Faculty Bibliography

PURPOSE/OBJECTIVE:To advance research in the field of machine learning for MR image reconstruction with an open challenge. METHODS:We provided participants with a dataset of raw k-space data from 1,594 consecutive clinical exams of the knee. The goal of the challenge was to reconstruct images from these data. In order to strike a balance between realistic data and a shallow learning curve for those not already familiar with MR image reconstruction, we ran multiple tracks for multi-coil and single-coil data. We performed a two-stage evaluation based on quantitative image metrics followed by evaluation by a panel of radiologists. The challenge ran from June to December of 2019. RESULTS:We received a total of 33 challenge submissions. All participants chose to submit results from supervised machine learning approaches. CONCLUSIONS:The challenge led to new developments in machine learning for image reconstruction, provided insight into the current state of the art in the field, and highlighted remaining hurdles for clinical adoption.

The ACR recognizes that radiology practices are grappling with when and how to safely resume routine radiology care during the coronavirus disease 2019 (COVID-19) pandemic. Although it is unclear how long the pandemic will last, it may persist for many months. Throughout this time, it will be important to perform safe, comprehensive, and effective care for patients with and patients without COVID-19, recognizing that asymptomatic transmission is common with this disease. Local idiosyncrasies prevent a single prescriptive strategy. However, general considerations can be applied to most practice environments. A comprehensive strategy will include consideration of local COVID-19 statistics; availability of personal protective equipment (PPE); local, state, and federal government mandates; institutional regulatory guidance; local safety measures; health care worker availability; patient and health care worker risk factors; factors specific to the indication(s) for radiology care; and examination or procedure acuity. An accurate risk-benefit analysis of postponing versus performing a given routine radiology examination or procedure often is not possible due to many unknown and complex factors. However, this is the overriding principle: If the risk of illness or death to a health care worker or patient from health care-acquired COVID-19 is greater than the risk of illness or death from delaying radiology care, the care should be delayed; however, if the opposite is true, the radiology care should proceed in a timely fashion.

RATIONALE AND OBJECTIVES/OBJECTIVE:The educational value of the daily resident readout, a vital component of resident training, has been markedly diminished due to a significant decrease in imaging volume and case mix diversity. The goal of this study was to create a "simulated" daily readout (SDR) to restore the educational value of the daily readout. MATERIALS AND METHODS/METHODS:To create the SDR the following tasks were performed; selection of cases for a daily worklist for each resident rotation, comprising a combination of normal and abnormal cases; determination of the correct number of cases and the appropriate mix of imaging modalities for each worklist; development of an "educational" environment consisting of separate "instances" of both our Picture Archive Communication System and reporting systems; and the anonymization of all of the cases on the worklists. Surveys of both residents and faculty involved in the SDR were performed to assess its effectiveness. RESULTS:Thirty-two residents participated in the SDR. The daily worklists for the first 20 days of the SDR included 3682 cases. An average of 480 cases per day was dictated by the residents. Surveys of the residents and the faculty involved in the SDR demonstrated that both agreed that the SDR effectively mimics a resident's daily work on rotations and preserves resident education during the Coronavirus Disease 2019 crisis. CONCLUSION/CONCLUSIONS:The development of the SDR provided an effective method of preserving the educational value of the daily readout experience of radiology residents, despite severe decreases in imaging exam volume and case mix diversity during the Coronavirus Disease 2019 pandemic.

Artificial intelligence (AI) has the potential to significantly disrupt the way radiology will be practiced in the near future, but several issues need to be resolved before AI can be widely implemented in daily practice. These include the role of the different stakeholders in the development of AI for imaging, the ethical development and use of AI in healthcare, the appropriate validation of each developed AI algorithm, the development of effective data sharing mechanisms, regulatory hurdles for the clearance of AI algorithms, and the development of AI educational resources for both practicing radiologists and radiology trainees. This paper details these issues and presents possible solutions based on discussions held at the 2019 meeting of the International Society for Strategic Studies in Radiology. KEY POINTS: • Radiologists should be aware of the different types of bias commonly encountered in AI studies, and understand their possible effects. • Methods for effective data sharing to train, validate, and test AI algorithms need to be developed. • It is essential for all radiologists to gain an understanding of the basic principles, potentials, and limits of AI.

A publicly available dataset containing k-space data as well as Digital Imaging and Communications in Medicine image data of knee images for accelerated MR image reconstruction using machine learning is presented.

OBJECTIVE. The purpose of this study is to assess the perceptions of radiologists and emergency medicine (EM) providers regarding the quality, value, and challenges associated with using outside imaging (i.e., images obtained at facilities other than their own institution). MATERIALS AND METHODS. We surveyed radiologists and EM providers at a large academic medical center regarding their perceptions of the availability and utility of outside imaging. RESULTS. Thirty-four of 101 radiologists (33.6%) and 38 of 197 EM providers (19.3%) responded. A total of 32.4% of radiologists and 55.3% of EM providers had confidence in the quality of images from outside community facilities; 20.6% and 44.7%, respectively, had confidence in the interpretations of radiologists from these outside facilities. Only 23.5% of radiologists and 5.3% of EM physicians were confident in their ability to efficiently access reports (for outside images, 47.1% and 5.3%). Very few radiologists and EM providers had accessed imaging reports from outside facilities through an available stand-alone portal. A total of 40.6% of radiologists thought that outside reports always or frequently reduced additional imaging recommendations (62.5% for outside images); 15.6% thought that reports changed interpretations of new examinations (37.5% for outside images); and 43.8% thought that reports increased confidence in interpretations of new examinations (75.0% for outside images). A total of 29.4% of EM providers thought that access to reports from outside facilities reduced repeat imaging (64.7% for outside images), 41.2% thought that they changed diagnostic or management plans (50.0% for outside images), and 50.0% thought they increased clinical confidence (67.6% for outside images). CONCLUSION. Radiologists and EM providers perceive high value in sharing images from outside facilities, despite quality concerns. Substantial challenges exist in accessing these images and reports from outside facilities, and providers are unlikely to do so using separate systems. However, even if information technology solutions for seamless image integration are adopted, providers' lack of confidence in outside studies may remain an important barrier.

Magnetic resonance imaging (MRI) is a leading image modality for the assessment of musculoskeletal (MSK) injuries and disorders. A significant drawback, however, is the lengthy data acquisition. This issue has motivated the development of methods to improve the speed of MRI. The field of artificial intelligence (AI) for accelerated MRI, although in its infancy, has seen tremendous progress over the past 3 years. Promising approaches include deep learning methods for reconstructing undersampled MRI data and generating high-resolution from low-resolution data. Preliminary studies show the promise of the variational network, a state-of-the-art technique, to generalize to many different anatomical regions and achieve comparable diagnostic accuracy as conventional methods. This article discusses the state-of-the-art methods, considerations for clinical applicability, followed by future perspectives for the field.

OBJECTIVE:Guideline-concordant follow-up of incidental lung nodules (ILNs) is suboptimal. We aimed to improve communication and tracking for follow-up of these common incidental findings detected on imaging examinations. METHODS:We implemented a process improvement program for reporting and tracking ILNs at a large urban academic health care system. A multidisciplinary committee designed, tested, and implemented a multipart tracking system in the electronic health record (EHR) that included Fleischner Society management recommendations for each patient. Plan-do-study-act cycles addressed gaps in the follow-up of ILNs, broken into phases of developing and testing components of the conceived EHR toolkit. RESULTS:The program resulted in standardized text macros with discrete categories and recommendations for ILNs, with ability to track each case in a work list within the EHR. The macros incorporated evidence-based guidelines and also input of collaborating clinical referrers in the respective specialty. The ILN macro was used 3,964 times over the first 2 years, increasing from 104 to over 300 uses per month. Usage spread across all subspecialty divisions, with nonthoracic radiologists currently accounting for 80% (56 of 70) of the radiologists using the system and 31% (1,230 of 3,964) of all captured ILNs. When radiologists indicated ILNs as warranting telephone communication to provider offices, completion was documented in 100% of the cases captured in the EHR-embedded tracking report. CONCLUSION/CONCLUSIONS:An EHR-based system for managing incidental nodules enables case tracking with exact recommendations, provider communication, and completion of follow-up testing. Future efforts will target consistent radiologist use of the system and follow-up completion.