Imaging Pearls ❯ December 2024
|
-- OR -- |
|
3D and Workflow
- ”A robust argument for AI in dentistry arises from current diagnostic deficiencies. For instance, 43% of cavities on radiographs go entirely undiagnosed, 20% of diagnosed cavities are not actually cavities, and almost 50% of periapical lesions on radiographs are overlooked.Additionally, there is a general distrust towards dentists, with only 55% trust in the accuracy of dental diagnoses, 61% of Americans surveyed saying they have switched dentists or refused treatment after a diagnosis, and 40% expressing skepticism.This distrust is further evidenced by a study from the Dental AI Council, where 136 practitioners diagnosing a single patient’s full set of radiographs found <50% diagnostic agreement, with treatment cost recommendations ranging from $300 to $36,000. A UCLA study also highlighted this issue, showing a disparity in diagnostic accuracy between students (51%) and an AI system (94%). The inconsistency in dental diagnoses is not a new issue. A 1997 Reader’s Digest article documented a journalist’s experience of receiving 50 different diagnoses from dentists across the United States, with treatment quotes ranging from $160 to $30,000. Clearly, there is a significant problem with consistency in the field. ”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - ”Numerous studies have shown that AI tools can help improve the detection and characterization of periapical pathology and periodontitis, improve operational efficiency, and provide crucial information in treatment planning. AI also enhances patient communication. Anatomical tooth segmentation can visually demonstrate how a cavity is encroaching into the dentin, and this visual aid can potentially increase patient treatment acceptance. The principal focus, however, is not on driving production, but on establishing a higher standard of truth and addressing the widespread problem of underdiagnosis in dentistry. AI-enabled dental co-pilot can also integrate multimodal data from dental radiographs, photographs, clinical parameters, procedure codes, and unstructured clinical notes to offer action-oriented opportunities and insights to all stakeholders within a practice.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - ”On a daily basis, AI-driven tools are elevating the standard of care in dentistry. Beyond improving diagnostic accuracy, AI tools can also improve the operational efficiency of the practices by automating appointment scheduling, billing, and reimbursement optimization. Virtual assistants can streamline patient triage and management by highlighting the relevant radiography and diagnostic needs for each patient. These AI tools enable the dentists to focus on delivering quality dental care during their patient facing time.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - “As AI becomes increasingly prominent in radiology, we should look to other allied fields, such as dentistry, and learn from their successes. First, AI algorithms have the ability to interpret mountains of data and perform certain tasks which exceed human abilities. For example, AI can equal or in many cases outperform radiologists in the early detection of certain cancers, such as lung and pancreatic cancer, and is increasingly used in breast screenings. Further, AI has a promising ability to utilize radiomic features (quantitative attributes extracted from medical images into mineable high-dimensional data) to predict disease progression and treatment outcomes. Working in conjunction with radiologists, there is a strong opportunity for AI to improve diagnostic accuracy and speed of treatment intervention.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - “Implementation of AI into radiological practice not only has the potential to improve the rate and accuracy of image analysis, but also to perform a variety of nondiagnostic duties. Evidenced by the integration of AI with patient management systems in dentistry, there is a tremendous opportunity to enhance operational efficiency in radiology departments, although admittedly through the much more complex milieu of the electronic medical record and the radiology information system. AI can optimize scheduling, machine utilization, and patient flow, reducing wait times and improving the overall radiological operation. Further, AI’s automation of routine tasks, such as measuring tumor sizes or identifying anatomical landmarks, alleviates the monotony and workload of radiologists, allowing them to concentrate on more complex diagnostic tasks.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) 
- “My journey with ADHD and dyslexia has shaped my leadership style and contributed to my success. Although ADHD is often perceived as a hindrance,my recent accomplishments attest to the potential of producing optimal outcomes with minimal effort. Moreover, my aptitude for problem solving has guided my leadership approach. Dyslexic thinking, in which I process information in unique ways, has empowered me to perceive solutions and possibilities that might elude others. This has been particularly helpful when confronting challenging decisions, such as weighing the pros and cons of a potential new service offering or making critical personnel choices. Creative thinking has fostered my bank of resilience and permeates my leadership style.”
Neurodiversity and Leadership
Heldfond D, Fishman EK, Chu LC, Rizk RC, Rowe SP.
J Am Coll Radiol. 2024 - “People with dyslexia are thought to excel in certain areas to compensate for weaknesses in other areas. The greatest gift of my dyslexia is keen perceptiveness. I am acutely attuned to my employees’ needs, our board’s demands, and the evolving requirements of our business. This perceptiveness also translates into persuasive and charismatic qualities. I understand how to build trust, rally individuals behind a vision, and inspire them toward a shared mission. Above all, my journey with dyslexia and ADHD has bestowed on me a serene confidence, which has been one key to my success.”
Neurodiversity and Leadership
Heldfond D, Fishman EK, Chu LC, Rizk RC, Rowe SP.
J Am Coll Radiol. 2024 - “Furthermore, the resilience and adaptability demonstrated by neurodiverse individuals are invaluable qualities in radiology and perhaps all medical fields. Rapidly evolving imaging technologies and diagnostic methods demand that radiologists remain flexible and prepared to adapt to handle complex diagnostic challenges and effectively incorporate new advancements into practice.”
Neurodiversity and Leadership
Heldfond D, Fishman EK, Chu LC, Rizk RC, Rowe SP.
J Am Coll Radiol. 2024 - “Perhaps the most important lesson is the value of perceptive and confident leadership. Such leadership recognizes and encourages the unique strengths and perspectives of all team members, leading to improved dynamics, efficiency, and patient care. Furthermore, as attending physicians, demonstrating confidence encourages positive work environments; cultivates respect as well as trust among residents, technologists, and other medical staff; and enhances team efficiency and morale, particularly important in high-pressure situations, such as complex cases. Additionally, perceptive leadership will allow us to be aware of neurodivergent members on our own teams. Certain behaviors may be apparent in training sessions or team collaborations, and it is important to try to identify and nurture their unique and valuable strengths.”
Neurodiversity and Leadership
Heldfond D, Fishman EK, Chu LC, Rizk RC, Rowe SP.
J Am Coll Radiol. 2024 
- “I love the word flavor—the essential character of something or someone. We say flavor is everything. Of course, it is what you taste in ice cream, but it is so much more. Flavor is character—it is what others say about you, what you stand for, the choices you make, what people remember about you, and how you make people feel. It is the magic we give to the world every day. It is who you are when no one is looking. It is about growing, questioning, and challenging yourself. It is what makes you unique.”
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - “In a similar vein, belonging is something we truly want to create. We believe that what you say and the words that you choose really matter. In a company, the first thing people typically think about is belonging. I want people to understand that we stem from the idea of starting small and working hard daily to build something alongside the community. Our mission is to be a community of people devoted to making better ice creams and uniting people—behind the scenes and with our products. We love ice cream—but we love service even more. We believe kindness and service are gifts that should be given freely to the world. And our four values are more valuable than any amount of money. Often in business, the focus becomes profiting, and I can say that the following four principles were much more important to me than that.”
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - 1. LIVE THE GOLDEN RULE Most challenges in life—and ice cream!—can be solved simply by treating people the way you want to be treated. The golden rule is so much more than that because before you can accomplish this, you have to have standards. You have to know how you want to be treated. You have to know how it feels to be treated well and how it feels to be treated poorly. But you have to be very careful because not everybody wants to be treated how you want to be treated, so it is important to be sensitive to that, as well. Essentially, this is about empathy, which is truly a superpower, while not assuming how others feel. I think it is a good starting point to ask yourself, how do I want to feel? How do I want to be treated? And then, how can I make other people feel that way too?
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - 2. PUT YOUR NAME ON IT Putting my name on my company turned out to be the most important thing that I have ever done. I remember the first 2 weeks into business, I was doing a massive load of dishes around 1:00 AM, after making and serving ice cream all day. I turned around and my feet were swollen. I was so tired. At any other ice cream shop, I would have just taken off my apron and gone home. I turned around and took off my apron, looked up, and there was my name on a sign made from paper blowing in the wind. I realized that I could never do that again. I could never just go home. I had to make sure that everything represented my standards. Your name is more than what is on your driver’s license—it is everything people know or believe about you. Only you have the power to define it. Have high standards and own your work. Earning the trust of those around you is important. You are not defined by 1 day’s work, but by the cumulative effect of relentless learning, doing, and improving. Be better today than yesterday, and be better tomorrow than you are today.
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - 3. ACT WITH PURPOSE “I have a chart that reads, “Trust yourself, do it, ask what’s next, and keep going.” I like the idea of think, do, repeat. In addition, be a self-starter— look for ways to create value and act. Creativity, hospitality, and quality all require initiative. Chances are you already know what to do. Trust your gut and do it. Right now! Move. Do not wait around for further instruction; act with urgency and anticipate what is next.”
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - 4. MAKE PEOPLE FEEL LOVED “Treating every single person equally— behind the counter and over it—is not just an expectation, it is a way of life. We have always been a place where many kinds of people come together. As a community, our differences are our strengths, and when each of us gives generously of ourselves, we are an unstoppable force for good.”
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - “In radiology, as in all fields of medicine, strong leadership is crucial, particularly as it plays a key role in shaping the next generation of radiologists and addressing the significant issue of burnout in the field. With nearly half of all radiologists experiencing burnout, and 39% of that largely attributed to perceived disrespect from peers and supervisors, it is imperative for our leaders to create an environment in which mutual respect and appreciation are standard. Perhaps we can adopt the golden rule—treating others as you would like to be treated—as not just a moral guideline but a professional necessity that should be present in every interaction within the department, whether with our patients, peers, or the numerous nonphysician staff that support our mission.”
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 - “Leaders must make it clear that the contributions of new hires are critical for delivering high-quality patient care and advancing medical knowledge. We should aim to offer comprehensive mentorship programs, continual feedback, and opportunities for new staff to engage with their colleagues and the broader medical community. Those initiatives should set clear expectations and demonstrate the collective impact of their work, ensuring that our leadership is focused on committing to a supportive and respectful workplace, which is vital for staff well-being and optimal patient outcomes.”
The Scents, Sense, and Cents in Jeni's Splendid Ice Creams: Implications for Radiology.
Britton J, Fishman EK, Rowe SP, Chu LC, Rizk RC.
Journal of the American College of Radiology Volume 21, Issue 11, November 2024, Pages 1835-1836 
- “We saw an opportunity to build tech for older adults and examined what was available and what we could do differently. Accessibility was clearly important, with any interface needing to be rendered user friendly. Entertainment, or consumer resonance, also emerged as an area of focus. The third opportunity for improvement we identified was the social impact of presenting the notion of health span and how we think about giving someone the power and tools to increase their health span. These three considerations were the rallying points as we built Bold.”
New Old Age Meets the Same Old Ageism: Leveraging Technology to Promote Healthier Aging
Amanda Rees, BSE, Elliot K. Fishman, MD, Linda C. Chu, MD, Steven P. Rowe, MD, PhD, Ryan C. Rizk, MS
Journal of the American College of Radiology, Volume 21, Issue 11, 1830 - 1831 - “The platform is a web-based application that can adapt dynamic programming for members with diverse experiences. For example, if two individuals want to improve their balance, and one has a history of falls, is fearful of falling, and doesn’t exercise regularly, while the other isn’t fearful but has arthritis and osteoporosis, both will receive recommendations rooted in evidence and based on building balance and strength, but the specifics of the programs will look different for each person. Ultimately, the way they engage will be adapted to what they share about themselves.”
New Old Age Meets the Same Old Ageism: Leveraging Technology to Promote Healthier Aging
Amanda Rees, BSE, Elliot K. Fishman, MD, Linda C. Chu, MD, Steven P. Rowe, MD, PhD, Ryan C. Rizk, MS
Journal of the American College of Radiology, Volume 21, Issue 11, 1830 - 1831 - “The prevalence of ageism in the field of radiology, as in many other professions, often overshadows the significant benefits older radiologists bring to their practices. It’s crucial to recognize and value these benefits, shifting the focus from the negative perceptions of aging to a more appreciative understanding of what older age contributes. Older radiologists often have a deeper and more nuanced understanding of medical imaging. Their years in the field allow them to recognize patterns and subtleties in images that less experienced radiologists and artificial intelligence might miss. That expertise is not only a result of their long-term exposure to a large variety of cases but also stems from their ability to integrate evolving technologies and methodologies over time.”
New Old Age Meets the Same Old Ageism: Leveraging Technology to Promote Healthier Aging
Amanda Rees, BSE, Elliot K. Fishman, MD, Linda C. Chu, MD, Steven P. Rowe, MD, PhD, Ryan C. Rizk, MS
Journal of the American College of Radiology, Volume 21, Issue 11, 1830 - 1831 - “Last, emotional well-being and social connections are important aspects of aging well. Radiologists should be encouraged to engage in social activities, peer mentoring, and community involvement. This enhances their quality of life and provides them with a sense of purpose and belonging, which are key to a positive outlook on aging. Redefining aging in radiology means appreciating the unique advantages that come with experience while also emphasizing the importance of holistic well-being. By doing so, older radiologists can continue to provide exceptional patient care, serve as mentors for the next generation, and lead fulfilling, active professional lives. This approach benefits not only the individual radiologists but also the medical community and patients we serve.”
New Old Age Meets the Same Old Ageism: Leveraging Technology to Promote Healthier Aging
Amanda Rees, BSE, Elliot K. Fishman, MD, Linda C. Chu, MD, Steven P. Rowe, MD, PhD, Ryan C. Rizk, MS
Journal of the American College of Radiology, Volume 21, Issue 11, 1830 - 1831
Deep Learning
- Radiology till the late 1990’s
- Film (8x10, 10x14, 14x17)
- Film jackets (one copy of a study)
- Film Processors and Film Alternators
- Kodak was the king of imaging
- Radiology and Pathology Change is Similar
- Film to digital display (Radiology)
- Glass slides to digital display (Pathology)
Both were held back in part by the attitude of the changes are not acceptable - “Histopathology image evaluation is indispensable for cancer diagnoses and subtype classification. Standard artificial intelligence methods for histopathology image analyses have focused on optimizing specialized models for each diagnostic task. Although such methods have achieved some success, they often have limited generalizability to images generated by different digitization protocols or samples collected from different populations3. Here, to address this challenge, we devised the Clinical Histopathology Imaging Evaluation Foundation (CHIEF) model, a general purpose weakly supervised machine learning framework to extract pathology imaging features for systematic cancer evaluation. CHIEF leverages two complementary pretraining methods to extract diverse pathology representations: unsupervised pretraining for tile-level feature identification and weakly supervised pretraining for whole-slide pattern recognition.”
A pathology foundation model for cancer diagnosis and prognosis prediction.
Wang X, Zhao J, Marostica E, et al.
Nature. 2024 Sep 4. doi: 10.1038/s41586-024-07894-z. Epub ahead of print. PMID: 39232164. - “We developed CHIEF using 60,530 whole-slide images spanning 19 anatomical sites. Through pretraining on 44 terabytes of high resolution pathology imaging datasets, CHIEF extracted microscopic representations useful for cancer cell detection, tumour origin identification, molecular profile characterization and prognostic prediction. We successfully validated CHIEF using whole-slide images from 32 independent slide sets collected from 24 hospitals and cohorts internationally. Overall, CHIEF outperformed the state-of-the-art deep learning methods by up to 36.1%, showing its ability to address domain shifts observed in samples from diverse populations and processed by different slide preparation methods. CHIEF provides a generalizable foundation for efficient digital pathology evaluation for patients with cancer.”
A pathology foundation model for cancer diagnosis and prognosis prediction.
Wang X, Zhao J, Marostica E, et al.
Nature. 2024 Sep 4. doi: 10.1038/s41586-024-07894-z. Epub ahead of print. PMID: 39232164. - ”We established the CHIEF model, a general-purpose machine learning framework for weakly supervised histopathological image analyses. Unlike commonly used self-supervised feature extractors, CHIEF leveraged two types of pretraining procedure: unsupervised pretraining on 15 million unlabelled tile images and weakly supervised pretraining on more than 60,000 WSIs. Tile-level unsupervised pretraining established a general feature extractor for haematoxylin–eosin-stained histopathological images collected from heterogeneous publicly available databases, which captured diverse manifestations of microscopic cellular morphologies. Subsequent WSI-level weakly supervised pretraining constructed a general-purpose model by characterizing thesimilarities and differences between cancer types.”
A pathology foundation model for cancer diagnosis and prognosis prediction.
Wang X, Zhao J, Marostica E, et al.
Nature. 2024 Sep 4. doi: 10.1038/s41586-024-07894-z. Epub ahead of print. PMID: 39232164. - CHIEF consistently attained superior performance in a variety of cancer identification tasks using either biopsy or surgical resection slides CHIEF achieved a macro-average area under the receiver operating characteristic curve (AUROC) of 0.9397 across 15 datasets representing 11 cancer types, which is approximately 10% higher than that attained by DSMIL (a macro-average AUROC of 0.8409), 12% higher than that of ABMIL (a macro-average AUROC of 0.8233) and 14% higher than that of CLAM (a macro-average AUROC of 0.8016). In all five biopsy datasets collected from independent cohorts, CHIEF possessed AUROCs of greater than 0.96 across several cancer types, including oesophagus (CUCH-Eso), stomach (CUCH-Sto), colon (CUCH-Colon) and prostate (Diagset-B and CUCH-Pros). On independent validation with seven surgical resection slide sets spanning five cancer types (that is, colon (Dataset-PT), breast (DROID-Breast), endometrium (SMCH-Endo and CPTAC-uterine corpus endometrial carcinoma (UCEC)), lung (CPTAC-lung squamous cell carcinoma (LUSC)) and cervix (SMCH-Cervix and TissueNet)), CHIEF attained AUROCs greater than 0.90
A pathology foundation model for cancer diagnosis and prognosis prediction.
Wang X, Zhao J, Marostica E, et al.
Nature. 2024 Sep 4. doi: 10.1038/s41586-024-07894-z. Epub ahead of print. PMID: 39232164. - “The CHIEF framework successfully characterized tumour origins, predicted clinically important genomic profiles, and stratified patients into longer-term survival and shorter-term survival groups. Furthermore, our approach established a general pathology feature extractor capable of a wide range of prediction tasks even with small sample sizes. Our results showed that CHIEF is highly adaptable to diverse pathology samplesobtained from several centres, digitized by various scanners, and obtained from different clinical procedures (that is, biopsy and surgicalresection). This new framework substantially enhanced model generalizability, a critical barrier to the clinical penetrance of conventional computational pathology models.”
A pathology foundation model for cancer diagnosis and prognosis prediction.
Wang X, Zhao J, Marostica E, et al.
Nature. 2024 Sep 4. doi: 10.1038/s41586-024-07894-z. Epub ahead of print. PMID: 39232164. - “In conclusion, CHIEF is a foundation model useful for a wide range of pathology evaluation tasks across several cancer types. We have demonstrated the generalizability of this foundation model across several clinical applications using samples collected from 24 hospitals and patient cohorts worldwide. CHIEF required minimal image annotations and extracted detailed quantitative features from WSIs, which enabled systematic analyses of the relationships among morphological patterns, molecular aberrations and important clinical outcomes. Accurate, robust and rapid pathology sample assessment provided by CHIEF will contribute to the development of personalized cancer management.”
A pathology foundation model for cancer diagnosis and prognosis prediction.
Wang X, Zhao J, Marostica E, et al.
Nature. 2024 Sep 4. doi: 10.1038/s41586-024-07894-z. Epub ahead of print. PMID: 39232164. - Application of AI to an array of diagnostic tasks using WSIs has rapidly expanded in recent years . Successes in AI for digital pathology can be found for many disease types, but particularly in examples applied to cancer. An important early study in 2017 by Bejnordi et al. described 32 AI models developed for breast cancer detection in lymph nodes through the CAMELYON16 grand challenge. The best model achieved an area under the curve (AUC) of 0.994 (95% CI 0.983–0.999), demonstrating similar performance to the human in this controlled environment.
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; - Following recent prominent discoveries in deep learning techniques, wider artificial intelligence (AI) applications have emerged for many sectors, including in healthcare. Pathology AI is of broad importance in areas across medicine, with implications not only in diagnostics, but in cancer research, clinical trials and AI-enabled therapeutic targeting4. Access to digital pathology through scanning of whole slide images (WSIs) has facilitated greater interest in AI that can be applied to these images. WSIs are created by scanning glass microscope slides to produce a high resolution digital image, which is later reviewed by a pathologist to determine the diagnosis. Opportunities for pathologists have arisen from this technology, including remote and flexible working, obtaining second opinions, easier collaboration and training, and applications in research, such as AI..
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; - ”Despite the many developments in pathology AI, examples of routine clinical use of these technologies remain rare and there are concerns around the performance, evidence quality and risk of bias for medical AI studies in general. Although, in the face of an increasing pathology workforce crisis, the prospect of tools that can assist and automate tasks is appealing. Challenging workflows and long waiting lists mean that substantial patient benefit could be realised if AI was successfully harnessed to assist in the pathology laboratory.”
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; - “AI has been extensively promoted as a useful tool that will transform medicine, with examples of innovation in clinical imaging, electronic health records (EHR), clinical decision making, genomics, wearables, drug development and robotics. The potential of AI in digital pathology has been identified by many groups, with discoveries frequently emerging and attracting considerable interest. Tools have not only been developed for diagnosis and prognostication, but also for predicting treatment response and genetic mutations from the H&E image alone. Various models have now received regulatory approval for applications in pathology, with some examples being trialled in clinical settings.”
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; - Of studies of other disease types included in the meta-analysis, AI models in liver cancer, lymphoma, melanoma, pancreatic cancer, brain cancer, lung cancer and rhabdomyosarcoma all demonstrated a high sensitivity and specificity. This emphasises the breadth of potential diagnostic tools for clinical applications with a high diagnostic accuracy in digital pathology. T
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; 
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038s41746-024-01106-8. PMID: 38704465;- “Performing an external validation on data from an alternative source to that on which an AI model was trained, providing details on the process for case selection and using large, diverse datasets would help to reduce the risk of bias of these studies. Overall, better quality study design, transparency, reporting quality and addressing substantial areas of bias is needed to improve the evidence quality in pathology AI and to therefore harness the benefits of AI for patients and clinicians.”
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; - “There are many promising applications for AI models in WSIs to assist the pathologist. This systematic review has outlined a high diagnostic accuracy for AI across multiple disease types. A larger body of evidence is available for gastrointestinal pathology, urological pathology and breast pathology. Many other disease areas are underrepresented and should be explored further in future. To improve the quality of future studies, reporting of sensitivity, specificity and raw data (true positives, false positives, false negatives, true negatives) for pathology AI models would help with transparency in comparing diagnostic performance between studies.”
Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy.
McGenity C, Clarke EL, Jennings C, et al..
NPJ Digit Med. 2024 May 4;7(1):114. doi: 10.1038/s41746-024-01106-8. PMID: 38704465; - Digital pathology, also referred to as whole slide imaging, is a sub-field of pathology, in which tissue specimens are digitized with a scanner before being examined. Biopsy or sample collection techniques, laboratory workflow and final reporting with treatment decisions remain largely unchanged, but the slide review phase of the pathology process happens in a digital way using a display and viewing software, in addition to or in combination with a microscope.
- Digital pathology enables the creation of digital images of glass slides that can be securely shared electronically with other pathologists to view, reducing transportation needs and speeding testing and results reporting. It also has the advantage of extending access to expert consults to geographic areas where pathologists are in short supply, such as in parts of rural America and internationally. It may also help alleviate workforce pressures due to a shortage of pathologists and histotechnologists, the skilled laboratory professionals who prepare tissue slides. Digital pathology enables the creation of digital images of glass slides that can be securely shared electronically with other pathologists to view, reducing transportation needs and speeding testing and results reporting. It also has the advantage of extending access to expert consults to geographic areas where pathologists are in short supply, such as in parts of rural America and internationally. It may also help alleviate workforce pressures due to a shortage of pathologists and histotechnologists, the skilled laboratory professionals who prepare tissue slides.
- The analysis of histopathology images with artificial intelligence aims to enable clinical decision support systems and precision medicine. The success of such applications depends on the ability to model the diverse patterns observed in pathology images. To this end, we present Virchow, the largest foundation model for computational pathology to date. In addition to the evaluation of biomarker prediction and cell identification, we demonstrate that a large foundation model enables pan-cancer detection, achieving 0.95 specimen-level area under the (receiver operating characteristic) curve across nine common and seven rare cancers. Furthermore, we show that with less training data, the pan-cancer detector built on Virchow can achieve similar performance to tissue-specific clinical-grade models in production and outperform them on some rare variants of cancer. Virchow’s performance gains highlight the value of a foundation model and open possibilities for many high-impact applications with limited amounts of labeled training data.
A foundation model for clinical-grade computational pathology and rare cancers detection
Eugene Vorontsov, Alican Bozkurt, Adam Casson et al.
Nat Med. 2024 Oct;30(10):2924-2935. - Computational pathology applies artificial intelligence (AI) to digitized WSIs to support the diagnosis, characterization and understanding of disease . Initial work has focused on clinical decision support tools to enhance current workflows, and in 2021 the first Food and Drug Administration-approved AI pathology system was launched. However, given the incredible gains in performance of computer vision, a subfield of AI focused on images, more recent studies attempt to unlock new insights from routine WSIs and reveal undiscovered outcomes such as prognosis and therapeutic response. If successful, such efforts would enhance the utility of hematoxylin and eosin (H&E)-stained WSIs and reduce reliance on specialized and often expensive immunohistochemistry (IHC) or genomic testing.
A foundation model for clinical-grade computational pathology and rare cancers detection
Eugene Vorontsov, Alican Bozkurt, Adam Casson et al.
Nat Med. 2024 Oct;30(10):2924-2935. - A major factor in the performance gains of computer vision models has been the creation of large-scale deep neural networks, termed foundation models. Foundation models are trained on enormous . datasets—orders of magnitude greater than any used historically for computational pathology—using a family of algorithms, referred to as self-supervised learning, which do not require curated labels. Foundation models generate data representations, called embeddings, that can generalize well to diverse predictive tasks. This offers a distinct advantage over current diagnostic-specific methods in computational pathology, which, limited to a subset of pathology images, are less likely to reflect the full spectrum of variations in tissue morphology and laboratory preparations necessary for adequate generalization in practice. The value of generalization from large datasets is even greater for applications with inadequate quantities of data to develop bespoke models, as is the case for the detection of uncommon or rare tumor types, as well as for less common diagnostic tasks such as the prediction of specific genomic alterations, clinical outcomes and therapeutic response
A foundation model for clinical-grade computational pathology and rare cancers detection
Eugene Vorontsov, Alican Bozkurt, Adam Casson et al.
Nat Med. 2024 Oct;30(10):2924-2935. - “Here, we present a million-image-scale pathology foundation model, Virchow, named in honor of Rudolf Virchow, who is regarded as the father of modern pathology and proposed the first theory of cellular pathology45. Virchow is trained on data from approximately 100,000 patients corresponding to approximately 1.5 million H&E stained WSIs acquired from Memorial Sloan Kettering Cancer Center (MSKCC), which is 4–10× more WSIs than in prior training datasets in pathology. The training data are composed of cancerous and benign tissues, collected via biopsy (63%) and resection (37%), from 17 high-level tissues.”
A foundation model for clinical-grade computational pathology and rare cancers detection
Eugene Vorontsov, Alican Bozkurt, Adam Casson et al.
Nat Med. 2024 Oct;30(10):2924-2935. - A key aim of our work was to develop a single model to detect cancer, including rare cancers (defined by the National Cancer Institute (NCI) as cancers with an annual incidence in the United States of fewer than 15 people per 100,000 (ref. 46)), across various tissues. The pan-cancer detection model infers the presence of cancer using Virchow embeddings as input. For evaluation, slides from MSKCC and slides submitted for consultation to MSKCC from numerous external sites globally are used. Stratified performance across nine common and seven rare cancer types is reported. Embeddings generated by Virchow, UNI41, Phikon37 and CTransPath35 are evaluated. Pan-cancer aggregators are trained using specimen-level labels, maintaining the same training protocol for all embeddings .
A foundation model for clinical-grade computational pathology and rare cancers detection
Eugene Vorontsov, Alican Bozkurt, Adam Casson et al.
Nat Med. 2024 Oct;30(10):2924-2935. - “Recent advances in computational pathology have been supported by increased dataset scale and reduced reliance on labels. Using multiple-instance learning with labels at the level of groups of slides has enabled clinically relevant diagnostics by scaling to training datasets on the order of 10,000 WSIs. These earlier works typically initialized the model’s embedding parameters using pretrained model weights, often those trained on ImageNet in a supervised setting. This process, called transfer learning, was motivated by the observation that model performance critically depends on the model’s ability to capture image features. In-domain transfer learning was not possible given the limited availability of labeled pathology datasets. Now self-supervised learning is enabling in-domain transfer by removing the label requirement, driving a second wave of scaling to tens of thousands of WSIs to inform image representation. Virchow marks a major increase in training data scale to 1.5 million WSIs—a volume of data that is over 3,000 times the size of ImageNet as measured by the total number of pixels. This large scale of data in turn motivates large models that can capture the diversity of image features in WSIs. In this work, we have demonstrated that this approach can form the foundation for clinical-grade models in cancer pathology.”
A foundation model for clinical-grade computational pathology and rare cancers detection
Eugene Vorontsov, Alican Bozkurt, Adam Casson et al.
Nat Med. 2024 Oct;30(10):2924-2935. - There has been a sharp rise in the development and application of AI tools, including image-based algorithms for the use in pathology service, and it is expected to dominate the field of pathology in the coming years. The deployment of computational pathology and application of pathology-related AI tools can be considered a paradigm shift that will change the way pathology services are managed and make them not only more efficient but also capable of meeting the needs of this era of precision medicine. Development of pathology-based AI tools needs input from a multidisciplinary team in which pathologists and users should have great input to improve the adoption of these technology-driven applications.
Current and future applications of artificial intelligence in pathology: a clinical perspective
Rakha EA, Toss M, Shiino S, et al.
J Clin Pathol 2021;74:409–414. - A combination of AI and pathologists can yield results that are more accurate, consistent, timely and useful beyond a human’s ability. AI can provide analytical tools to streamline the complex, multistep pathology case life cycle in pathology laboratories, from accession to archiving. This can provide not only workflow automation but also analytics dashboard and data repository that can improve efficiency by self-learning from previous experience and help understand laboratory productivity, quality and efficiency, in addition to helping to allocate future resources to areas in more need. AI can also improve streamlining the whole process by aligning the laboratory technical components of the case pathway with the pathologists reporting components.
Current and future applications of artificial intelligence in pathology: a clinical perspective
Rakha EA, Toss M, Shiino S, et al.
J Clin Pathol 2021;74:409–414. - “AI can also improve streamlining the whole process by aligning the laboratory technical components of the case pathway with the pathologists reporting components. Improving the efficiency of pathology service workflow, trainee and junior pathologists reporting, timely reporting by pathologists, costefficient diagnostic, prognostic/predictive algorithms and production of multidimensional output of pathology reports, and combining with image and genomic/genetic data are some of the expected benefits of AI technology application in routine practice.”
Current and future applications of artificial intelligence in pathology: a clinical perspective
Rakha EA, Toss M, Shiino S, et al.
J Clin Pathol 2021;74:409–414. - AI applications will also lead to an advanced diagnostic, enabling researchers and clinical teams to share knowledge and use computational algorithms to assess and contribute valuable insights that can ultimately lead to a more informed and detailed pathology diagnosis. This integration will help advance the future of precision oncology and can result in personalised care plans.
Current and future applications of artificial intelligence in pathology: a clinical perspective
Rakha EA, Toss M, Shiino S, et al.
J Clin Pathol 2021;74:409–414. - ► Using Computational Pathology and Artificial Intelligence in clinical histopathology service is expected to extend in the near future.
► Understanding the current limitation and challenges of AI applications will help to improve its performance and applicability.
► These new technologies are aiming to complement the human resources rather than replacing them.
Current and future applications of artificial intelligence in pathology: a clinical perspective
Rakha EA, Toss M, Shiino S, et al.
J Clin Pathol 2021;74:409–414. - “But what AI is, and how exactly it might prove useful in pathology, is still not clear to many of us. It feels new and mysterious, triggering a lot of anxiety about what it means for our profession, our practices, and our patients. I’ve heard a lot of fear that AI may eventually even replace us.”
Donald S. Karcher, MD
President, The College of American Pathologists
Sept 2024 - ”Like many other “disruptive” technologies, AI is just another tool that will allow us to make better and more actionable diagnoses—nothing more and nothing less. It will not take our place and it will not eliminate the need for our expertise. But we should be prepared. There are loads of pathology AI tools that are on the verge of being FDA approved, and I expect they’ll start coming out by the dozens in the next few years. Image analysis tools will be quite common (there are already FDA-approved AI programs for image analysis in pathology) and so too will be products to analyze “big data” across many clinical sources, including laboratory data. We will have to know how to assess them, select the best ones for our practices, and implement them.”
Donald S. Karcher, MD
President, The College of American Pathologists
Sept 2024 - Objective.—To review use cases of GAI in CP, with a particular focus on large language models. Specific examples are provided for the applications of GAI in the subspecialties of clinical chemistry, microbiology, hematopathology, and molecular diagnostics. Additionally, the review addresses potential pitfalls of GAI paradigms
Conclusions.—GAI is a powerful tool with the potential to revolutionize health care for patients and practitioners alike. However, GAI must be implemented with much caution considering various shortcomings of the technology such as biases, hallucinations, practical challenges of implementing GAI in existing CP workflows, and end-user acceptance. Human-in-the-loop models of GAI implementation have the potential to revolutionize CP by delivering deeper, meaningful insights into patient outcomes both at an individual and population level.
Evaluating Use of Generative Artificial Intelligence in Clinical Pathology Practice
Peter McCaffrey, Ronald Jackups, Jansen Seheult et al.
Arch Pathol Lab Med. doi: 10.5858/arpa.2024-0208-RA) - “Within the field of pathology, the use of AI/machine learning has been confined to nongenerative settings and much attention has rightfully been paid to whole slide imaging classification and image segmentation tasks. This has led to more widespread digitization, which, in turn, has enabled image-based AI to access pathology workflows much in the same way that AI has advanced through the specialty of radiology.”
Evaluating Use of Generative Artificial Intelligence in Clinical Pathology Practice
Peter McCaffrey, Ronald Jackups, Jansen Seheult et al.
Arch Pathol Lab Med. doi: 10.5858/arpa.2024-0208-RA) - However, ignorance and nonparticipation in AI technologies is not an option for laboratories (and pathologists) to remain competitive in the marketplace. If we collectively maintain an open dialogue about challenges and successes while adapting this transformational technology in a collaborative manner, there will be ample room for everyone to be successful. In the famous words of Peter Drucker, “The best way to predict the future is to create it.”
Evaluating Use of Generative Artificial Intelligence in Clinical Pathology Practice
Peter McCaffrey, Ronald Jackups, Jansen Seheult et al.
Arch Pathol Lab Med. doi: 10.5858/arpa.2024-0208-RA) - BACKGROUND While previous studies of artificial intelligence (AI) have shown its potential for diagnosing diseases using imaging data, clinical implementation lags behind. AI models require training with large numbers of examples, which are only available for common diseases. In clinical reality, however, the majority of diseases are less frequent, and current AI models overlook or misclassify them. An effective, comprehensive technique is needed for the full spectrum of real-world diagnoses.
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - Methods We collected two large real-world datasets of gastrointestinal (GI) biopsies, which are prototypical of the problem. Herein, the 10 most common findings accounted for approximately 90% of cases, whereas the remaining 10% contained 56 disease entities, including many cancers. Seventeen million histological images from 5423 cases were used for training and evaluation. We propose a deep anomaly detection (AD) approach that only requires training data from common diseases to also detect all less frequent diseases.
Results Without specific training for the diseases, our best-performing model reliably detected a broad spectrum of infrequent (“anomalous”) pathologies with 95.0% (stomach) and 91.0% (colon) area under the receiver operating characteristic curve (AUROC) and was able to generalize between scanners and hospitals. Cancers were detected with 97.7% (stomach) and 96.9% (colon) AUROC. Heatmaps reliably highlighted anomalous areas and can guide pathologists during the diagnostic process.
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - “In this study, we establish the first effective clinical application of AI-based AD in histopathology and demonstrate high performance on a unique real-world collection of GI biopsies. The proposed novel AD can flag anomalous cases, facilitate case prioritization, and reduce missed diagnoses, providing critical support for pathologists. By design, it can be expected to detect any pathological alteration including rare primary or metastatic cancers in GI biopsies. To our knowledge, no other published AI tool is capable of zero-shot pan-cancer detection. AD may enhance the safety of AI models in histopathology, thereby driving AI adoption and automation in routine diagnostics and beyond.”
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - “Diagnostic pathology faces serious challenges due to a shortage of pathologists in many parts of the world and too few doctors entering the profession. Meanwhile, both the diagnostic workload and cancer burden are rising. Diagnostic procedures are increasing in complexity due to the demands of precision medicine. Studies have shown significant diagnostic errors in a range of 0.1% to 10% of cases.”
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - In this study, we hypothesize that infrequent findings in histopathological images can be detected using AI-based anomaly detection (AD). In contrast to supervised learning, AD assumes that certain data inputs are too infrequent to be sufficiently represented during model training. Instead of trying to learn insufficiently represented patterns, AD methods aim to very precisely characterize the frequent findings, which in our setting includes normal cases and common pathologies that can be learned by supervised methods. Samples deviating from the learned common characteristics are consequently deemed “anomalies.” Since only frequent findings are used for AD model training, there is no need for extensive data collection or annotation gathering of rarer instances from the tail of the disease distribution. is no need for extensive data collection or annotation gathering of rarer instances from the tail of the disease distribution.
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - We were largely successful in this task, as almost all diseases from various diagnostic groups resulted in considerably elevated anomaly scores. Importantly, malignant tumors of very different morphology and histogenesis, such as carcinomas, neuroendocrine tumors, lymphomas, metastatic melanomas, or sarcomas, were reliably assigned high anomaly scores. In fact, of all the diagnostic groups, slide-AUROCs were highest for malignancies, with 97.72% for stomach and 96.97% for colon, respectively (with the overall best-performing OE model). This is crucial, as detecting malignancy is the most consequential task in histopathological diagnostics. Infrequent benign and precancerous neoplastic changes were also reliably detected (slide-AUROC of 88.45% for stomach, 95.72% for colon). Additionally, the AD model effectively recognized inflammation of the colon (slide-AUROC of 94.42%). For stomach, most types of inflammation are frequent and therefore nonanomalous.
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - “Our AI AD can be implemented in the clinical workflow in two main ways. First, it can be used as a standalone clinical AI assistant that prescreens every stomach and colon sample, identifies and prioritizes “suspicious” cases, and creates corresponding warning labels. The heatmaps can then further guide the pathologists during their assessment. This has the potential to substantially improve diagnostic efficiency and quality while reducing missed diagnoses. Critically, the AI AD’s design ensures the reliable detection of any kind of primary or metastatic cancer in stomach and colon samples, even beyond those evaluated here. To our knowledge, no other published AI tool is capable of this in a zero-shot manner, even across other tissues. Second, the AI AD can be integrated with the supervised detection of common findings (e.g., for GI samples) .”
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11) - “In this proposed workflow, common pathologies are classified automatically by a supervised module, while AI-AD flags anomalous cases for manual review by a pathologist. Our results indicate that with the current performance, already up to a third of biopsies with frequent findings could be automatically diagnosed in this way without the risk of missing any less frequent and potentially severe diseases. This fraction can be expected to grow with future model improvements. Ultimately, only a subset of cases may require manual review, drastically reducing pathologists’ workloads and enabling largely automated and safe AI-based histopathological diagnostics.”
AI-Based Anomaly Detection for Clinical-Grade Histopathological Diagnostics
Jonas Dippel, Niklas Prenißl, Julius Hense, et al
NEJM AI 2024;1(11)
- IMPORTANCE Advances in artificial intelligence (AI) must be matched by efforts to better understand and evaluate how AI performs across health care and biomedicine as well as develop appropriate regulatory frameworks. This Special Communication reviews the history of the US Food and Drug Administration’s (FDA) regulation of AI; presents potential uses of AI in medical product development, clinical research, and clinical care; and presents concepts that merit consideration as the regulatory system adapts to AI’s unique challenges.
CONCLUSIONS AND RELEVANCE Strong oversight by the FDA protects the long-term success of industries by focusing on evaluation to advance regulated technologies that improve health. The FDA will continue to play a central role in ensuring safe, effective, and trustworthy AI tools to improve the lives of patients and clinicians alike. However, all involved entities will need to attend to AI with the rigor this transformative technology merits.
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024. - OBSERVATIONS The FDA has authorized almost 1000 AI-enabled medical devices and has received hundreds of regulatory submissions for drugs that used AI in their discovery and development. Health AI regulation needs to be coordinated across all regulated industries, the US government, and with international organizations. Regulators will need to advance flexible mechanisms to keep up with the pace of change in AI across biomedicine and health care. Sponsors need to be transparent about and regulators need proficiency in evaluating the use of AI in premarket development. A life cycle management approach incorporating recurrent local postmarket performance monitoring should be central to health AI development. Special mechanisms to evaluate large language models and their uses are needed. Approaches are necessary to balance the needs of the entire spectrum of health ecosystem interests, from large firms to start-ups. The evaluation and regulatory system will need to focus on patient health outcomes to balance the use of AI for financial optimization for developers, payers, and health systems.
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024. 
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024.- The FDA’s first approval of a partially AI-enabled medical device took place in 1995,when the FDA approved PAPNET, a software that used neural networks to prevent misdiagnosis of cervical cancer in women undergoing Papanicolaou tests. Although PAPNET was shown to be more accurate than human pathologists, it was not adopted in clinical practice due to inadequate cost-effectiveness. Since then, the FDA has authorized approximately 1000 AI-enabled medical devices, with their most common use being in radiology, followed by cardiology.
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024. 
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024.- Although the FDA does not regulate the practice of medicine, it has a strong mission to both advance public health and biomedical innovation. Therefore, there is concern that a disproportionate focus of AI applications on financial return on investment could harm patient outcomes and reduce acceptance and trust in this technology. Many AI innovations that could benefit patients may come at the price of traditional jobs, capital structures, and revenue streams in health care. Yet too many US residents live in health care deserts, with primary care shortages even in many physician-dense areas, and AI algorithms could point to more preventive services that currently are not profitable. Furthermore, AI could significantly improve the efficiency of clinical services, there by freeing clinicians to do the one thing that ultimately no machine can: forge a human connection with the patient.
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024. - Historic advances in AI applied to biomedicine and health caremust be matched by continuous complementary efforts to better understand how AI performs in the settings in which it is deployed. This will entail a comprehensive approach reaching far beyond the FDA, spanning the consumer and health care ecosystems to keep pace with accelerating technical progress. If not, there is a risk that AI could disappoint similar to other general-purpose technologies deployed in health care settings or even create significant harm if untended models’ performance deteriorates or focuses on financial return without adequate attention to impact on clinical outcomes.
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024. - “Strong oversight by the FDA and other agencies aims to protect the long-term success of regulated products by maintaining a high grade of public trust in the regulated space. It is in the interest of the biomedical, digital, and health care industries to identify and deal with irresponsible actors and to avoid misleading hyperbole. Regulated industries, academia, and the FDA will need to develop and optimize the tools needed to assess the ongoing safety and effectiveness of AI in health care and biomedicine. The FDA will continue to play a central role with a focus on health outcomes, but all involved sectors will need to attend to AI with the care and rigor this potentially transformative technology merits.”
FDA Perspective on the Regulation of Artificial Intelligence in Health Care and Biomedicine
Haider J. Warraich, Troy Tazbaz, Robert M. Califf
JAMA. doi:10.1001/jama.2024.21451Published online October 15, 2024.
- To date, the development of machine learning (ML) and artificial intelligence(AI) in medicine has been characterized by steady progress with times of rapid growth—each with promise to improve patientoutcomes and clinical practice and, at times, reduce costs. With the expansion from traditional AI (ML, computer vision, and natural language processing) to generative AI using generative pretrained transformers (GPTs), staggering new opportunities exist to both develop insights and present them to clinicians and patients. However, to maximize the positive impact of these innovations, a framework is needed for clinicians and patients to understand AI in the context of clinical practice, including the evidence of efficacy, safety, and monitoring in real-world clinical use. We believe that the progress and adoption of ML and AI tools in medicine will be accelerated by a clinical framework for AI development and testing that links evidence generation to indication and benefit and risk and allows clinicians to immediately understand in the context of existing practice guidelines.
Translating AI for the Clinician
Manesh R. Patel, Suresh Balu, Michael J. Pencina
JAMA Published online October 15, 2024 - “To realize their full potential, current development of health AI technologies needs to focus on the clinical use case or indication thatnthe technologies aim to improve. Specifically, developers should prioritize aligning the technologies with clinical indication and use cases to maximize impact. We believe this first step is a conceptual sea change from the current development pathway, which focuses on the advanced computational techniques and available health data sources being used, with emphasis on variety ,amount, and breadth. Although this is necessary for AI algorithm and model formation, it is not sufficient. For successful adoption of AI technologies in the clinic, we must first articulate the specific problems or use cases that would benefit from the incorporation of AI. “
Translating AI for the Clinician
Manesh R. Patel, Suresh Balu, Michael J. Pencina
JAMA Published online October 15, 2024 - “Although regulatory agencies have provided some guidance,2 risk or indication-based testing and monitoring are key to rapid development and implementation of high-quality health AI technologies. Specifically, for low-risk AI technologies (those that might improve health behaviors, such as encouraging more movement or more sleep),one can imagine study designs incorporating these tools in prospective observational studies. These study designs still need representative patients and broad uptake and change measures for meaningful use. In contrast, high-risk AI technologies, such as tools to improve clinical performance during either a therapeutic (eg, percutaneous angioplasty) or diagnostic (eg, AI algorithm for detection of a dangerous cardiac rhythm) procedure, would require a randomized trial with a control group to ensure clinical evidence for use. In this way, the indication and risk of the AI technology would matchthe methodology for clinical testing and adoption.”
Translating AI for the Clinician
Manesh R. Patel, Suresh Balu, Michael J. Pencina
JAMA Published online October 15, 2024 - “The next decades of health care innovation will undoubtedly be dependent on the volume of health data generated in the daily conduct of health delivery. Coupled with the technological breakthroughs affordedby the rapid growth of AI capabilities, these health care innovations could truly revolutionize the practice of medicine as we know it. However, this potential will not be realized without a refocusing of AI technology development toward a closer alignment with the health goals that clinicians and patients understand are required to ensure widespread adoption and maximal impact to improve human health. We need clearly articulated clinical indications, well-defined risk-based clinical testing processes and evidence generation, and continuous monitoring linked to these indications. Without this type of paradigm shift, we fear that the use of AI technologies will struggle to gain sufficient trust among clinicians and patients, which in turn will limit its adoption and impact on health.”
Translating AI for the Clinician
Manesh R. Patel, Suresh Balu, Michael J. Pencina
JAMA Published online October 15, 2024 
Translating AI for the Clinician
Manesh R. Patel, Suresh Balu, Michael J. Pencina
JAMA Published online October 15, 2024
- “Integrating artificial intelligence (AI) in emergency radiology, particularly in the context of WB-CTbscans for polytraumatized patients, represents a substantial advancement in the diagnosis and subsequent management of traumatic injuries. A review of the current literature underscores the promising potential of AI to enhance the accuracy, efficiency, and predictive capabilities of radiological assessments in the emergency and trauma settings. Zhou and colleagues conducted a study on 133 patients comparing the results of rib fracture detection using initial CT and follow-up CT, revealing that AI-assisted diagnosis statistically significantly improved the overall accuracy for the detection of rib fractures.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “The innovative application of AI extends beyond these examples. AI has been employed in the quantitative analysis of traumatic hemoperitoneum,70 highlighting its ability to aid in diagnosing and managing intrabdominal bleeding. Seyam and colleagues implemented an AI-based detection tool for intracranial hemorrhage (ICH) and evaluated its diagnostic performance, assessing clinical workflow metrics compared to pre-AI implementation.71 The study found that the tool demonstrated practical diagnostic performance with an overall accuracy of 93.0%, sensitivity of 87.2%, and a negative predictive value of 97.8%. However, it identified lower detection rates for specific subtypes of ICH, such as 69.2% for subdural hemorrhage and 77.4% for acute subarachnoid hemorrhage. It is important to define a clear framework for clinical integration, recognizing the limitations of AI.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076
- One of the major limitations of the AI system highlighted by this study is its susceptibility to false negative and false positive findings. Of the 71 fractures identified by CT, 13 were missed by the AI solution, while the radiologist missed only six fractures [2]. False negatives are of particular concern in the clinical setting, because missed fractures can lead to delayed or inappropriate treatment, increasing the risk of complications and potentially worsening patient outcomes. Moreover, the AI solution also generated 15 false positive findings, which can result in unnecessary further imaging or treatment, increasing patient anxiety and healthcare costs. This result underscore the need for sufficient training data, such as a low prevalence of rare fractures, which is a constant issue for AI applications [3]. As the study suggests, AI should be considered as a complementary tool rather than a replacement tool for human expertise, at least until further fine tuning can address these shortcomings. Future studies could focus on the combination of radiologists and AI tools, which may be a good balance to maximize the accuracy of bone fracture detection.
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press) - In their study, Pastor et al. compared the diagnostic performance of a deep learning algorithm (Rayvolve®, AZmed) with that of experienced radiologists in detecting bone fractures in adult patients using radiographs [2]. With 94 patients included, this study evaluated both the sensitivity and specificity of the AI solution and human radiologists, using computed tomography (CT) as ground truth. The results demonstrated that while the AI solution performed reasonably well, it was consistently outperformed by the radiologists. The AI solution achieved a sensitivity (i.e., the ability to correctly identify fractures) of 82 % and a specificity (i.e., the ability to correctly rule out fractures in patients without fractures) of 69 % [2]. By comparison, the radiologists achieved a sensitivity of 92 % and a specificity of 88 %, demonstrating that human expertise remains critical in the clinical setting.
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press) - “ AI tools have varying levels of performance, with some favoring sensitivity and others favoring accuracy, depending on the specific goal to achieve. Future studies should focus on improving the sensitivity and specificity of AI solutions, particularly in detecting fractures in challenging anatomical regions such as the hands, wrists, and feet, which are often missed by both AI and radiologists. As the technology continues to evolve, the role of AI in healthcare will undoubtedly grow. However, the study by Pastor et al. underscores the need for caution in adopting AI without first addressing its limitations. By maintaining a balance between technological innovation and human expertise, we can ensure that AI enhances, rather than diminishes, the quality of patient care. ”
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press)
- “The combined model showed improved performance compared to both the clinical and radiomics models in the test group, with an AUC of 0.844, accuracy of 0.767, sensitivity of 0.806, and specificity of 0.667. Subsequently, DCA of the combined model indicated optimal clinical value for predicting PNI status. Machine learning radiomics models can accurately predict PNI status in patients with pancreatic ductal adenocarcinoma. The combined model, which incorporates clinical and radiomics features, enhances preoperative diagnostic performance and aids in the selection of treatment methods.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1 - “Multivariate logistic regression was employed to identify independent predictors and establish clinical models. A combined model was constructed by integrating clinical and radiomics features. Model performances were assessed by receiver operating characteristic curves (ROCs) and decision curve analyses (DCAs). A total of 788 radiomics features were extracted from CECT images, of which 14 were identified as significant through the three-step selection process. Among the machine learning models, the SVM radiomics model exhibited the highest predictive performance in the test group, achieving an area under the curve (AUC) of 0.831, accuracy of 0.698, sensitivity of 0.677, and specificity of 0.750. After logistic regression screening, the clinical model was established using carbohydrate antigen 19–9 (CA199) levels as one independent predictor. In the test group, the clinical model demonstrated an AUC of 0.644, accuracy of 0.744, sensitivity of 0.871, and specificity of 0.417.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1 - “In fact, clinical factors and imaging features also need to be considered in the clinic. Our study demonstrated that the combined models surpassed the performance of models based solely on radiomics signatures or clinical features in the test group, achieving an AUC of 0.844. Lee et al. integrated clinical and radiomics features to develop a hybrid model for predicting early recurrence of pancreatic ductal adenocarcinoma, achieving the highest diagnostic power in the test set (AUC, 0.830), which aligned with our research findings. This demonstrated that creating a combined model was more beneficial for improving diagnostic performance.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1 - “This study had several limitations. Firstly, as a retrospective study, it is inherently subject to biases such as selection bias and potential confounders. In the future, a larger sample size multi-center retrospective study or a prospective study is suggested to improve the robustness of the model. Secondly, ROIs of tumors were delineated manually, and the next step will be to try to use automated or semi-automated segmentation methods for automatic segmentation and delineation of the tumors to improve and provide more consistent and objective results. Thirdly, our study classified PNI as positive or negative; however, the extent of PNI may be a more valuable prognostic factor. Radiomics models could offer an efficient method for differentiating between various grades of PNI, which should be further explored and validated infuture studies.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1
- TAKE-HOME POINTS
- A comprehensive and multidimensional ROI calculator was designed to assess both monetary and nonmonetary benefits of integration of digital applications for AI in radiology settings.
- Relative to current practices, the use of the platform resulted in reductions of 16 days of waiting time and 78 days of triage time (through scan reprioritization), 10 days of reading time, and 41 days of reporting time, totaling 145 days saved overall.
- The use of the platform also provided gains for the hospital through optimizing diagnostic and therapeutic procedures with more, but often shorter, hospitalizations, to the benefit of the patients.
Unlocking the Value: Quantifying the Return on Investment of Hospital Artificial Intelligence
Prateek Bharadwaj et al.
J Am Coll Radiol 2024;21:1677-1685. 
Unlocking the Value: Quantifying the Return on Investment of Hospital Artificial Intelligence
Prateek Bharadwaj et al.
J Am Coll Radiol 2024;21:1677-1685.
Unlocking the Value: Quantifying the Return on Investment of Hospital Artificial Intelligence
Prateek Bharadwaj et al.
J Am Coll Radiol 2024;21:1677-1685.- “Results: In the calculator, the introduction of an AI platform into the hospital radiology workflow resulted in labor time reductions and delivery of an ROI of 451% over a 5-year period. The ROI was increased to 791% when radiologist time savings were considered. Time savings for radiologists included more than 15 8-hour working days of waiting time, 78 days in triage time, 10 days in reading time, and41 days in reporting time. Using the platform also provided revenue benefits for the hospital in bringing in patients for clinically beneficial follow-up scans, hospitalizations, and treatment procedures. Results were sensitive to the time horizon, health center setting, and number of scans performed. Among those, the most influential outcome was the number of additional necessary treatments performed because of AI identification of patients. Conclusions: The authors demonstrate a substantial 5-year ROI of implementing an AI platform in a stroke management–accredited hospital. The ROI calculator may be useful for decision makers evaluating AI-powered radiology platforms.”
Unlocking the Value: Quantifying the Return on Investment of Hospital Artificial Intelligence
Prateek Bharadwaj et al.
J Am Coll Radiol 2024;21:1677-1685. - “The ROI calculator demonstrated that on average, health care organizations could expect a positive ROI for each year of investment. This finding was consistent with previous studies that have shown the financial benefits of AI technologies in health care, including radiology. Additionally, the calculator highlighted significant efficiency gains, such as automating tasks, reducing interpretation time, and enhancing accuracy. These benefits contribute to increased radiologist productivity, reduced turnaround time for reports, improved patient care, and potential avoidance of costly litigations from missed diagnoses.”
Unlocking the Value: Quantifying the Return on Investment of Hospital Artificial Intelligence
Prateek Bharadwaj et al.
J Am Coll Radiol 2024;21:1677-1685.
- ”A robust argument for AI in dentistry arises from current diagnostic deficiencies. For instance, 43% of cavities on radiographs go entirely undiagnosed, 20% of diagnosed cavities are not actually cavities, and almost 50% of periapical lesions on radiographs are overlooked.Additionally, there is a general distrust towards dentists, with only 55% trust in the accuracy of dental diagnoses, 61% of Americans surveyed saying they have switched dentists or refused treatment after a diagnosis, and 40% expressing skepticism.This distrust is further evidenced by a study from the Dental AI Council, where 136 practitioners diagnosing a single patient’s full set of radiographs found <50% diagnostic agreement, with treatment cost recommendations ranging from $300 to $36,000. A UCLA study also highlighted this issue, showing a disparity in diagnostic accuracy between students (51%) and an AI system (94%). The inconsistency in dental diagnoses is not a new issue. A 1997 Reader’s Digest article documented a journalist’s experience of receiving 50 different diagnoses from dentists across the United States, with treatment quotes ranging from $160 to $30,000. Clearly, there is a significant problem with consistency in the field. ”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - ”Numerous studies have shown that AI tools can help improve the detection and characterization of periapical pathology and periodontitis, improve operational efficiency, and provide crucial information in treatment planning. AI also enhances patient communication. Anatomical tooth segmentation can visually demonstrate how a cavity is encroaching into the dentin, and this visual aid can potentially increase patient treatment acceptance. The principal focus, however, is not on driving production, but on establishing a higher standard of truth and addressing the widespread problem of underdiagnosis in dentistry. AI-enabled dental co-pilot can also integrate multimodal data from dental radiographs, photographs, clinical parameters, procedure codes, and unstructured clinical notes to offer action-oriented opportunities and insights to all stakeholders within a practice.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - ”On a daily basis, AI-driven tools are elevating the standard of care in dentistry. Beyond improving diagnostic accuracy, AI tools can also improve the operational efficiency of the practices by automating appointment scheduling, billing, and reimbursement optimization. Virtual assistants can streamline patient triage and management by highlighting the relevant radiography and diagnostic needs for each patient. These AI tools enable the dentists to focus on delivering quality dental care during their patient facing time.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - “As AI becomes increasingly prominent in radiology, we should look to other allied fields, such as dentistry, and learn from their successes. First, AI algorithms have the ability to interpret mountains of data and perform certain tasks which exceed human abilities. For example, AI can equal or in many cases outperform radiologists in the early detection of certain cancers, such as lung and pancreatic cancer, and is increasingly used in breast screenings. Further, AI has a promising ability to utilize radiomic features (quantitative attributes extracted from medical images into mineable high-dimensional data) to predict disease progression and treatment outcomes. Working in conjunction with radiologists, there is a strong opportunity for AI to improve diagnostic accuracy and speed of treatment intervention.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press) - “Implementation of AI into radiological practice not only has the potential to improve the rate and accuracy of image analysis, but also to perform a variety of nondiagnostic duties. Evidenced by the integration of AI with patient management systems in dentistry, there is a tremendous opportunity to enhance operational efficiency in radiology departments, although admittedly through the much more complex milieu of the electronic medical record and the radiology information system. AI can optimize scheduling, machine utilization, and patient flow, reducing wait times and improving the overall radiological operation. Further, AI’s automation of routine tasks, such as measuring tumor sizes or identifying anatomical landmarks, alleviates the monotony and workload of radiologists, allowing them to concentrate on more complex diagnostic tasks.”
What can radiologists learn from the AI evolution in Dentistry?
Ophir Tanz, Ryan C. Rizk, Seven P. Rowe, Elliot K. Fishman, Linda C. Chu
Current Problems in Diagnostic Radiology 2024 (in press)
Kidney
- “Urinary tract carcinoma is the fourth most common type of cancer in the United States and can be classified by location and histopathologic level. Approximately 90%–95% of the cases arise from the lower urinary tract, predominantly involving the bladder and, to a lesser extent, the urethra . The remaining 5%–10% of cases involve the upper urinary tract, most commonly the pelvicalyceal system, followed by the ureter. Upper tract urothelial carcinoma(UTUC) accounts for 5%–7% of all renal tumors. Due to advancements in diagnosis, including contrast-enhanced imaging and endoscopic techniques, the incidence of upper tract tumors has risen in recent decades.”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz CV, Stanton ML, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “UTUC and urothelial carcinoma in the bladder have similar endoscopic morphology, classification systems (TNM staging and pathologic grading), and risk factors. However, several clinical and biologic differences suggest that these two diseases are distinct entities. UTUC has more aggressive behavior, with a higher frequency of invasive disease at diagnosis (60%) compared with that of bladder cancer (15–25%) . Also, patients with UTUC have a worse 5-year mortality rate (>50%) than those with bladder cancer (<25%).”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz CV, Stanton ML, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “The incidence of urinary tract carcinoma peaks in patients aged 70–90 years, and it is twice as common in men than in women. Hematuria is the most common symptom and is seen in approximately 75% of cases. Less common symptoms include flank pain, a palpable mass, and gynecomastia due to β-hCG secretion. A pelvicalyceal location is twice as common as ureteral lesions. Within the ureter, the distal ureter is the most common site and accounts for 73% of cases. Key features of urothelial carcinoma are multiplicity and a high incidence of metachronous tumors. Approximately 19%–34% of patients with UTUC have a previous history of bladder cancer, and 8%–13% of patients have concurrent bladder cancer at diagnosis.”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz CV, Stanton ML, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - • CT urography has the highest diagnostic accuracy for UTUC, with high sensitivity (92%) and specificity (95%). Technical considerations in CT urography include appropriate timing of imaging, optimal distention of the collecting system, and minimizing the radiation dose.
• The imaging appearance of UTUC can be broadly categorized intopapillary soft-tissue projections (filling defects), wall thickening, and mass-forming lesions. Window settings should be adjusted per different contrast phases to detect different lesion types.
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz CV, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “Lynch syndrome is an autosomal dominant genetic disorder caused by germline mutations in DNA mismatch repair genes. Many types of cancer risks increase, and UTUC is the third most common malignancy after colorectal and endometrioid carcinoma. The lifetime risk of developing UTUC is 3% . Lynch syndrome–associated UTUC is typically diagnosed in patients at a younger age (62 years)compared with diagnosis of sporadic cases (70 years).”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz CV, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “Two standard image acquisitions are the nephrographic and excretory phases, which are complementary. Their combined use is recommended. The former can be used to detect a wide range of parenchymal and vascular abnormalities, while the latter allows antegrade contrast agent filling of the upper urinary tract to depict urothelial abnormalities. Although less common, use of the urothelial phase 60 seconds after intravenous contrast agent administration has been advocated because it improves the detection of UTUC compared with use of the excretory phase alone and can potentially enable the use of single-phase CT urography without the excretory phase.”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz CV, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “Optimal opacification and distention of the urinary tract are important to detect small UTUCs . Oral hydration, commonly 500–1000 mL of water 20–60 minutes prior to CT urography, is a widely used technique and has similar effects to intravenous hydration. Use of the diuretic furosemide improves distention beyond just hydration. Additional delayed imaging to attempt complete opacification yields no clinical benefit, because nonopacified ureteral segments without associated soft-tissue components are unlikely to harbor UTUC . Coronal and sagittal reformations can often help detect and characterize urothelial abnormalities.”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz C, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - UTUC often appears as papillary soft-tissue projections in the nephrographic phase and as papillary filling defects in the excretory phase. Use of the excretory phase with bone windows is typically the best method, especially to find small lesions. Coronal and sagittal reformations can also help with detection. A focally dilated ureter against an occlusive convex filling defect (“goblet” or “champagne” sign) is a characteristicretrograde pyelographic finding of an obstructing papillary UTUC. This sign may also be identified with CT urography and MR urography.
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz C, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “UTUC can appear as wall thickening. Unlike filling defects, the best imaging phase for detection of wall thickening varies. Because detection of mild wall thickening can be challenging in a single phase, use of nephrographic and excretory phases combined is recommended because these are complementary. The positive predictive value of CT urography for wall thickening (70%) is lower than that for filling defects (88%), so diagnostic ureteroscopy or short-term follow- up CT urography should be considered. Wall thickeningis the most common abnormal finding of upper tract carcinoma in situ, although the sensitivity is limited even for CT urography (42%–54%).”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz C, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - “Many mass-forming UTUCs are thought to have progressed from papillary projections or wall thickening. Carcinoma in situ can appear as a mass in the renal collecting system, although rarely in the ureter. Even a large pelvicalyceal mass could be stage T2 UTUC or lower. However, excluding the presence of peripelvic fat invasion is sometimes challenging because expansile tumors confined within the renal collecting system can result in effacement of the surrounding renal sinus fat. Also, detection of mass-forming UTUC can be challenging if the renal calyx is filled with tumor (phantom calyx). Therefore, it is recommended to check the influx of contrast agent into each of the renal calyces in the axial plane as well as on coronal or sagittal reformations.”
Imaging of Upper Tract Urothelial Carcinoma.
Nakai H, Takahashi H, Wellnitz C, Takahashi N, Kawashima A.
Radiographics. 2024 Nov;44(11):e240056. doi: 10.1148/rg.240056. - UTUC: Differential Dx
- Inflammation/infection
- Hemorrhage
- Systemic disease (amyloidosis)
- Structural abnormalities (papillary necrosis)
- Benign tumors (papilloma)
- Malignant tumors (RCC or Metastases)
- “Bladder cancer (BC) represents a significant global health burden, with an estimated 83,190 new cases and 16,840 deaths in the United States in 2024 alone.The worldwide incidence of BC was 4.3 per 100,000 and mortality of 2 per 100,000 in both genders in 2022. Despite advancements in treatment modalities, the management of BC remains complex, necessitating a multidisciplinary approach for accurate diagnosis, staging, and treatment planning. Imaging plays a pivotal role in this paradigm, serving as a cornerstone for disease evaluation and guiding therapeutic strategies.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “CTU is most commonly used for upper and lower urinary tract imaging. Approximately 1-6% of patients with BC would have synchronous upper tract urothelial carcinoma (UTUC) and 0.5-5% will develop metachronous UTUC. CTU has 67- 100% sensitivity and 93-99% specificity for detecting UTUC. A multiphasic CTU protocol involves acquiring images of the urinary tract at different phases of contrast enhancement including noncontrast imaging to provide baseline anatomic information, nephrographic phase at 100 seconds post contrast injection to assess for renal lesions, and an excretory phase scan at 15 minutes post injection to allow visualization of contrast excretion into the renal pelvis, ureters, and bladder.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “Urothelial carcinoma (UC), previously referred to as transitional cell carcinoma, stands as the most prevalent cellular type of BC, accounting for approximately 90% of cases in western countries. This malignancy originates from the urothelial cells lining the inner surface of the bladder.1 Morphologically, UC can manifest in several forms, including papillary, non-papillary (flat), and mixed variants— such as aggressive micropapillary carcinoma, with a poor prognosis. Papillary UCs typically present with finger-like projections into the bladder lumen, while non-papillary tumors exhibit a flat morphology. Recent research has identified molecular subtypes of UC, each associated with distinct clinical behaviors and treatment responses.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “Squamous cell carcinoma is the second most common histologic type (3–8%), arising from squamous epithelial cells within the bladder mucosa. This subtype is often associated with chronic irritation or inflammation, such as recurrent urinary tract infections or prolonged catheter use. Histologically, squamous cell carcinoma presents as infiltrative masses with features like keratinization and intercellular bridges. Despite its relatively low frequency compared to UC, squamous cell carcinoma tends to be aggressive and associated with a poorer prognosis.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “Hematuria, urinary voiding symptoms such as urgency and frequency, and recurrent urinary tract infections are commonly observed in patients with BC. The prevalence of BC symptoms varies, with approximately 10-20% of patients presenting with gross hematuria and 3-5% with microscopic hematuria. In 2020, the AUA updated their guidelines for microscopic hematuria from the 2012 guidelines, providing a new individualized risk stratified approach based on age, smoking history, and quality and quantity of hematuria. In these guidelines, low-risk patients are recommended to be involved with shared decision making for either repeating urinalysis or undergoing cystoscopy plus renal ultrasound, intermediate-risk patients to undergo cystoscopy and ultrasound, and high-risk patients undergoing to cystoscopy and CTU. The guidelines further emphasize that urine cytology or urine-based tumor markers should not be employed in the initial evaluation of microscopic hematuria.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “fCTU has become the preferred imagingmodality for diagnosing and staging BC. As a primary imagingmodality for evaluating BC, it offers comprehensive visualization of the entire urinary tract, including the bladder, ureters, and kidneys, allowing for the detection of other possible sources of hematuria, primary tumors, local invasion, lymph node involvement, and distant metastases. As one of the first imaging modalities for evaluating hematuria along with cystoscopy, it allows characterization of the upper urinary tract with approximately 2% of patients demonstrating a synchronous tumor and 6% developing a metachronous tumor. Hence, it is recommended by both the American College of Radiology (ACR) and NCCN.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press 
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press- “Because the bladder wall reacts to infectious/inflammatory stimuli often with wall thickening and/ or nodularity, these etiologies form a large portion of BC mimics. Although this wall thickening is usually uniform, an isolated inflammatory process near the bladder (eg, appendicitis, colitis, and diverticulitis) can cause focal wall thickening. Recurrent bouts of cystitis or prior surgery can also form heterogenous wall thickening, causing polypoid or mass-like thickening (pseudotumor) with typically delayed central enhancement because of internal fibrosis. Differentiating pseudotumor from BC on imaging alone may be difficult and may ultimately rely on cystoscopy and biopsy for confirmation.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “Patients undergoing treatment of pelvic tumors (including BC) can demonstrate findings that mimic BC. In the setting of Bacillus Calmette-Guerin treatment of BC, wall thickening with granulomatous changes is often seen. This can also be seen in the setting of pelvic radiation that can also cause hematuria (hemorrhagic cystitis). Immune compromised patients and those with autoimmune disease may develop malacoplakia of the bladder (most common site involved in the genitourinary system), demonstrated by hyperplasia of the epithelium with more flat and plaque-like features; however, nodular and mass-like presentations can also be encountered and biopsy is often necessary for accurate diagnosis.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “Finally, benign tumors or blood products/clot can easily mimic BC. Although rare, leiomyoma arising from smooth muscle in the bladder wall can exhibit intraluminal components. Other benign masses arise from the bladder wall, including paragangliomas, hemangiomas, lipomas, neurofibromas, and lymphangiomas. Lastly, benign extrinsic bladder lesions such as endometrial implants may infiltrate the bladder wall, exhibiting a heterogeneous appearance based on the menstrual cycle.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “An important frontier in radiology has been the deployment of machine learning to aid both in segmenting lesions on imaging and deriving quantitative features beyond those visibly interpreted by a radiologist. Texture analysis and Radiomic algorithms seek to extract quantitative features, many of which are now available via open-source software, including mathematical features like gray level size zone matrix. These are then used to build models that help identify a particular pathology, such as clinically significant prostate cancer. These paradigms are emerging more broadly in radiology, with applications for renal mass analysis, liver metastases, pancreatic cancer, and other abdominal malignancies. Radiomics has also been utilized in BC detection.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press - “CTU is the initial examination for hematuria, providing information on the upper and lower urinary tract imaging. It also has become the preferred imaging method for systemic staging BC, allowing the detection of possible sources of hematuria, including synchronous and metachronous UC. It also provides information for lymph node involvement, and distant metastases.”
Urologic Imaging of the Bladder: Cancers and Mimics
Haleh Amirian, Felipe B. Franco, Borna Dabiri, MD, Francesco Alessandrino
Urol Clin N Am - (2025) in press
Musculoskeletal
- One of the major limitations of the AI system highlighted by this study is its susceptibility to false negative and false positive findings. Of the 71 fractures identified by CT, 13 were missed by the AI solution, while the radiologist missed only six fractures [2]. False negatives are of particular concern in the clinical setting, because missed fractures can lead to delayed or inappropriate treatment, increasing the risk of complications and potentially worsening patient outcomes. Moreover, the AI solution also generated 15 false positive findings, which can result in unnecessary further imaging or treatment, increasing patient anxiety and healthcare costs. This result underscore the need for sufficient training data, such as a low prevalence of rare fractures, which is a constant issue for AI applications [3]. As the study suggests, AI should be considered as a complementary tool rather than a replacement tool for human expertise, at least until further fine tuning can address these shortcomings. Future studies could focus on the combination of radiologists and AI tools, which may be a good balance to maximize the accuracy of bone fracture detection.
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press) - In their study, Pastor et al. compared the diagnostic performance of a deep learning algorithm (Rayvolve®, AZmed) with that of experienced radiologists in detecting bone fractures in adult patients using radiographs [2]. With 94 patients included, this study evaluated both the sensitivity and specificity of the AI solution and human radiologists, using computed tomography (CT) as ground truth. The results demonstrated that while the AI solution performed reasonably well, it was consistently outperformed by the radiologists. The AI solution achieved a sensitivity (i.e., the ability to correctly identify fractures) of 82 % and a specificity (i.e., the ability to correctly rule out fractures in patients without fractures) of 69 % [2]. By comparison, the radiologists achieved a sensitivity of 92 % and a specificity of 88 %, demonstrating that human expertise remains critical in the clinical setting.
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press) - “ AI tools have varying levels of performance, with some favoring sensitivity and others favoring accuracy, depending on the specific goal to achieve. Future studies should focus on improving the sensitivity and specificity of AI solutions, particularly in detecting fractures in challenging anatomical regions such as the hands, wrists, and feet, which are often missed by both AI and radiologists. As the technology continues to evolve, the role of AI in healthcare will undoubtedly grow. However, the study by Pastor et al. underscores the need for caution in adopting AI without first addressing its limitations. By maintaining a balance between technological innovation and human expertise, we can ensure that AI enhances, rather than diminishes, the quality of patient care. ”
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press)
Pancreas
- “The combined model showed improved performance compared to both the clinical and radiomics models in the test group, with an AUC of 0.844, accuracy of 0.767, sensitivity of 0.806, and specificity of 0.667. Subsequently, DCA of the combined model indicated optimal clinical value for predicting PNI status. Machine learning radiomics models can accurately predict PNI status in patients with pancreatic ductal adenocarcinoma. The combined model, which incorporates clinical and radiomics features, enhances preoperative diagnostic performance and aids in the selection of treatment methods.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1 - “Multivariate logistic regression was employed to identify independent predictors and establish clinical models. A combined model was constructed by integrating clinical and radiomics features. Model performances were assessed by receiver operating characteristic curves (ROCs) and decision curve analyses (DCAs). A total of 788 radiomics features were extracted from CECT images, of which 14 were identified as significant through the three-step selection process. Among the machine learning models, the SVM radiomics model exhibited the highest predictive performance in the test group, achieving an area under the curve (AUC) of 0.831, accuracy of 0.698, sensitivity of 0.677, and specificity of 0.750. After logistic regression screening, the clinical model was established using carbohydrate antigen 19–9 (CA199) levels as one independent predictor. In the test group, the clinical model demonstrated an AUC of 0.644, accuracy of 0.744, sensitivity of 0.871, and specificity of 0.417.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1 - “In fact, clinical factors and imaging features also need to be considered in the clinic. Our study demonstrated that the combined models surpassed the performance of models based solely on radiomics signatures or clinical features in the test group, achieving an AUC of 0.844. Lee et al. integrated clinical and radiomics features to develop a hybrid model for predicting early recurrence of pancreatic ductal adenocarcinoma, achieving the highest diagnostic power in the test set (AUC, 0.830), which aligned with our research findings. This demonstrated that creating a combined model was more beneficial for improving diagnostic performance.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1 - “This study had several limitations. Firstly, as a retrospective study, it is inherently subject to biases such as selection bias and potential confounders. In the future, a larger sample size multi-center retrospective study or a prospective study is suggested to improve the robustness of the model. Secondly, ROIs of tumors were delineated manually, and the next step will be to try to use automated or semi-automated segmentation methods for automatic segmentation and delineation of the tumors to improve and provide more consistent and objective results. Thirdly, our study classified PNI as positive or negative; however, the extent of PNI may be a more valuable prognostic factor. Radiomics models could offer an efficient method for differentiating between various grades of PNI, which should be further explored and validated infuture studies.”
Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast‑Enhanced CT Imaging
Wenzheng Lu · Yanqi Zhong · Xifeng Yang ·et al.
Journal of Imaging Informatics in Medicine https://doi.org/10.1007/s10278-024-01325-1
Small Bowel
- KEY POINTS
-Non-invasive imaging with whole-body computed tomography (WBCT) is an excellent tool for the prompt screening, diagnosis, management, and surveillance of potentially life-threatening trauma related injuries in the significantly or severely injured patient; however, its role in those without obvious injury is debatable.
- WBCT may be used to identify unexpected critical injuries and incidental findings that may affect mortality and morbidity, thereby making it appropriate despite associated costs and radiation exposure. -The decision to utilize WBCT in trauma is ultimately that of the managing medical/surgical team.
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Multiple studies have shown that WB-CT has sensitivity greater than 95% and a negative predictive value approaching 100% to identify treatable polytrauma injuries. It can be used to detect unexpected injuries in approximately 22% of patients, and to identify additional traumatic findings that may lead to change in management in up to 34% of patients. There is also questionable evidence demonstrating the beneficial effect on the survival of polytrauma patients evaluated by immediate WB-CT in all such cases, and there are data showing shorter time to treatment and length of stay for polytrauma patients who underwent assessment with WBCT.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - Perienteric findings associated with small bowel CD include perienteric edema or fat stranding, which are present in more severe disease. Engorged vasa recta, also known as the “comb sign,” results from current or previous bowel inflammation. Findings of chronicity include sacculations and fibrofatty proliferation. Sacculations result from shortening along the mesenteric border with ballooning along the antimesenteric border. Fibrofatty proliferation is hypertrophy of the mesenteric fat adjacent to diseased bowel segments due to repeated inflammation. Mesenteric venous thrombosis or occlusion has been described in CD. It is important to distinguish between acute mesenteric thrombosis and chronic mesenteric vein thrombosis. Chronic peripheral mesenteric vein occlusion has been shown to anatomically correspond with small bowel segments with active or prior CD inflammation.
Evidence-Based Review of Current Cross-Sectional Imaging of Inflammatory Bowel Disease
Jesi Kim, MDa,b,c, Bari Dane, MDc
Radiol Clin N Am 62 (2024) 1025–1034
- “Extraintestinal findings include sacroiliitis, primary sclerosing cholangitis (PSC), avascular necrosis of the hips, nephrolithiasis, cholelithiasis, and pancreatitis (type 2 autoimmune pancreatitis and due to cholelithiasis).10 Many patients with CD present with back pain, and the detection of sacroiliitis allows for identification of the cause and therapy facilitation. PSC can be seen in both CD and UC; however, it is more common in UC.”
Evidence-Based Review of Current Cross-Sectional Imaging of Inflammatory Bowel Disease
Jesi Kim, MDa,b,c, Bari Dane, MDc
Radiol Clin N Am 62 (2024) 1025–1034 - Perienteric findings associated with small bowel CD include perienteric edema or fat stranding, which are present in more severe disease. Engorged vasa recta, also known as the “comb sign,” results from current or previous bowel inflammation. Findings of chronicity include sacculations and fibrofatty proliferation. Sacculations result from shortening along the mesenteric border with ballooning along the antimesenteric border. Fibrofatty proliferation is hypertrophy of the mesenteric fat adjacent to diseased bowel segments due to repeated inflammation. Mesenteric venous thrombosis or occlusion has been described in CD. It is important to distinguish between acute mesenteric thrombosis and chronic mesenteric vein thrombosis. Chronic peripheral mesenteric vein occlusion has been shown to anatomically correspond with small bowel segments with active or prior CD inflammation.
Evidence-Based Review of Current Cross-Sectional Imaging of Inflammatory Bowel Disease
Jesi Kim, MDa,b,c, Bari Dane, MDc
Radiol Clin N Am 62 (2024) 1025–1034 - “Recently, there have been efforts to quantify disease activity using CTE and MRE. Dual-energy CT allows for the determination of iodine density, which reflects only the iodine content within a voxel. Iodine density has been shown to be a surrogate marker of perfusion. Prior studies have shown that dual-energy CTE-obtained iodine density correlates with active inflammation in CD. In addition, photon-counting CT enterography iodine density can be used to distinguish mild from moderate-to-severe active inflammation.”
Evidence-Based Review of Current Cross-Sectional Imaging of Inflammatory Bowel Disease
Jesi Kim, MDa,b,c, Bari Dane, MDc
Radiol Clin N Am 62 (2024) 1025–1034
Trauma
- One of the major limitations of the AI system highlighted by this study is its susceptibility to false negative and false positive findings. Of the 71 fractures identified by CT, 13 were missed by the AI solution, while the radiologist missed only six fractures [2]. False negatives are of particular concern in the clinical setting, because missed fractures can lead to delayed or inappropriate treatment, increasing the risk of complications and potentially worsening patient outcomes. Moreover, the AI solution also generated 15 false positive findings, which can result in unnecessary further imaging or treatment, increasing patient anxiety and healthcare costs. This result underscore the need for sufficient training data, such as a low prevalence of rare fractures, which is a constant issue for AI applications [3]. As the study suggests, AI should be considered as a complementary tool rather than a replacement tool for human expertise, at least until further fine tuning can address these shortcomings. Future studies could focus on the combination of radiologists and AI tools, which may be a good balance to maximize the accuracy of bone fracture detection.
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press) - In their study, Pastor et al. compared the diagnostic performance of a deep learning algorithm (Rayvolve®, AZmed) with that of experienced radiologists in detecting bone fractures in adult patients using radiographs [2]. With 94 patients included, this study evaluated both the sensitivity and specificity of the AI solution and human radiologists, using computed tomography (CT) as ground truth. The results demonstrated that while the AI solution performed reasonably well, it was consistently outperformed by the radiologists. The AI solution achieved a sensitivity (i.e., the ability to correctly identify fractures) of 82 % and a specificity (i.e., the ability to correctly rule out fractures in patients without fractures) of 69 % [2]. By comparison, the radiologists achieved a sensitivity of 92 % and a specificity of 88 %, demonstrating that human expertise remains critical in the clinical setting.
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press) - “ AI tools have varying levels of performance, with some favoring sensitivity and others favoring accuracy, depending on the specific goal to achieve. Future studies should focus on improving the sensitivity and specificity of AI solutions, particularly in detecting fractures in challenging anatomical regions such as the hands, wrists, and feet, which are often missed by both AI and radiologists. As the technology continues to evolve, the role of AI in healthcare will undoubtedly grow. However, the study by Pastor et al. underscores the need for caution in adopting AI without first addressing its limitations. By maintaining a balance between technological innovation and human expertise, we can ensure that AI enhances, rather than diminishes, the quality of patient care. ”
Artificial intelligence for bone fracture detection: A promising tool but no substitute for human expertise
Daphne Guenouna, Mickael Tordjman
Diagnostic and Interventional Imaging 2024 (in press)
- ”WB-CT is typically performed with modern multi-detector CT (MDCT) scanners with a continuous acquisition, with thin (0.5–0.6 mm) collimation images reconstructed at 1 mm and 3 mm slices for imaging interpretation—the anatomic coverage extending from the head to the symphysis pubis. The authors’ institutional protocol involves a biphasic injection of 100 mL of iodinated contrast (350 mg/mL) at 4 cc/s for 15 seconds, then at a rate of 3 cc/s, followed by a 40-cc saline bolus at 4 cc/s. The fixed-time delay method is generally suitable, but issues arise in cases of abnormal anatomy, cardiac function, or severe arterial disease, impacting blood transit time. While bolus tracking is often favored in such scenarios, research shows improved efficiency witha fixed-time empiric delay of 20 or 25 seconds for patients over 55.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Post-processing techniques, including coronal and sagittal multiplanar reformations (MPR) and maximum intensity projections (MIPS) of the torso are also generated and submitted for interpretation. Separate dedicated coronal and sagittal reformations of the neck and thoracolumbar spine in both bone and soft-tissue algorithms are generated, with the liberal use of additional postprocessing techniques by the radiologist.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Second to intracranial hemorrhage, traumatic aortic injuries (TAI) are the next most common cause of motor vehicle collision (MVC) deaths, and it is estimated that 80% of patients with aortic injuries die before arriving at the hospital.MDCT has been found to be nearly 100% sensitive and specific for the detection of TAIs. Most TAIs occur at the isthmus. Abdominal TAIs are rare, and 25% represent an extension of a thoracic aortic injury.The infrarenal aortic segment is twice as commonly involved as the supra-renal segment. The presence of a vertebral body fracture should prompt the interpreter to closely scrutinize the aorta. Sagittal MPR reconstructions are especially useful in assessing the aorta.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Extraluminal gas, bowel wall discontinuity, mesenteric vessel irregularity, occlusion, or extravasation, and focal bowel wall thickening all have high specificities for bowel injury on CT. Other sources of free air in the setting of blunt trauma may be related to the extension of a pneumothorax or pneumomediastinum, barotrauma from mechanical ventilation, DI, chest tube placement, diagnostic peritoneal lavage (DPL), and intraperitoneal rupture of a recently instrumented bladder. Free fluid on CT, particularly if hyperdense, is the most sensitive finding for a bowel injury (BI), with a 90% to 100% sensitivity, but has only a 15% to 25% specificity.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Delayed CT imaging alone is not reliable in demonstrating a bladder injury, as the bladder is usually not fully distended. Therefore, a dedicated CT-cystogram should be performed with an attempt for optimal bladder distension. A total of 70% of bladder injuries are extraperitoneal and are associated with pelvic fractures (managed by placing a supra-pubic catheter), and 20% are intraperitoneal (requiring operative repair).”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Anatomic variants may also potentially mimic traumatic injuries. An aortic ductus diverticulum or aortic spindle may be confused for a pseudoaneurysm (PSA), a remnant patent or calcified ductus arteriosum may be misconstrued for a TAI, thymus may be confused for a mediastinal hematoma, while a splenic cleft or pancreatic lobulation can mimic a laceration. The absence of associatedfat stranding or fluid may hint that the finding is non-traumatic. Calcifications or foreign bodies may also resemble areas of hemorrhage, which can be differentiated by the area’s density and stable appearance in all phases of imaging.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “Integrating artificial intelligence (AI) in emergency radiology, particularly in the context of WB-CTbscans for polytraumatized patients, represents a substantial advancement in the diagnosis and subsequent management of traumatic injuries. A review of the current literature underscores the promising potential of AI to enhance the accuracy, efficiency, and predictive capabilities of radiological assessments in the emergency and trauma settings. Zhou and colleagues conducted a study on 133 patients comparing the results of rib fracture detection using initial CT and follow-up CT, revealing that AI-assisted diagnosis statistically significantly improved the overall accuracy for the detection of rib fractures.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076 - “The innovative application of AI extends beyond these examples. AI has been employed in the quantitative analysis of traumatic hemoperitoneum,70 highlighting its ability to aid in diagnosing and managing intrabdominal bleeding. Seyam and colleagues implemented an AI-based detection tool for intracranial hemorrhage (ICH) and evaluated its diagnostic performance, assessing clinical workflow metrics compared to pre-AI implementation.71 The study found that the tool demonstrated practical diagnostic performance with an overall accuracy of 93.0%, sensitivity of 87.2%, and a negative predictive value of 97.8%. However, it identified lower detection rates for specific subtypes of ICH, such as 69.2% for subdural hemorrhage and 77.4% for acute subarachnoid hemorrhage. It is important to define a clear framework for clinical integration, recognizing the limitations of AI.”
Trauma and ‘Whole’ Body Computed Tomography Role, Protocols, Appropriateness, and Evidence to Support its Use and When
Daniela Galan et al.
Radiol Clin N Am 62 (2024) 1063–1076