular osteitis, cartilage volume, joint fluid, enthesitis, and surface cortex irregularities were evaluated. Metaphyseal-equivalent signal intensity was evaluated using an ordinal grading system (types I-IV). Intraclass correlation coefficients were calculated to assess interrater reliability. RESULTS. Increased metaphyseal-equivalent signal intensity (types I and II) was present in most prepubertal children and in less than 10% of the group approaching skeletal maturity. More prepubertal girls had type I signal than boys, but signal progressed to type IV signal faster in girls than in boys. None of the subjects had subchondral marrow edema, but four subjects had nonperiarticular osteitis. Cartilage volumes decreased with advancing age in girls and, on average, were lower in girls than in boys of the same age. One subject had measurable joint fluid. Cortex irregularities were common (57.1%), most frequently seen along the iliac bone (51.4% ilium vs 11.4% sacrum, p < 0.01) and in the upper quadrants (42.8% upper vs 27.1% lower, p = 0.01). CONCLUSION. We provide valuable reference MRI descriptions of the healthy pediatric sacroiliac joint that should improve our ability to distinguish between normal and pathologic findings.

Clinical decision support: practical implementation at two pediatric hospitals.

Clinical decision support has been identified by the United States government as a method to decrease inappropriate imaging exams and promote judicious use of imaging resources. The adoption of this method will be incentivized by requiring appropriate use criteria to qualify for Medicare reimbursement starting in January 2020. While Medicare reimbursement is unlikely to directly impact pediatric imaging because of largely disparate patient populations, insurance providers typically use Medicare to benchmark their reimbursement guidelines. Therefore soon after their adoption these guidelines could become relevant to pediatric imaging. In this article we discuss how pediatric imaging was initially underrepresented in the clinical decision support realm, and how this was addressed by a subcommittee involving both American College of Radiology and Society for Pediatric Radiology members. We also present the experience of implementing clinical decision support software at two standalone pediatric hospitals and summarize the lessons learned from these deployments.

Clinical decision support: the role of ACR Appropriateness Criteria.

Clinical decision support is a way to decrease inappropriate imaging exams and promote judicious use of imaging resources. The adoption of clinical decision support will be incentivized by requiring the use of approved mechanisms to qualify for Medicare reimbursement starting in January 2020. Insurance providers base their reimbursement policies on Medicare, so clinical decision support could soon become relevant to pediatric imaging. We present the process behind the American College of Radiology (ACR) Appropriateness Criteria (a set of appropriate use criteria developed by the ACR) that will form the basis for software that can be used to fulfill the criteria for clinical decision support. For most organizations, this software is expected to be the easiest way to implement clinical decision support. Clinical decision support will affect how providers order imaging exams. This article should help readers understand how clinical decision support is expected to change the practice of the ordering providers, how the ACR Appropriateness Criteria are related to clinical decision support and how the ACR Appropriateness Criteria are developed. This will help the interpreting radiologist better communicate with the referring clinician, including informing the latter about how the clinical decision support software is making decisions.

Buckle fracture of the proximal tibia in children and frequency of association with trampoline and inflatable bouncer use.

BACKGROUND: Proximal tibial buckle fractures are often called "trampoline fractures," but mechanisms are more varied. OBJECTIVE: To assess the frequency of trampoline or bouncer etiology among children with buckle fracture of the proximal tibia, characterizing demographics, alternative etiologies, and fracture patterns. MATERIALS AND METHODS: We performed retrospective chart review of mechanisms, age and gender of children with radiographs interpreted as proximal tibial buckle fracture between 2010 and 2016. Two pediatric radiologists assessed presence of cortical impaction, cortical break, increased anterior scoop deformity, and oblique extension of fracture toward the physis. We studied the associations among demographics, etiology and fracture appearance using analysis of variance and chi-square/Fisher exact tests. RESULTS: We identified 145 exams interpreted as proximal tibial buckle fracture (median age 34 months, 64% girls). Bouncing surface was the most common mechanism (44%), and 80% of those cases reported multiple people on the bouncing surface. Falls were the second most common mechanism (30%). Children injured while bouncing were older (median 41 months) than others (median 21 months, P<0.005) and more likely to have oblique extension of fracture toward the physis (P<0.05). Buckle deformity was associated with a younger age (F=8.67; P<0.01), while oblique extension to the physis and concurrent fibula fracture were associated with older age (F=18.62, P<0.001; and F=8.02, P<0.01, respectively). CONCLUSION: Trampoline use was the most common single mechanism of injury in children with proximal tibial fracture interpreted as buckle deformity. However, non-bouncing mechanisms were overall more common and occurred in a younger age group at risk for "toddler fractures."

Learning, teaching and writing with reference managers.

Reference managers, also known as citation managers, were designed to create bibliographic citations but have evolved into powerful tools for education. In addition to their core competency of easily managing in-text citations and automatically generating a list of works cited, modern reference managers can help researchers manage increasingly large libraries of portable document format (PDF) files, web pages and other documents. Users can annotate, store, organize, tag and search PDFs. These programs fetch useful associated metadata such as publication information, author lists and article abstracts. Additionally, the programs create a searchable index that can include the article abstract and often the complete text of many articles' PDFs. Modern reference managers allow for rapid creation, organization and classification of a curated collection of reference articles that can be made available anywhere, including on mobile devices. In this article the author describes how pediatric radiologists can use reference managers to facilitate learning, teaching and writing.

Ultrasound features of purulent skin and soft tissue infection without abscess.

PURPOSE: Ultrasound (US) aids clinical management of skin and soft tissue infection (SSTI) by differentiating non-purulent cellulitis from abscess. However, purulent SSTI may be present without abscess. Guidelines recommend incision and drainage (I & D) for purulent SSTI, but US descriptions of purulent SSTI without abscess are lacking. METHODS: We retrospectively reviewed pediatric emergency department patients with US of the buttock read as negative for abscess. We identified US features of SSTI with adequate interobserver agreement (kappa > 0.45). Six independent observers then ranked presence or absence of these features on US exams. We studied association between US features and positive wound culture using logistic regression models (significance at p < 0.05). RESULTS: Of 217 children, 35 patients (16%) had cultures positive for pathogens by 8 h after US and 61 patients (32%) had cultures positive by 48 h after US. We found kappa > 0.45 for focal collection > 1.0 cm (κ = 0.57), hyperemia (κ = 0.57), swirling with compression (κ = 0.52), posterior acoustic enhancement (κ = 0.47), and cobblestoning or branching interstitial fluid (κ = 0.45). Only cobblestoning or interstitial fluid was associated with positive wound cultures in logistic regression models at 8 and 48 h. CONCLUSIONS: Cobblestoning or interstitial fluid on US may indicate presence of culture-positive, purulent SSTI in patients without US appearance of abscess. Although our study has limitations due to its retrospective design, this US appearance should alert imagers that the patient may benefit from early I & D.

Up-To-Date Practical Imaging Evaluation of Neonatal Soft-Tissue Tumors: What Radiologists Need to Know.

OBJECTIVE: The purposes of this article are to provide an up-to-date overview of neonatal soft-tissue tumors, including information regarding their unique nature, and to present practical imaging techniques and characteristic imaging findings. CONCLUSION: Neonatal soft-tissue tumors are a unique set of neoplasms that often have characteristic clinical and imaging findings. Imaging evaluation, mainly with ultrasound and MRI, plays an important role in the initial diagnosis, staging, preoperative assessment, and follow-up evaluation. Clear understanding of practical imaging techniques combined with up-to-date knowledge of characteristic imaging findings can help the radiologist provide a timely and accurate diagnosis of these neoplasms and can lead to optimal neonatal patient care.

Intraabdominal Lymphatic Malformations: Pearls and Pitfalls of Diagnosis and Differential Diagnoses in Pediatric Patients.

OBJECTIVE: The purpose of this article is to review the practical imaging workup and characteristic imaging appearances of intraabdominal lymphatic malformations (LMs) in the pediatric population with a brief discussion of some common differential diagnoses found in a vascular anomaly clinic. CONCLUSION: LMs are uncommon pediatric lesions. Because of their rarity among LMs overall, a tendency to present later in life than superficial LMs, and often incidental identification, intraabdominal LMs pose a particular diagnostic challenge, and pathologic entities that are more prevalent must be carefully excluded first. Although the diagnosis of most intraabdominal LMs can be reliably based on clear understanding of characteristic imaging findings, histologic correlation may be necessary in some cases.

MRI of Native Knee Cartilage Delamination Injuries.

OBJECTIVE: The purpose of this article is to describe the normal imaging appearance of cartilage and the pathophysiologic findings, imaging appearance, and surgical management of cartilage delamination. CONCLUSION: Delamination injuries of knee cartilage signify surgical lesions that can lead to significant morbidity without treatment. These injuries may present with clinical symptoms identical to those associated with meniscal injury, and arthroscopic identification can be difficult, thereby creating a role for imaging diagnosis. A low sensitivity of imaging identification of delamination injury of the knee is reported in the available literature, although vast improvements in MRI of cartilage have since been introduced.

Ultrasound assessment of the bowel: inflammatory bowel disease and conditions beyond.

Ultrasound (US) is a versatile imaging study for the evaluation of the bowel in children. US imaging of the bowel can be used as the initial examination or in follow-up for many common pediatric diseases. In this article, we highlight our bowel US technique and describe how US can depict the features of a select group of bowel pathologies relevant to pediatric practice.

Surveying Academic Radiology Department Chairs Regarding New and Effective Strategies for Medical Student Recruitment.

OBJECTIVE: The number of 4th-year medical student applications to the field of diagnostic radiology has decreased from 2009 to 2015. The purpose of this study was to learn how radiology departments are recruiting medical students. MATERIALS AND METHODS: An anonymous online survey hyperlink was distributed to the members of the Society of Chairs of Academic Radiology Departments regarding both innovative and proven recruitment strategies. The results were synthesized with a recently published survey of medical students about factors influencing them to go into radiology. RESULTS: Forty of 126 radiology departments completed the survey. Most felt that radiology exposure and curricula require alteration given recent downward trends in medical student applications. A majority (79%) had changed their outreach to medical students in response to these trends. The responding department chairs felt that interactive learning while on rotation was the most important strategy for recruitment. The presence of a diversity program, dedicated medical school educator, or rotating daily assignment for students did not affect the likelihood of filling residency spots in the main match. CONCLUSION: Many radiology departments are changing their outreach to medical students to improve recruitment. Effective strategies to focus on include early active outreach by involving students in the radiology department, thereby framing radiologists as clinicians.

Meniscal pathology in children: differences and similarities with the adult meniscus.

The normal meniscus undergoes typical developmental changes during childhood, reaching a mature adult appearance by approximately 10 years of age. In addition to recognizing normal meniscal appearances in children, identifying abnormalities - such as tears and the different types of discoid meniscus and meniscal cysts, as well as the surgical implications of these abnormalities - is vital in pediatric imaging. The reported incidence of meniscal tears in adolescents and young adults has increased because of increased sports participation and more widespread use of MRI. This review discusses the normal appearance of the pediatric meniscus, meniscal abnormalities, associated injuries, and prognostic indicators for repair.