Integrating human factors engineering into your pediatric radiology practice.

Human factors engineering involves the study and development of methods aimed at enhancing performance, improving safety, and optimizing user satisfaction. The focus of human factors engineering encompasses the design of work environments and an understanding of human mental processes to prevent errors. In this review, we summarize the history, applications, and impacts of human factors engineering on the healthcare field. To illustrate these applications and impacts, we provide several examples of how successful integration of a human factors engineer in our pediatric radiology department has positively impacted various projects. The successful integration of human factors engineering expertise has contributed to projects including improving response times for portable radiography requests, deploying COVID-19 response resources, informing the redesign of scheduling workflows, and implementation of a virtual ergonomics program for remote workers. In sum, the integration of human factors engineering insight into our department has resulted in tangible benefits and has also positioned us as proactive contributors to broader hospital-wide improvements.

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 development in pediatric body magnetic resonance imaging: best ideas to adapt from adults.

Emerging manifestations of artificial intelligence (AI) have featured prominently in virtually all industries and facets of our lives. Within the radiology literature, AI has shown great promise in improving and augmenting radiologist workflow. In pediatric imaging, while greatest AI inroads have been made in musculoskeletal radiographs, there are certainly opportunities within thoracoabdominal MRI for AI to add significant value. In this paper, we briefly review non-interpretive and interpretive data science, with emphasis on potential avenues for advancement in pediatric body MRI based on similar work in adults. The discussion focuses on MRI image optimization, abdominal organ segmentation, and osseous lesion detection encountered during body MRI in children.

MRI evaluation of pediatric tibial eminence fractures: comparison between conventional and "CT-like" ultrashort echo time (UTE) images.

OBJECTIVE: UTE MRI offers a radiation-free alternative to CT for bone depiction, but data on children is lacking. The purpose of this study was to determine whether UTE images improve detection and characterization of pediatric tibial eminence fractures. METHODS: Fifteen MRIs with UTE from 12 children (10 boys, 2 girls; mean age: 12.6 ± 3.3 years) with tibial eminence fractures (2018-2020) and 15 age-matched MRIs without fractures were included. After randomization, 5 readers reviewed images without and with UTE, at least 1 month apart, and recorded the presence of fracture and preferred images. If fracture is present, radiologists also recorded fragment size, number, and displacement; surgeons assigned Meyers-McKeever grade and management. Disagreements on management were resolved through consensus review. Kappa and intra-class correlation (ICC), sensitivity, and specificity were used to compare agreement between readers and fracture detection between images without and with UTE. RESULTS: For fracture detection, inter-reader agreement was almost perfect (κ-range: 0.91-0.93); sensitivity and specificity were equivalent between images without and with UTE (range: 95-100%). For fracture characterization, UTE improved agreement on size (ICC = 0.88 to 0.93), number (ICC = 0.52 to 0.94), displacement (ICC = 0.74 to 0.86), and grade (ICC = 0.92 to 0.93) but reduced agreement on management (κ = 0.68 to 0.61), leading to a change in consensus management in 20% (3/15). Radiologists were more likely to prefer UTE for fracture and conventional images for non-fracture cases (77% and 77%, respectively, p < 0.001). CONCLUSION: While UTE did not improve diagnosis, it improved agreement on characterization of pediatric tibial eminence fractures, ultimately changing the preferred treatment in 20%.

Pediatric musculoskeletal pathologies: are there differences in triage of diagnoses and preferences for communication between radiology and orthopedics?

OBJECTIVE: To define the clinical importance of various pediatric musculoskeletal diagnoses, determine preferred communication methods based on the acuity level of findings, and investigate differences between specialties utilizing the Delphi methodology. METHODS: Radiologists, orthopedic surgeons, and sports-medicine pediatricians at a tertiary children's hospital were surveyed (n = 79) twice using REDCap (Research Electronic Data Capture). Surveys were conducted anonymously and at least 1 year apart, first eliciting all potentially non-routine findings and various communication methods (round 1), and later categorizing the acuity (emergent, urgent, or non-urgent) of different diagnosis categories and selecting the preferred communication method (verbal, written electronic messages, and report) and timeframe (round 2). Chi-square, Fisher's exact, and Kruskal-Wallis H tests were used to compare variables between specialties. RESULTS: Round 1 produced 267 entries for non-routine findings (grouped into 19 diagnoses) and 71 for communication methods (grouped into 3 categories). Round 2 found no significant difference in the acuity assignments for the 19 predetermined diagnoses (p = 0.66) between the 3 specialties; however, there was reduced agreement for the top urgent diagnoses within and between specialties. Most pediatricians preferred written electronic messages. The preferred communication timeframe for urgent diagnoses was significantly different (< 2 h for pediatricians, < 4 h for radiologists, and < 8 h for surgeons; p = 0.003) between specialties whereas no difference was found for emergent (p = 1) and non-urgent diagnoses (p = 0.80). CONCLUSION: Acuity assignment for the 19 pediatric-specific musculoskeletal diagnoses was not significantly different between specialties, but the preferred communication timeframe for urgent diagnoses was significantly different, ranging between 2 and 8 h.

Deep Learning Measurement of Leg Length Discrepancy in Children Based on Radiographs.

Background Radiographic measurement of leg length discrepancy (LLD) is time consuming yet cognitively simple for pediatric radiologists. Purpose To compare deep learning (DL) measurements of LLD in pediatric patients to measurements performed by radiologists. Materials and Methods For this HIPAA-compliant retrospective study, radiographs obtained to evaluate LLD in children between January and August 2018 were identified. LLD was automatically measured by means of image segmentation followed by leg length calculation. On training data, a DL model was trained to segment femurs and tibias on radiographs. The validation set was used to select the optimized model. On testing data, leg lengths were calculated from segmentation masks and compared with measurements from the radiology report. Statistical analysis was performed by using a paired Wilcoxon signed-rank test to compare DL calculations and radiology reports. In addition, the measurement time was manually assessed by a pediatric radiologist and automatically assessed by the DL model on a randomly chosen group of 26 cases; the values were compared with the paired Wilcoxon signed-rank test. Results Radiographs obtained to evaluate LLD in 179 children (mean age ± standard deviation, 12 years ± 3; age range, 5-19 years; 89 boys and 90 girls) were evaluated. Radiographs were randomly divided into training, validation, and testing sets and consisted of studies from 70, 32, and 77 patients, respectively. In the training and validation sets, the DL model showed a high spatial overlap between manual and automatic segmentation masks of pediatric legs (Dice similarity coefficient, 0.94). For the testing set, the correlation between radiology reports and DL-calculated lengths of separated femurs and tibias (r = 0.99; mean absolute error [MAE], 0.45 cm), full pediatric leg lengths (r = 0.99; MAE, 0.45 cm), and full LLD (r = 0.92; MAE, 0.51 cm) was high (P < .001 for all correlations). Calculation time for the DL method per radiograph was faster than the mean time for radiologist manual calculation (1 second vs 96 seconds ± 7, respectively; P < .001). Conclusion A deep learning algorithm measured pediatric leg lengths with high spatial overlap compared with manual measurement at a rate 96 times faster than that of subspecialty-trained pediatric radiologists. © RSNA, 2020 See also the editorial by van Rijn and De Luca in this issue.

The growing issue of burnout in radiology - a survey-based evaluation of driving factors and potential impacts in pediatric radiologists.

BACKGROUND: Burnout in medicine, and specifically radiology, has been receiving more attention. Little data-driven literature is available regarding risk factors/causes to ultimately help guide the development of potential solutions. OBJECTIVE: To survey pediatric radiologists, a cohort with a documented high prevalence of burnout, and to understand the impact of clinical demands on nonclinical tasks and the implications of burnout on mental health. MATERIALS AND METHODS: A survey of Society for Pediatric Radiology (SPR) North America attendings was performed regarding institutional factors contributing to burnout, including call burden, clinical demands, departmental support and administrative/academic tasks. Questions regarding mental health and wellness resources were also included. Generalized linear modeling assuming binomial distribution was used for analyses with SAS 9.4. RESULTS: The response rate was 305/1,282 (24%) with 53% of respondents female. Respondents reported that both the number and complexity of clinical cases have increased since they first started practice as an attending, while the time for interpretation has not changed, P<0.0001. Using a scale of 0 (never), 1 (rarely), 2 (sometimes), 3 (frequently) and 4 (always), covering multiple hospitals (2.2) and administrative tasks (2.4) were the most stressful job factors. For those in administrative roles, the most stressful job factors were job-related tasks affected teaching duties (2.0) and decreased overall job satisfaction (2.0). Of the respondents, 52% said they know a physician affected by work stress-related mental illness and 25% know a physician who has contemplated or committed suicide. While 39% of the respondents have resources available to address burnout, only 33% utilize these resources. CONCLUSION: Increasing clinical demands and additional institutional/departmental factors play a potential role in burnout, which has serious implications for the mental health of pediatric radiologists.

'Walk in my shoes': intradepartmental role shadowing to increase workplace collegiality and wellness in a large pediatric radiology department.

BACKGROUND: Our nearly 500-member department implemented the shadowing program "Walk in My Shoes" to improve intradepartmental relationships and build a stronger sense of community. The program provides both clinical and non-clinical employees an opportunity to shadow colleagues in their various roles and learn more about one another's contribution to the overarching mission of caring for children and their families. The goal of this study was to understand the impact of the shadowing program on employee perceptions of various roles. OBJECTIVE: To bridge the gap of understanding among colleagues in order to strengthen workplace interrelatedness, increase understanding of various roles, and decrease preconceived notions about roles, through shadowing. MATERIALS AND METHODS: A preliminary survey distributed to our department in August 2018 assessed the level of interest in new wellness initiatives, including the shadowing program. The survey gauged which roles participants were interested in shadowing. The survey results revealed that 67 employees were interested in the shadowing program. We selected 39 participants and matched them to a coworker in their area of interest. The roles for shadowing included administrator, Child Life specialist, information technologist, medical assistant, nurse, radiologist, researcher and technologist. Participants were required to complete pre- and post-shadowing surveys to assess their experience. Individuals who hosted the shadow experience also completed a survey. RESULTS: A total of 39 clinical and non-clinical staff members participated in the program. We summarized the pre- and post-survey data using median and interquartile range (IQR) and compared the results using the Wilcoxon signed-rank test. The distribution of preconceived notions about each role was not significantly different between the pre- and post-surveys (P=0.094). However, participants' value, understanding of the role they shadowed, and understanding of how the roles relate to each other were significantly greater (P<0.001). In addition, participants showed great interest in shadowing the specific role again (82%) and shadowing another role (92%). Furthermore, almost all hosts would repeat the experience (96%). CONCLUSION: Our study showed that intradepartmental shadowing can improve clinical and non-clinical staff employees' perceptions and understanding of each other's roles in overall patient care, which in turn contributes to the broader initiative of workplace wellness. The enthusiasm and willingness of the hosts were essential for sustainability of the program and demonstrated that this type of program is feasible in a large, busy department.

A primer for pediatric radiologists on infection control in an era of COVID-19.

Pediatric radiology departments across the globe face unique challenges in the midst of the current COVID-19 pandemic that have not been addressed in professional guidelines. Providing a safe environment for personnel while continuing to deliver optimal care to patients is feasible when abiding by fundamental recommendations. In this article, we review current infection control practices across the multiple pediatric institutions represented on the Society for Pediatric Radiology (SPR) Quality and Safety committee. We discuss the routes of infectious transmission and appropriate transmission-based precautions, in addition to exploring strategies to optimize personal protective equipment (PPE) supplies. This work serves as a summary of current evidence-based recommendations for infection control, and current best practices specific to pediatric radiologists.

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.

If Disney ran your pediatric radiology department: a different approach to improving the patient and family experience.

The concepts behind the patient experience and patient- and family-centered care have their roots in the 1980s. Prioritization and implementation of programs to improve the patient experience have received increased attention since the passage of legislation tying health insurance reimbursement to patient satisfaction surveys. Radiology has joined these efforts with the Radiology 3.0 initiative, and departments are applying established patient- and family-centered care models and quality-improvement methods to improve patient experience and satisfaction. While these approaches are valuable, they should be supplemented with more qualitative, humanistic and empathetic approaches. We present a "Disney model" for improving the patient and family experience in pediatric radiology and examples of practical implementation.

Machine learning concepts, concerns and opportunities for a pediatric radiologist.

Machine learning, a subfield of artificial intelligence, is a rapidly evolving technology that offers great potential for expanding the quality and value of pediatric radiology. We describe specific types of learning, including supervised, unsupervised and semisupervised. Subsequently, we illustrate two core concepts for the reader: data partitioning and under/overfitting. We also provide an expanded discussion of the challenges of implementing machine learning in children's imaging. These include the requirement for very large data sets, the need to accurately label these images with a relatively small number of pediatric imagers, technical and regulatory hurdles, as well as the opaque character of convolution neural networks. We review machine learning cases in radiology including detection, classification and segmentation. Last, three pediatric radiologists from the Society for Pediatric Radiology Quality and Safety Committee share perspectives for potential areas of development.

Imaging of developmental dysplasia of the hip: ultrasound, radiography and magnetic resonance imaging.

Developmental dysplasia of the hip (DDH) describes a broad spectrum of developmental abnormalities of the hip joint that are traditionally diagnosed during infancy. Because the development of the hip joint is a dynamic process, optimal treatment depends not only on the severity of the dysplasia, but also on the age of the child. Various imaging modalities are routinely used to confirm suspected diagnosis, to assess severity, and to monitor treatment response. For infants younger than 4 months, screening hip ultrasound (US) is recommended only for those with risk factors, equivocal or positive exam findings, whereas for infants older than 4-6 months, pelvis radiography is preferred. Following surgical hip reduction, magnetic resonance (MR) imaging is preferred over computed tomography (CT) because MR can not only confirm concentric hip joint reduction, but also identify the presence of soft-tissue barriers to reduction and any unexpected postoperative complications. The routine use of contrast-enhanced MR remains controversial because of the relative paucity of well-powered and validated literature. The main objectives of this article are to review the normal and abnormal developmental anatomy of the hip joint, to discuss the rationale behind the current recommendations on the most appropriate selection of imaging modalities for screening and diagnosis, and to review routine and uncommon findings that can be identified on post-reduction MR, using an evidence-based approach. A basic understanding of the physiology and the pathophysiology can help ensure the selection of optimal imaging modality and reduce equivocal diagnoses that can lead to unnecessary treatment.

Practical considerations for establishing and maintaining a magnetic resonance imaging safety program in a pediatric practice.

Magnetic resonance imaging is a multipurpose imaging modality that is largely safe, given the lack of ionizing radiation. However there are electromagnetic and biological effects on human tissue when exposed to magnetic environments, and hence there is a risk of adverse events occurring with these exams. It is imperative to understand these risks and develop methods to minimize them and prevent consequent adverse events. Implementing these safety practices in pediatric MR imaging has been somewhat limited because of gaps in information and knowledge among the personnel who are closely involved in the MR environment. The American College of Radiology has provided guidelines on MR safety practices that are helpful in minimizing such adverse events. This article provides an overview of the issues related to MR safety and practical ways to implement them across different health care facilities.

Practical considerations when implementing peer learning conferences.

Peer learning represents a shift away from traditional peer review. Peer learning focuses on improvement of diagnostic performance rather than on suboptimal performance. The shift in focus away from random selection and toward identification of cases with valuable teaching points can encourage more active radiologist engagement in the learning process. An effective peer learning program relies on a trusting environment that lessens the fear of embarrassment or punitive action. Here we describe the shortcomings of traditional peer review, and the benefits of peer learning. We also provide tips for a successful peer learning program and examples of implementation.

Survey of peer review programs among pediatric radiologists: report from the SPR Quality and Safety Committee.

During the last 15 years, peer review has been widely incorporated into radiology quality improvement programs. However, current implementations are variable and carry concerns, including subjectivity of numerical scores and a sense of merely satisfying regulatory requirements. The Society for Pediatric Radiology (SPR) Quality and Safety Committee sought to evaluate the state of peer review programs in pediatric radiology practices, including implementation methods, perceived functions, strengths and weaknesses, and opportunities for improvement. We distributed an online 16-question survey to SPR members. Questions pertained to the type of peer review system, the use of numerical scores and comments, how feedback on discordances is given and received, and the use of peer learning conferences. We collected 219 responses (15% of survey invitations), 80% of which were from children's hospitals. Fifty percent of respondents said they use a picture archiving and communication system (PACS)-integrated peer review system. Comment-enhanced feedback for interpretive discordances was either very important or somewhat important to performance improvement in 86% of responses, compared to 48% with a similar perception of numerical scores. Sixty-eight percent of respondents said they either rarely or never check their numerical scores, and 82% either strongly or somewhat agreed that comments are more effective feedback than numerical scores. Ninety-three percent either strongly or somewhat agreed that peer learning conferences would be beneficial to their practice. Forty-eight percent thought that their current peer review system should be modified. Survey results demonstrate that peer review systems in pediatric radiology practices are implemented variably, and nearly half of respondents believe their systems should be modified. Most respondents prefer feedback in the form of comments and peer learning conferences, which are thought to be more beneficial for performance improvement than numerical scores.

Biofunctionalized gadolinium-containing prussian blue nanoparticles as multimodal molecular imaging agents.

Molecular imaging agents enable the visualization of phenomena with cellular and subcellular level resolutions and therefore have enormous potential in improving disease diagnosis and therapy assessment. In this article, we describe the synthesis, characterization, and demonstration of core-shell, biofunctionalized, gadolinium-containing Prussian blue nanoparticles as multimodal molecular imaging agents. Our multimodal nanoparticles combine the advantages of MRI and fluorescence. The core of our nanoparticles consists of a Prussian blue lattice with gadolinium ions located within the lattice interstices that confer high relaxivity to the nanoparticles providing MRI contrast. The relaxivities of our nanoparticles are nearly nine times those observed for the clinically used Magnevist. The nanoparticle MRI core is biofunctionalized with a layer of fluorescently labeled avidin that enables fluorescence imaging. Biotinylated antibodies are attached to the surface avidin and confer molecular specificity to the nanoparticles by targeting cell-specific biomarkers. We demonstrate our nanoparticles as multimodal molecular imaging agents in an in vitro model consisting of a mixture of eosinophilic cells and squamous epithelial cells. Our nanoparticles specifically detect eosinophilic cells and not squamous epithelial cells, via both fluorescence imaging and MRI in vitro. These results suggest the potential of our biofunctionalized Prussian blue nanoparticles as multimodal molecular imaging agents in vivo.

Artificial Intelligence Curriculum Needs Assessment for a Pediatric Radiology Fellowship Program: What, How, and Why?

RATIONALE AND OBJECTIVES: Artificial intelligence (AI) holds enormous potential for improvements in patient care, efficiency, and innovation in pediatric radiology practice. Although there is a pressing need for a radiology-specific training curriculum and formalized AI teaching, few resources are available. The purpose of our study was to perform a needs assessment for the development of an AI curriculum during pediatric radiology training and continuing education. MATERIALS AND METHODS: A focus group study using a semistructured moderator-guided interview was conducted with radiology trainees' and attending radiologists' perceptions of AI, perceived competence in interpretation of AI literature, and perceived expectations from radiology AI educational programs. The focus group was audio-recorded, transcribed, and thematic analysis was performed. RESULTS: The focus group was held virtually with seven participants. The following themes we identified: (1) AI knowledge, (2) previous training, (3) learning preferences, (4) AI expectations, and (5) AI concerns. The participants had no previous formal training in AI and variability in perceived needs and interests. Most preferred a case-based approach to teaching AI. They expressed incomplete understanding of AI hindered its clinical applicability and reiterated a need for improved training in the interpretation and application of AI literature in their practice. CONCLUSION: We found heterogeneity in perspectives about AI; thus, a curriculum must account for the wide range of these interests and needs. Teaching the interpretation of AI research methods, literature critique, and quality control through implementation of specific scenarios could engage a variety of trainees from different backgrounds and interest levels while ensuring a baseline level of competency in AI.