Optimal Large Language Model Characteristics to Balance Accuracy and Energy Use for Sustainable Medical Applications.

Background Large language models (LLMs) for medical applications use unknown amounts of energy, which contribute to the overall carbon footprint of the health care system. Purpose To investigate the tradeoffs between accuracy and energy use when using different LLM types and sizes for medical applications. Materials and Methods This retrospective study evaluated five different billion (B)-parameter sizes of two open-source LLMs (Meta's Llama 2, a general-purpose model, and LMSYS Org's Vicuna 1.5, a specialized fine-tuned model) using chest radiograph reports from the National Library of Medicine's Indiana University Chest X-ray Collection. Reports with missing demographic information and missing or blank files were excluded. Models were run on local compute clusters with visual computing graphic processing units. A single-task prompt explained clinical terminology and instructed each model to confirm the presence or absence of each of the 13 CheXpert disease labels. Energy use (in kilowatt-hours) was measured using an open-source tool. Accuracy was assessed with 13 CheXpert reference standard labels for diagnostic findings on chest radiographs, where overall accuracy was the mean of individual accuracies of all 13 labels. Efficiency ratios (accuracy per kilowatt-hour) were calculated for each model type and size. Results A total of 3665 chest radiograph reports were evaluated. The Vicuna 1.5 7B and 13B models had higher efficiency ratios (737.28 and 331.40, respectively) and higher overall labeling accuracy (93.83% [3438.69 of 3665 reports] and 93.65% [3432.38 of 3665 reports], respectively) than that of the Llama 2 models (7B: efficiency ratio of 13.39, accuracy of 7.91% [289.76 of 3665 reports]; 13B: efficiency ratio of 40.90, accuracy of 74.08% [2715.15 of 3665 reports]; 70B: efficiency ratio of 22.30, accuracy of 92.70% [3397.38 of 3665 reports]). Vicuna 1.5 7B had the highest efficiency ratio (737.28 vs 13.39 for Llama 2 7B). The larger Llama 2 70B model used more than seven times the energy of its 7B counterpart (4.16 kWh vs 0.59 kWh) with low overall accuracy, resulting in an efficiency ratio of only 22.30. Conclusion Smaller fine-tuned LLMs were more sustainable than larger general-purpose LLMs, using less energy without compromising accuracy, highlighting the importance of LLM selection for medical applications. © RSNA, 2024 Supplemental material is available for this article.

Environmental Sustainability and AI in Radiology: A Double-Edged Sword.

According to the World Health Organization, climate change is the single biggest health threat facing humanity. The global health care system, including medical imaging, must manage the health effects of climate change while at the same time addressing the large amount of greenhouse gas (GHG) emissions generated in the delivery of care. Data centers and computational efforts are increasingly large contributors to GHG emissions in radiology. This is due to the explosive increase in big data and artificial intelligence (AI) applications that have resulted in large energy requirements for developing and deploying AI models. However, AI also has the potential to improve environmental sustainability in medical imaging. For example, use of AI can shorten MRI scan times with accelerated acquisition times, improve the scheduling efficiency of scanners, and optimize the use of decision-support tools to reduce low-value imaging. The purpose of this Radiology in Focus article is to discuss this duality at the intersection of environmental sustainability and AI in radiology. Further discussed are strategies and opportunities to decrease AI-related emissions and to leverage AI to improve sustainability in radiology, with a focus on health equity. Co-benefits of these strategies are explored, including lower cost and improved patient outcomes. Finally, knowledge gaps and areas for future research are highlighted.

Planetary Health and Radiology: Why We Should Care and What We Can Do.

Climate change adversely affects the well-being of humans and the entire planet. A planetary health framework recognizes that sustaining a healthy planet is essential to achieving individual, community, and global health. Radiology contributes to the climate crisis by generating greenhouse gas (GHG) emissions during the production and use of medical imaging equipment and supplies. To promote planetary health, strategies that mitigate and adapt to climate change in radiology are needed. Mitigation strategies to reduce GHG emissions include switching to renewable energy sources, refurbishing rather than replacing imaging scanners, and powering down unused scanners. Radiology departments must also build resiliency to the now unavoidable impacts of the climate crisis. Adaptation strategies include education, upgrading building infrastructure, and developing departmental sustainability dashboards to track progress in achieving sustainability goals. Shifting practices to catalyze these necessary changes in radiology requires a coordinated approach. This includes partnering with key stakeholders, providing effective communication, and prioritizing high-impact interventions. This article reviews the intersection of planetary health and radiology. Its goals are to emphasize why we should care about sustainability, showcase actions we can take to mitigate our impact, and prepare us to adapt to the effects of climate change. © RSNA, 2024 Supplemental material is available for this article. See also the article by Ibrahim et al in this issue. See also the article by Lenkinski and Rofsky in this issue.

Gender Affirmation Surgery: A Primer on Imaging Correlates for the Radiologist.

OBJECTIVE. The purpose of this article is to provide an overview of common gender affirmation surgical therapies, define key anatomy, and describe select complications using multidisciplinary, multimodality approaches. CONCLUSION. Gender affirmation therapy may be tailored to the needs of each individual patient. There are three major categories of gender affirmation surgery: genital reconstruction (comprising vaginoplasty and either metoidioplasty or phalloplasty), body contouring, and maxillofacial contouring (facial feminization or masculinization). If encountered in diagnostic imaging, routine evaluation should take into consideration normal postsurgical anatomy and key associated unique complications.

General-Purpose Large Language Models Versus a Domain-Specific Natural Language Processing Tool for Label Extraction From Chest Radiograph Reports.

GPT-4 outperformed a radiology domain-specific natural language processing model in classifying imaging findings from chest radiograph reports, both with and without predefined labels. Prompt engineering for context further improved performance. The findings indicate a role for large language models to accelerate artificial intelligence model development in radiology by automating data annotation.

Proceedings From the 2022 ACR-RSNA Workshop on Safety, Effectiveness, Reliability, and Transparency in AI.

Despite the surge in artificial intelligence (AI) development for health care applications, particularly for medical imaging applications, there has been limited adoption of such AI tools into clinical practice. During a 1-day workshop in November 2022, co-organized by the ACR and the RSNA, participants outlined experiences and problems with implementing AI in clinical practice, defined the needs of various stakeholders in the AI ecosystem, and elicited potential solutions and strategies related to the safety, effectiveness, reliability, and transparency of AI algorithms. Participants included radiologists from academic and community radiology practices, informatics leaders responsible for AI implementation, regulatory agency employees, and specialty society representatives. The major themes that emerged fell into two categories: (1) AI product development and (2) implementation of AI-based applications in clinical practice. In particular, participants highlighted key aspects of AI product development to include clear clinical task definitions; well-curated data from diverse geographic, economic, and health care settings; standards and mechanisms to monitor model reliability; and transparency regarding model performance, both in controlled and real-world settings. For implementation, participants emphasized the need for strong institutional governance; systematic evaluation, selection, and validation methods conducted by local teams; seamless integration into the clinical workflow; performance monitoring and support by local teams; performance monitoring by external entities; and alignment of incentives through credentialing and reimbursement. Participants predicted that clinical implementation of AI in radiology will continue to be limited until the safety, effectiveness, reliability, and transparency of such tools are more fully addressed.

Evaluation of Climate-Aware Metrics Tools for Radiology Informatics and Artificial Intelligence: Toward a Potential Radiology Ecolabel.

Radiology is a major contributor to health care's impact on climate change, in part due to its reliance on energy-intensive equipment as well as its growing technological reliance. Delivering modern patient care requires a robust informatics team to move images from the imaging equipment to the workstations and the health care system. Radiology informatics is the field that manages medical imaging IT. This involves the acquisition, storage, retrieval, and use of imaging information in health care to improve access and quality, which includes PACS, cloud services, and artificial intelligence. However, the electricity consumption of computing and the life cycle of various computer components expands the carbon footprint of health care. The authors provide a general framework to understand the environmental impact of clinical radiology informatics, which includes using the international Greenhouse Gas Protocol to draft a definition of scopes of emissions pertinent to radiology informatics, as well as exploring existing tools to measure and account for these emissions. A novel standard ecolabel for radiology informatics tools, such as the Energy Star label for consumer devices or Leadership in Energy and Environmental Design certification for buildings, should be developed to promote awareness and guide radiologists and radiology informatics leaders in making environmentally conscious decisions for their clinical practice. At this critical climate juncture, the radiology community has a unique and pressing obligation to consider our shared environmental responsibility in innovating clinical technology for patient care.

Economic and Environmental Costs of Cloud Technologies for Medical Imaging and Radiology Artificial Intelligence.

Radiology is on the verge of a technological revolution driven by artificial intelligence (including large language models), which requires robust computing and storage capabilities, often beyond the capacity of current non-cloud-based informatics systems. The cloud presents a potential solution for radiology, and we should weigh its economic and environmental implications. Recently, cloud technologies have become a cost-effective strategy by providing necessary infrastructure while reducing expenditures associated with hardware ownership, maintenance, and upgrades. Simultaneously, given the optimized energy consumption in modern cloud data centers, this transition is expected to reduce the environmental footprint of radiologic operations. The path to cloud integration comes with its own challenges, and radiology informatics leaders must consider elements such as cloud architectural choices, pricing, data security, uptime service agreements, user training and support, and broader interoperability. With the increasing importance of data-driven tools in radiology, understanding and navigating the cloud landscape will be essential for the future of radiology and its various stakeholders.

Children Are Not Small Adults: Addressing Limited Generalizability of an Adult Deep Learning CT Organ Segmentation Model to the Pediatric Population.

Deep learning (DL) tools developed on adult data sets may not generalize well to pediatric patients, posing potential safety risks. We evaluated the performance of TotalSegmentator, a state-of-the-art adult-trained CT organ segmentation model, on a subset of organs in a pediatric CT dataset and explored optimization strategies to improve pediatric segmentation performance. TotalSegmentator was retrospectively evaluated on abdominal CT scans from an external adult dataset (n = 300) and an external pediatric data set (n = 359). Generalizability was quantified by comparing Dice scores between adult and pediatric external data sets using Mann-Whitney U tests. Two DL optimization approaches were then evaluated: (1) 3D nnU-Net model trained on only pediatric data, and (2) an adult nnU-Net model fine-tuned on the pediatric cases. Our results show TotalSegmentator had significantly lower overall mean Dice scores on pediatric vs. adult CT scans (0.73 vs. 0.81, P < .001) demonstrating limited generalizability to pediatric CT scans. Stratified by organ, there was lower mean pediatric Dice score for four organs (P < .001, all): right and left adrenal glands (right adrenal, 0.41 [0.39-0.43] vs. 0.69 [0.66-0.71]; left adrenal, 0.35 [0.32-0.37] vs. 0.68 [0.65-0.71]); duodenum (0.47 [0.45-0.49] vs. 0.67 [0.64-0.69]); and pancreas (0.73 [0.72-0.74] vs. 0.79 [0.77-0.81]). Performance on pediatric CT scans improved by developing pediatric-specific models and fine-tuning an adult-trained model on pediatric images where both methods significantly improved segmentation accuracy over TotalSegmentator for all organs, especially for smaller anatomical structures (e.g., > 0.2 higher mean Dice for adrenal glands; P < .001).

Imaging the acute complications of gender-affirming surgeries: a primer for radiologists in the emergency setting.

Gender-affirming surgery (GAS) is increasingly being performed. GAS is tailored to the patient leading to a diverse spectrum of radiologic post-operative findings. Radiologists who are unfamiliar with expected anatomic alterations after GAS may misdiagnose important complications leading to adverse patient outcomes. This collaborative multi-institutional review aims to: Describe relevant embryology and native anatomy. Describe relevant Gender-Affirming Surgery (GAS) techniques and expected neo-anatomy with associated complications, including common terminology. Review expected imaging appearance of neo-anatomy/postoperative findings. Review multi-modality [ultrasound, plain film, retrograde urethrogram, computed tomography] emergent imaging findings. Understand unique patient evaluation and imaging protocol considerations in the GAS population. Discuss pearls and pitfalls of imaging in the acute post-GAS setting.

Building Diversity, Equity, and Inclusion Within Radiology Artificial Intelligence: Representation Matters, From Data to the Workforce.

Diversity, equity, and inclusion (DEI) is both a critical ingredient and moral imperative in shaping the future of radiology artificial intelligence (AI) for improved patient care, from design to deployment. At the design level: Potential biases and discrimination within data sets results in inaccurate radiology AI models, and there is an urgent need to purposefully embed DEI principles throughout the AI development and implementation process. At the deployment level: Diverse representation in radiology AI leadership, research, and career development is necessary to avoid worsening structural and historical health inequities. To create an inclusive and equitable AI-enabled future in healthcare, a DEI radiology AI leadership training program may be needed to cultivate a diverse and sustainable pipeline of leaders in the field.

Developing a Comprehensive Resident-driven Research Training Pathway: A Chief Resident's Perspective.

Wide variation exists in research training, experience, opportunities, and exposure across various radiology residency training programs, ranging from having a dedicated research track to no exposure to hypothesis driven projects. Studies conducted at different residency training programs with varied resources and National Institutes of Health funding have shown that resident-driven research initiatives and mentorship programs have the potential to improve research experience during residency training, engage more medical students in research, increase departmental peer-reviewed publications and increase peer-reviewed publications of early-career faculty physicians. In an attempt to standardize the research training during radiology residency, we propose a standardized resident-led program which institutions may adapt, as well as resources that the American Alliance of Academic Chief Residents in Radiology (A3CR2) might compile in collaboration with other national organizations to improve trainee's research experience during their radiology residency training.

Repeat Annual Lung Cancer Screening After Baseline Screening Among Screen-Negative Individuals: No-Cost Coverage Is Not Enough.

PURPOSE: Adherence to lung cancer screening (LCS) is central to effective screening. The authors evaluated the likelihood of repeat annual LCS in a national commercially insured population and associations with individual characteristics, insurance characteristics, and annual out-of-pocket cost (OOPC) burden. METHODS: Using claims data from an employer-insured population (Clinformatics), individuals 55 to 80 years of age undergoing LCS between January 1, 2015, to September 30, 2019, with "negative" LCS were included. Repeat LCS was defined as low-dose chest CT occurring 10 to 15 months after the preceding LCS. Analysis was conducted over a 6-year period. Multivariable logistic regression was used to evaluate associations between repeat LCS and individual characteristics, insurance characteristics, and total OOPC incurred by the individual in the year of the index LCS, even if unrelated to LCS. RESULTS: Of 14,943 individuals with negative LCS, 4,561 (30.5%) underwent repeat LCS. Likelihood of repeat LCS was decreased for men (adjusted odds ratio [aOR], 0.91; 95% confidence interval [CI], 0.86-0.97), Hispanic ethnicity (aOR, 0.82; 95% CI, 0.69-0.97), and indemnity insurance plans (aOR, 0.36; 95% CI, 0.25-0.53). Relative to New England, individuals in nearly all US geographic regions were less likely to undergo repeat LCS. Finally, individuals with total OOPC in the highest two quartiles were less likely to undergo repeat LCS (aOR, 0.85 [95% CI, 0.77-0.92] for OOPC >$1,069.02-$2,475.09 vs $0-$351.82; aOR, 0.75 [95% CI, 0.68-0.82] for OOPC >$2,475.09 vs $0-$351.82). CONCLUSIONS: Although federal policies facilitate LCS without cost sharing, individuals incurring high OOPC, even when unrelated to LCS, are less likely to undergo repeat LCS. Future policy design should consider the permeative burden of OOPC across the health continuum on preventive services use.

Prevalence of Diversity Statements and Disability Inclusion Among Radiology Residency Program Websites.

INTRODUCTION: Radiology has widely acknowledged the need to improve inclusion of racial, ethnic, gender, and sexual minorities, with recent discourse also underscoring the importance of disability diversity and inclusion efforts. Yet studies have shown a paucity of diversity among radiology residents, despite increasing efforts to foster diversity and inclusion. Thus, the purpose of this study is to assess radiology residency program websites' diversity statements for inclusion of race and ethnicity, gender, sexual orientation, and disability as commonly underrepresented groups. METHODS: A cross-sectional, observational study of websites of all diagnostic radiology programs in the Electronic Residency Application Service directory was conducted. Program websites that met inclusion criteria were audited for presence of a diversity statement; if the statement was specific to the residency program, radiology department, or institution; and if it was presented or linked on the program or department website. All statements were evaluated for the inclusion of four diversity categories: race or ethnicity, gender, sexual orientation, and disability. RESULTS: One hundred ninety-two radiology residencies were identified using Electronic Residency Application Service. Programs with missing or malfunctioning hyperlinks (n = 33) or required logins (n = 1) were excluded. One hundred fifty-eight websites met inclusion criteria for analysis. Two-thirds (n = 103; 65.1%) had a diversity statement within their residency, department, or institution, with only 28 (18%) having residency program-specific statements and 22 (14%) having department-specific statements. Of the websites with diversity statements, inclusion of gender diversity was most frequent (43.0%), followed by race or ethnicity (39.9%), sexual orientation (32.9%), and disability (25.3%). Race or ethnicity was most included in institution-level diversity statements. CONCLUSIONS: Less than 20% of radiology residency websites include a diversity statement, and disability is the least-included category among the diversity statements. As radiology continues to lead diversity and inclusion efforts in health care, a more comprehensive approach with equitable representation of different groups, including those with disabilities, would foster a broader sense of belonging. This comprehensive approach can help to overcome systemic barriers and bridge gaps in disability representation.

Exploring the Clinical Translation of Generative Models Like ChatGPT: Promise and Pitfalls in Radiology, From Patients to Population Health.

Generative artificial intelligence (AI) tools such as GPT-4, and the chatbot interface ChatGPT, show promise for a variety of applications in radiology and health care. However, like other AI tools, ChatGPT has limitations and potential pitfalls that must be considered before adopting it for teaching, clinical practice, and beyond. We summarize five major emerging use cases for ChatGPT and generative AI in radiology across the levels of increasing data complexity, along with pitfalls associated with each. As the use of AI in health care continues to grow, it is crucial for radiologists (and all physicians) to stay informed and ensure the safe translation of these new technologies.

Predicting Patient Demographics From Chest Radiographs With Deep Learning.

BACKGROUND: Deep learning models are increasingly informing medical decision making, for instance, in the detection of acute intracranial hemorrhage and pulmonary embolism. However, many models are trained on medical image databases that poorly represent the diversity of the patients they serve. In turn, many artificial intelligence models may not perform as well on assisting providers with important medical decisions for underrepresented populations. PURPOSE: Assessment of the ability of deep learning models to classify the self-reported gender, age, self-reported ethnicity, and insurance status of an individual patient from a given chest radiograph. METHODS: Models were trained and tested with 55,174 radiographs in the MIMIC Chest X-ray (MIMIC-CXR) database. External validation data came from two separate databases, one from CheXpert and another from a multihospital urban health care system after institutional review board approval. Macro-averaged area under the curve (AUC) values were used to evaluate performance of models. Code used for this study is open-source and available at https://github.com/ai-bias/cxr-bias, and pixelstopatients.com/models/demographics. RESULTS: Accuracy of models to predict gender was nearly perfect, with 0.999 (95% confidence interval: 0.99-0.99) AUC on held-out test data and 0.994 (0.99-0.99) and 0.997 (0.99-0.99) on external validation data. There was high accuracy to predict age and ethnicity, ranging from 0.854 (0.80-0.91) to 0.911 (0.88-0.94) AUC, and moderate accuracy to predict insurance status, with AUC ranging from 0.705 (0.60-0.81) on held-out test data to 0.675 (0.54-0.79) on external validation data. CONCLUSIONS: Deep learning models can predict the age, self-reported gender, self-reported ethnicity, and insurance status of a patient from a chest radiograph. Visualization techniques are useful to ensure deep learning models function as intended and to demonstrate anatomical regions of interest. These models can be used to ensure that training data are diverse, thereby ensuring artificial intelligence models that work on diverse populations.

Rare presentations of COVID-19: PRES-like leukoencephalopathy and carotid thrombosis.

Coronavirus disease 2019 (COVID-19) is a global pandemic, and it is increasingly important that physicians recognize and understand its atypical presentations. Neurological symptoms such as anosmia, altered mental status, headache, and myalgias may arise due to direct injury to the nervous system or by indirectly precipitating coagulopathies. We present the first COVID-19 related cases of carotid artery thrombosis and acute PRES-like leukoencephalopathy with multifocal hemorrhage.