Comparing Ultrasound with Magnetic Resonance Imaging in the Evaluation of Subungual Glomus Tumors and Subungual Myxoid Cysts.

Introduction: Subungual myxoid cysts and subungual glomus tumors demonstrate characteristic features on magnetic resonance imaging (MRI) and ultrasound (US). It is not yet well-established whether US is congruent to MRI in diagnostic evaluation of these subungual lesions. Methods: Participants with clinical suspicion for subungual glomus tumors or subungual myxoid cysts were recruited. After clinical evaluation, participants underwent radiography, MRI, and US plus biopsy, aspiration, or excision where possible. Differential diagnoses were revised after review of imaging, and imaging findings were compared to definitive diagnosis by pathology, aspiration, or clinical course. Results: All lesions were visible on both US and MRI and size estimates agreed between the two modalities. US and MRI findings of subungual glomus tumors and subungual myxoid cysts agreed with their known respective imaging characteristics. Conclusions: Diagnosis of subungual myxoid cysts and subungual glomus tumors agreed between US and MRI. We provide sample MRI and US imaging parameters for optimal evaluation of subungual myxoid cysts and glomus tumors. We demonstrate that subungual MRI evaluation can be performed without special equipment, allowing for evaluation by most radiology departments. Lastly, US is user-dependent and may be non-inferior for a sonographer familiar with subungual US.

Teaching with Technology-Matching Pedagogy with Purpose in Radiology Education.

The response to pandemic-related teaching disruption has revealed dynamic levels of learning and teaching flexibility and rapid technology adoption of radiology educators and trainees. Shutdowns and distancing requirements accelerated the adoption of technology as an educational tool, in some instances supplanting in-person education entirely. Despite the limitations of remote interaction, many educational advantages were recognized that can be leveraged in developing distance learning paradigms. The specific strategies employed should match modern learning science, enabling both students and educators to mutually grow as lifelong learners. As panel members of the "COVID: Faculty perspective" Task Force of the Association of University Radiologists Radiology Research Alliance, we present a review of key learning principles which educators can use to identify techniques that enhance resident learning and present an organized framework for applying technology-aided techniques aligned with modern learning principles. Our aim is to facilitate the purposeful integration of learning tools into the training environment by matching these tools to established educational frameworks. With these frameworks in mind, radiology educators have the opportunity to re-think the balance between traditional curricular design and modern digital teaching tools and models.

A "Bumper-Car" Curriculum for Teaching Deep Learning to Radiology Residents☆.

RATIONALE AND OBJECTIVES: Our goal was to create an artificial intelligence (AI) training curriculum for residents that taught them to create, train, evaluate and refine deep learning (DL) models. Hands-on training of models was emphasized and didactic presentations of the mathematical and programmatic underpinnings of DL were minimized. MATERIALS AND METHODS: We created a three-session, 6-hour curriculum based on a "no-code" machine learning system called Lobe.ai. This class met weekly in June 2021. Pre-class homework included reading assignments, software installation, dataset downloads, and image-collection and labeling. The class sessions included several short, didactic presentations, but were largely devoted to hands-on training of DL models. After the course, our residents completed a short, anonymous, online survey about the course. RESULTS: Our residents learned to acquire and label a wide variety of image datasets. They quickly learned to train DL models to classify these datasets, as well as how to evaluate and refine these models. Our survey showed that most residents felt AI to be important and worth learning, but most were not very interested in learning to program. Most felt that the course taught them useful things about DL, and they were now more interested in the topic. Most would recommend the course to other residents, as well as to medical students and to radiology faculty. CONCLUSION: The course met our objectives of teaching our residents to create, train, evaluate, and refine DL models. We hope that the hands-on experience they gained in this course will enable them to recognize problems in diagnostic AI systems, and to help solve such problems in their own radiology practices.

Bone and Soft Tissue Tumors: Horizons in Radiomics and Artificial Intelligence.

AI can improve the quality of CT, MR and PET/CT images, while simultaneously reducing imaging time, and doses of radiation and contrast. AI can improve radiologist workflow and decrease interpretation times. AI may someday be capable of accurately locating, classifying and segmenting bone and soft tissue tumors. The goal of radiomics is to use radiomic and other biomarkers to achieve "precision medicine", ie, to predict the right diagnosis and the right treatment of the right patient at the right time. Radiomic information may be helpful in guiding biopsies, classifying and grading tumors and predicting prognosis and treatment response.

Variation in Fetal Weight Percentile Estimates.

OBJECTIVES: Weight percentiles are generally reported without any indication of error. This variation can lead a fetus being mistakenly classified erroneously as having intrauterine growth restriction (IUGR) or macrosomia. The goal of this study was to compare estimated weight percentiles with the actual observed weight percentile for each gestational age in a large cohort of fetuses being scanned in our institution. METHODS: After IRB approval the radiology information system data base was retrospectively searched for all obstetrical US reports obtained during the late second and third trimesters from July 1, 2014, until July 1, 2020. Demographic information, fetal weight, and weight percentile information were obtained from these reports. Quantile-quantile plots were created for all gestational ages and all ethnicities. RESULTS: Our study included 6259 ultrasounds in 4060 patients. Mean maternal age of the total group was 31.68 years (ranging 15-53 years). When all subjects were considered, the median values in our QQ plots approximated the line of identity. However, there was considerable variation for a given estimate, implying that estimated fetal weight percentiles are only very rough predictors of the actual percentile. CONCLUSION: Estimated fetal weight percentiles are only very rough predictors of the actual percentile. We therefore suggest that estimates of the weight percentile should be reported along with an estimate of the expected variation. Recognition of variations in weight percentile should be considered in the greater clinical context, and could potentially prevent misdiagnosis of growth restriction and macrosomia as well as the subsequent overutilization of resources, unnecessary interventions, and maternal stress.

Exophytic External Occipital Protuberance Prevalence Pre- and Post-iPhone Introduction: A Retrospective Cohort.

Introduction: In controversial fashion, the presence of an enlarged external occipital protuberance has been recently linked to excessive use of handheld electronic devices. We sought to determine the prevalence of this protuberance in a diverse age group of adults from two separate time periods, before and approximately 10 years after the release of the iPhone, to further characterize this theory, as if indeed valid, such a relationship could direct preventative behavior. Materials and Methods: Eighty-two cervical spine radiographs between March 7, 2007 through June 29, 2007 and 147 cervical spine radiographs between October 25, 2017 through January 1, 2018 were reviewed for the presence or absence of an exophytic external occipital protuberance. Influence of sex and age were also assessed. Results: There were 41/82 (50%) patients within the 2007 pre-iPhone group with an exophytic external occipital protuberance, ranging from 2.7-33.8 mm in length. Twenty-seven out of 82 (32.9%) had an external occipital protuberance at or above 10 mm. There were 49/147 (33.3%) patients within the 2017 post-iPhone group with an exophytic external occipital protuberance, ranging from 4.4-53.8 mm in length. Thirty-three out of 147 (22.4%) had an external occipital protuberance at or above 10 mm. When considering accessibility to the iPhone, sex, and age to the presence of an exophytic external occipital protuberance, only sex has a statistically significant association, p=0.000000033. Conclusion: We found no significant association with iPhone accessibility and an exophytic external occipital protuberance. Due to inherent limitations in the retrospective nature of the study, future research is needed to better examine the association of handheld electronic devices with exophytic external occipital protuberances.

Review of Artificial Intelligence Training Tools and Courses for Radiologists.

Artificial intelligence (AI) systems play an increasingly important role in all parts of the imaging chain, from image creation to image interpretation to report generation. In order to responsibly manage radiology AI systems and make informed purchase decisions about them, radiologists must understand the underlying principles of AI. Our task force was formed by the Radiology Research Alliance (RRA) of the Association of University Radiologists to identify and summarize a curated list of current educational materials available for radiologists.

Noninterpretive Uses of Artificial Intelligence in Radiology.

We deem a computer to exhibit artificial intelligence (AI) when it performs a task that would normally require intelligent action by a human. Much of the recent excitement about AI in the medical literature has revolved around the ability of AI models to recognize anatomy and detect pathology on medical images, sometimes at the level of expert physicians. However, AI can also be used to solve a wide range of noninterpretive problems that are relevant to radiologists and their patients. This review summarizes some of the newer noninterpretive uses of AI in radiology.

MR Protocol Optimization With Deep Learning: A Proof of Concept.

PURPOSE: This study was performed to demonstrate that a properly trained convolutional neural net (CNN) can provide an acceptable surrogate for human readers when performing a protocol optimization study. Tears of the anterior cruciate ligament (ACL) were used as a proof of concept for this study. METHODS: Following institutional review board approval, a curated set of 2007 paired knee MR images was extracted from the author's picture archival and communications system for 1523 normal knees and 484 knees with torn ACLs. A pair (1 fat-saturated (FS) and 1 non-fat-saturated (NFS)) of midline sagittal images was extracted from each knee. CNNs were trained for both the FS and NFS image sets and used to make predictions on a previously unseen test set of images. RESULTS: Receiver operating characteristic area under the curve for the NFS and FS CNNs were, respectively, 0.9983 and 0.9988. Specificity was identical (0.993) for both NFS and FS images. FS sensitivity (0.98) and NFS sensitivity (0.88) were statistically significantly different (P = 0.0253). CONCLUSIONS: Both FS and NFS performed very well for the diagnosis of ACL tears, although FS sensitivity was superior to NFS sensitivity. The CNNs provided an acceptable surrogate for a human reader in this study. Pulse sequence optimization studies such as this can be opportunistically performed on image sets collected for many other machine learning purposes.

Capsular thinning on magnetic resonance arthrography is associated with intra-operative hip joint laxity in women.

Hip microinstability is a recognized cause of hip pain in young patients. Intra-operative evaluation is used to confirm the diagnosis, but limited data exist associating magnetic resonance arthrography (MRA) findings with hip microinstability. To determine if a difference exists in the thickness of the anterior joint capsule and/or the width of the anterior joint recess on MRA in hip arthroscopy patients with and without an intra-operative diagnosis of hip laxity. Sixty-two hip arthroscopy patients were included in the study. Two musculoskeletal radiologists blinded to surgical results reviewed the MRAs for two previously described findings: (i) anterior joint capsule thinning; (ii) widening of the anterior joint recess distal to the zona orbicularis. Operative reports were reviewed for the diagnosis of joint laxity. In all patients with and without intra-operative laxity, there were no significant differences with either MRA measurement. However, twenty-six of 27 patients with intra-operative laxity were women compared with 11 of 35 patients without laxity (P < 0.001). In subgroup analysis of women, the intra-operative laxity group had a higher rate of capsular thinning compared with the non-laxity group (85% versus 45%; P = 0.01). A 82% of women with capsular thinning also had intra-operative laxity, compared with 40% without capsular thinning (P = 0.01). There were no differences regarding the width of the anterior joint recess. In this study, there was an association between capsular thinning and intra-operative laxity in female patients. Measuring anterior capsule thickness on a pre-operative MRA may be useful for the diagnosis of hip microinstability.

Measuring and Teaching Confidence Calibration Among Radiologists: A Multi-Institution Study.

OBJECTIVE: Our purpose was to assess the calibration of resident, fellow, and attending radiologists on a simple image classification task (presence or absence of an anterior cruciate ligament [ACL] tear based on interpretation of sagittal proton density, fat-saturated MR images) and to assess whether teaching residents could improve their calibration. METHODS: We created a test containing 30 randomized, sagittal, proton density, fat-saturated MR images of the ACL (15 normal, 15 torn). This test was administered in person to 20 trainees and 3 attendings at one medical center in one state. An online version of the test was given to 23 trainees and 14 attendings from 11 other medical centers in nine other states. Subjects were asked to give their confidence level (0%-100%) that each ACL was torn. RESULTS: Cross-sectional data were collected from 60 radiologists (mean time after medical school = 9.3 years, minimum = 1 year, maximum = 36 years). This demonstrated a statistically significant improvement in calibration as a function of increasing experience (P = .020). Longitudinal data were collected from 12 trainees at the start and end of their musculoskeletal radiology rotation, with an intervening review of the primary and secondary signs of ACL tear on MR. A statistically significant improvement in calibration was noted during the rotation (P = .028). CONCLUSIONS: Confidence calibration is a promising tool for quality improvement and radiologist self-assessment. Our study showed that calibration loss improves with experience in radiologists tested on a common and clinically important image classification task. We also demonstrated that calibration can be successfully taught to residents over a relatively short period (2-4 weeks).

Confidence Calibration: An Introduction With Application to Quality Improvement.

A probabilistic forecast is one that assigns a probability (or likelihood) to the occurrence of an event. Radiologists commonly make probabilistic judgments in their reports, even if these predictions are not explicitly expressed as numbers. There are calls for radiologists to commit to their probabilistic predictions in a standardized fashion; however, without a mechanism for feedback, there is no opportunity for improvement. Analysis techniques familiar to radiologists (eg, calculation of sensitivity and specificity and construction of receiver operating characteristics curves) have a blind spot with regard to calibration of these probabilities to reality and are the main obstacle to improvement along this axis. We review statistical and graphical methods for calibration analysis in wider use outside the medical literature and present a framework for implementation of these techniques for quality improvement and radiologist self-assessment.

Disease progression in relation to pre-onset parity among women with rheumatoid arthritis.

OBJECTIVE: Rheumatoid arthritis (RA) often ameliorates during pregnancy and flares postpartum, but the relationship of pregnancy and childbirth to RA prognosis is unclear. We examined RA severity for association with parity prior to RA onset and asked whether time from birth (latency) and/or the mother's HLA genotype influenced results. METHODS: A cohort study was conducted of 222 women previously identified in a prospective study of newly diagnosed RA, who returned for follow-up evaluation a median of 8 years later. Stratified analyses using Mantel-Haenszel methods were conducted to evaluate 5 RA severity measures based on hand and wrist radiographs, physical exams, and Health Assessment Questionnaires for association with parity. RESULTS: Overall, we observed little evidence of altered risk of progression to severe RA in relation to pre-onset parity, adjusting for RA onset age and time to follow-up. Stratifying parous women who had only live births by latency (<15 years/15+ years) showed no difference in risk of severe RA compared to nulligravid women. Live birth deliveries were significantly protective for women with 0 but not for those with 1 or 2 copies of the RA risk-associated HLA-DRB1 shared epitope sequence for erosion score (RR 0.26 95% CI 0.09-0.89) and joint count (RR 0.28 95% CI 0.09-0.87). CONCLUSION: We observed little evidence of difference in severe RA by pre-onset parity overall. However, among women not predisposed to RA by possessing the risk-associated HLA genotype, parous women who had only live births had lower risk of progression to severe RA as measured by erosion score and joint count.

Peer Learning Through Multi-Institutional Web-based Case Conferences: Perceived Value (and Challenges) From Abdominal, Cardiothoracic, and Musculoskeletal Radiology Case Conference Participants.

RATIONALE AND OBJECTIVES: Peer learning is a case-based group-learning model intended to improve performance. In this descriptive paper, we describe multi-institutional, multi-subspecialty, web-based radiology case conferences and summarize the participants' experiences. MATERIALS AND METHODS: A semi-structured, 27-question survey was administered to radiologists participating in abdominal, cardiothoracic, and musculoskeletal case conferences. Survey questions included demographics, perceived educational value and challenges experienced. Survey question formats were continuous, binary, five-point Likert scale or text-based. The measures of central tendencies, proportions of responses and patterns were tabulated. RESULTS: From 57 responders, 12/57 (21.1%) were abdominal, 16/57 (28.1%) were cardiothoracic, and 29/57 (50.8%) were musculoskeletal conference participants; 50/56 (89.3%) represented academic practice. Median age was 45 years (range 35-74); 43/57 (75.4%) were male. Geographically, 16/52 (30.8%) of participants were from the East Coast, 16/52 (30.8%) Midwest, 18/52 (34.6%) West Coast, and 2/52 (3.8%) International. The median reported educational value was 5/5 (interquartile range 5-5). Benefits of the case conference included education (50/95, 52.6%) and networking (39/95, 41.1%). Participants reported presenting the following cases: "great call" 32/48 (66.7%), learning opportunity 32/48 (66.7%), new knowledge 41/49 (83.7%), "zebras" 46/49 (93.9%), and procedural-based 16/46 (34.8%). All 51/51 (100%) of responders reportedly gained new knowledge, 49/51 (96.1%) became more open to group discussion, 34/51 (66.7%) changed search patterns, and 50/51 (98%) would continue to participate. Reported challenges included time zone differences and support from departments for a protected time to participate. CONCLUSION: Peer learning through multi-institutional case conferences provides educational and networking opportunities. Current challenges and desires include having department-supported protected time and ability to receive continuing medical education credit.

Advanced, Interactive, Image-based Education: Technology and Pedagogy.

An audience response system (ARS) is an excellent tool for improving interactive learning in radiology residents. Traditional ARSs have long allowed text-based interactions between teacher and students. However, little attention has been given to techniques which allow students in large groups to interact directly with an image. Fortunately, a growing number of ARSs are beginning to add this ability. However, it is not the technology but the pedagogy that matters the most. The purpose of this article is to review those ARSs, and to present an array of pedagogical techniques that can take advantage of this technology.

Immunogenicity of a rheumatoid arthritis protective sequence when acquired through microchimerism.

HLA class II genes provide the strongest genetic contribution to rheumatoid arthritis (RA). HLA-DRB1 alleles encoding the sequence DERAA are RA-protective. Paradoxically, RA risk is increased in women with DERAA+ children born prior to onset. We developed a sensitive qPCR assay specific for DERAA, and found 53% of DERAA-/- women with RA had microchimerism (Mc; pregnancy-derived allogeneic cells) carrying DERAA (DERAA-Mc) vs. 6% of healthy women. DERAA-Mc quantities correlated with an RA-risk genetic background including DERAA-binding HLA-DQ alleles, early RA onset, and aspects of RA severity. CD4+ T cells showed stronger response against DERAA+ vs. DERAA- allogeneic cell lines in vitro, in line with an immunogenic role of allogeneic DERAA. Results indicate a model where DERAA-Mc activates DERAA-directed T cells that are naturally present in DERAA-/- individuals and can have cross-reactivity against joint antigens. Moreover, we provide an explanation for the enigmatic observation that the same HLA sequence differentially affects RA risk through Mendelian inheritance vs. microchimeric cell acquisition.

A Review of Innovative Teaching Methods.

Teaching is one of the important roles of an academic radiologist. Therefore, it is important that radiologists are taught how to effectively educate and, in turn, to act as role models of these skills to trainees. This is reinforced by the Liaison Committee on Medical Education which has the requirement that all residents who interact with and teach medical students must undergo training in effective methods of teaching. Radiologists are likely familiar with the traditional didactic lecture-type teaching format. However, there are many newer innovative teaching methods that could be added to the radiologist's teaching repertoire, which could be used to enhance the traditional lecture format. The Association of University Radiologists Radiology Research Alliance Task Force on Noninterpretive Skills therefore presents a review of several innovative teaching methods, which include the use of audience response technology, long-distance teaching, the flipped classroom, and active learning.