Imaging Pearls ❯ September 2023
-- OR -- |
|
3D and Workflow
- “Cinematic rendering, a postprocessing technique that uses complex light paths and high dynamic range light maps, has enabled more photorealistic CT image reconstruction than contemporary postprocessing techniques such as volume rendering. In a study of 16 medical students, participants were able to read and comprehend cinematic rendering–reconstructed musculoskeletal anatomy models faster than conventional volume rendering reconstructions. Cinematic rendering reconstructions possess more depth and increased surface detail compared with volume rendering, attributable to cinematic rendering’s complex global lighting model. In addition, pictorial essays have highlighted the detail and clarity of cinematic rendering reconstructions in evaluating multifaceted clinical musculoskeletal pathologies, including complex fractures. As such, cinematic rendering represents an important innovation in postprocessing CT reconstruction, with both pedagogic and clinical utility in musculoskeletal imaging.”
Musculoskeletal CT Imaging: State-of-the-Art Advancements and Future Directions.
Demehri S, Baffour FI, Klein JG, Ghotbi E, Ibad HA, Moradi K, Taguchi K, Fritz J, Carrino JA, Guermazi A, Fishman EK, Zbijewski WB.
Radiology. 2023 Aug;308(2):e230344. doi: 10.1148/radiol.230344. PMID: 37606571. - “Cinematic rendering reconstructions possess more depth and increased surface detail compared with volume rendering, attributable to cinematic rendering’s complex global lighting model. In addition, pictorial essays have highlighted the detail and clarity of cinematic rendering reconstructions in evaluating multifaceted clinical musculoskeletal pathologies, including complex fractures. As such, cinematic rendering represents an important innovation in postprocessing CT reconstruction, with both pedagogic and clinical utility in musculoskeletal imaging.”
Musculoskeletal CT Imaging: State-of-the-Art Advancements and Future Directions.
Demehri S, Baffour FI, Klein JG, Ghotbi E, Ibad HA, Moradi K, Taguchi K, Fritz J, Carrino JA, Guermazi A, Fishman EK, Zbijewski WB.
Radiology. 2023 Aug;308(2):e230344. doi: 10.1148/radiol.230344. PMID: 37606571.
Chest
- ”AI-supported mammography screening resulted in a similar cancer detection rate compared with standard double reading, with a substantially lower screen-reading workload, indicating that the use of AI in mammography screening is safe. The trial was thus not halted and the primary endpoint of interval cancer rate will be assessed in 100 000 enrolled participants after 2-years of follow up.”
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44 - “The results from this randomised trial support the findings of earlier retrospective studies, indicating a general potential of AI to improve screening efficacy and reduce workload. The clinical safety analysis concludes that the AI-supported screen-reading procedure can be considered safe. Implementation of AI in clinical practice to reduce the screen-reading workload could therefore be considered to help address workforce shortages. The assessment of the primary endpoint of interval cancer rate, together with a characterisation of detected cancers in the entire study population, will provide further insight into the efficacy of screening, possible side-effects such as overdiagnosis, and the prognostic implications of using AI in mammography screening, taking cost-effectiveness into account.”
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44 - To our knowledge, this is the first randomised controlled trial investigating the use of AI in mammography screening. In this first report, the objective was to assess the safety of an AI-supported screen-reading procedure, involving triage and detection support. AI-supported screening resulted in 20% more cancers being detected and exceeded the lowest acceptable limit for safety compared with standard double reading without AI, without affecting the false positive rate. The AI supported screen-reading procedure enabled a 44·3% reduction in the screen-reading workload. The results indicate that the proposed screening strategy is safe.
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44 - “In summary, this clinical safety analysis of the MASAI trial, in which an AI system was used to triage screening examinations to single or double reading and as detection support, showed that AI-supported mammography screening can be considered safe, since it resulted in a similar rate of screen-detected cancer—exceeding the lowest acceptable limit for safety—without increasing rates of recalls, false positives, or consensus meetings, and while substantially reducing the screen-reading workload compared with screening by means of standard double reading.”
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44
Colon
- “Gastrointestinal (GI) lymphoma accounts for 5–20% of extranodal lymphomas: the stomach is the most common site, followed by small intestine (ileum (60–65%), jejunum (20%−25%), and duodenum (6%–8%) and then colorectal lymphomas (6–12%)). GI lymphomas most commonly occur around the sixth decade of life and, although rare in childhood, they are the most common GI tumours in this age.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Risk factors implicated in the pathogenesis of GI lymphoma are some infections due to Helicobacter pylori, human immunodeficiency virus infection, Campylobacter jejuni, Epstein-Barr virus, hepatitis B virus, human T-cellmlymphotropic virus-1, and some inflammatory conditions asmceliac disease, inflammatory bowel disease, atrophic gastritis, and parasitic infection.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Clinical findings are not specific and this causes a delay in the diagnosis.The most common symptoms are epigastric pain, weight loss, and anorexia; nausea and vomiting in case of gastric lymphoma is uncommon, except in the later stage of the disease. Other symptoms encountered in these patients are GI bleeding and the presence of an abdominal mass and bowel perforation, mainly in the small bowel.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “The majority of GI lymphomas are of B-cell origin, while just 8%–10% show a T-cell origin . Most low-grade B-cell GI lymphomas are of mucosa-associated lymphoid tissue (MALT) type, while enteropathy-associated T-cell lymphoma is the most common primary gastrointestinal T-cell lymphoma. GI lymphomas represent a heterogeneous group of entities originating from different cell lineage, with lymphoid cell at different stage of development, and with different biologic behaviour. Certain histological subtypes most commonly occur in a precise location as MALT lymphoma in stomach, mantle cell lymphoma in terminal ileum, jejunum, and colon, enteropathy-associated T-cell lymphoma injejunum, and follicular lymphoma in duodenum .”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Primary esophageal lymphomas account for less than 1% in all primary GI lymphomas, while usually result from lymph node metastasis of the lymphomas from the cervical or mediastinal region. Both findings on barium studies, as irregular filling defects, and on CT, as thickened esophageal wall with narrowed lumen, are nonspecific and mimic esophageal adenocarcinoma. However, CT may be useful to differentiate primary esophageal lymphoma from lymph node involvements in the cervical or mediastinal regions, in staging of the disease and in evaluating response to therapy.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “The most common CT patterns of gastric lymphoma are the presence of diffuse or segmental wall thickening of 2–5 cm with low contrast enhancement and extensive lateral extension of the tumour due to submucosal spread; moreover, CT can assess the presence of lymphadenopathies. Less commonly, gastric lymphoma may present on CT as a polypoidal mass, an ulcerative lesion, or a mucosal nodularity. Considering the CT features of lymphoma, in low-grade ones there is less severe gastric wall thickening than in highgrade lymphoma, and abdominal lymphadenopathy is less common.The absence of abnormality or the presence of just minimal gastric wall thickening or a shallow lesion at CT suggests low-grade MALT lymphoma; yet, CT is of limited value in its diagnosis. A greater thickening may indicate transformation to a higher grade lymphoma.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “CT is particularly useful both for staging and in the follow-up after surgery or chemoradiotherapy.Nowadays,CT allows the evaluation of wall thickness, mesenteric vasculature, and any associated extramural findings. Small bowel CT, or entero-CT, performed through a multislice CT scanner has led to considerable advances in the detection and staging of intestinal diseases.The advantage of this technique lies in its panoramic view, which allows the evaluation of the intestinal wall thickness, the degree of bowel distension, and the circular folds. Yet, ileal loops and also those of the deep pelvis, the mesentery, the surrounding adipose tissue, and other abdominal organs are studied.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - The most common CT/MR patterns of small bowel lymphoma are:
(i) Polypoid/nodular pattern.
(ii) Infiltrative pattern.
(iii) Aneurismal pattern.
(iv) Exophytic mass.
(v) Stenosing mass (rare).
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “The polypoid pattern is characterized by the presence of a solid nodule, with a homogeneous signal density/intensity, that develops in the submucosa and protrudes into the lumen appearing as a polypoid mass. There is no wall thickening and/or lymph adenopathy and the mucosa is intact. This mass may cause intussusception. The infiltrative form is characterized by segmental symmetrical or slightly asymmetrical infiltrating lesions with a medium diameter of 1.5 cm and 2 cm, associated with mild circumferential thickening of the small bowel wall. Usually, the infiltrative lesions show ill-defined margins and a homogeneous contrast enhancement; the latter may rarely be inhomogeneous because of the presence of hypodense/ hypointense areas due to development of necrosis and/or ischemia in the context of the lesion. These lesions may extend to the whole bowel thickness, from the endoluminal mucosa to the tunica serosa.The length of the thickened small bowel segment is variable.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “The aneurismal pattern (diameter of dilatation of the lumen over 4 cm), firstly diagnosed byCupps et al. in 1969, represents 31% of small bowel lymphomas. It usually coexists with the infiltrative form since it can represent its natural evolution. Several factors are responsible for the aneurismal dilation secondary to infiltrative growth of neoplastic lesion, as a progressive destruction of myenteric plexus, destruction of muscle layers with stretching of the muscle fibers, and loss of contractile cells; on the other hand, the infiltration of arterial and lymphatic vessels determines anoxia and necrosis within the lesion. According to some authors, this tumour necrosis could lead to cavitation and be also responsible for the aneurismal dilatation.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Differential diagnosis includes all inflammatory, neoplastic, and metastatic lesions involving the small bowel. Primary carcinoma, metastases (especially those from melanoma and renal cancer), and the intestinal leiomyosarcoma are characterized by large necrotic/colliquative cavitations. In rare cases, inflammatory conditions, such as Crohn’s disease and intestinal tuberculosis, have to be differentiated: the significant thickening of the bowel wall (greater than 2 cm), the presence of lymphomatous nodules, and the coexistence of perivisceral multiple lymph nodes are CT features that are suggestive for a lymphoproliferative process.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Primary lymphoma of the large bowel accounts for 0.4% of all tumours of the colon, and colorectal lymphomas constitute 6%–12% of gastrointestinal lymphomas.The cecum and rectum are most commonly affected parts compared to other tracts of the large bowel. Primary large bowel lymphoma may appear as localized, large, extraluminal masses or constricting simulating annular-type carcinomas and may present with different radiological patterns that are often quite similar to other large bowel tumours or inflammatory diseases, thus leading to a difficult differential diagnosis These patterns include bulky polypoidal mass, focal infiltrative tumour, and aneurismal dilatation.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Colonic lymphoma usually presents with larger lesions and involves a longer segment compared to adenocarcinoma; moreover, colonic lymphoma is usually located near the ileocaecal valve and grows into the terminal ileum, not invading or obstructing neighbouring viscera.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Thanks to CT it is possible to study not only the GI tract using enterography technique, but also local and distant lymph nodes and other thoracic and abdominal organs that can be affected also by the disease, thus allowing an imaging staging of the disease according to the classification of the Consensus Conference of Lugano. The role of CT is also considered pivotal in the evaluation of complications of the disease, as perforation, fistulisation, and obstruction, and in the differential diagnosis with other neoplastic or inflammatory conditions, whichmay also coexist with the lymphoma. Lastly, CTmust be actually considered also the preferred technique for the evaluation of response to therapy when medical therapy with targeted therapy is used; in this case according to the used drug, the imaging appearance may be substantially different.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143
Deep Learning
- Purpose: To present a deep learning segmentation model that can automatically and robustly segment all major anatomic structures on body CT images.
Materials and Methods: In this retrospective study, 1204 CT examinations (from 2012, 2016, and 2020) were used to segment 104 anatomic structures (27 organs, 59 bones, 10 muscles, and eight vessels) relevant for use cases such as organ volumetry, disease characterization,and surgical or radiation therapy planning. The CT images were randomly sampled from routine clinical studies and thus represent a real-world dataset (different ages, abnormalities, scanners, body parts, sequences, and sites). The authors trained an nnUNet segmentation algorithm on this dataset and calculated Dice similarity coefficients to evaluate the model’s performance. The trained algorithm was applied to a second dataset of 4004 whole-body CT examinations to investigate age-dependent volume and attenuation changes.
TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images
Jakob Wasserthal, et al.
Radiology: Artificial Intelligence 2023; 5(5):e230024 - Results: The proposed model showed a high Dice score (0.943) on the test set, which included a wide range of clinical data with major abnormalities. The model significantly outperformed another publicly available segmentation model on a separate dataset (Dice score, 0.932 vs 0.871; P < .001). The aging study demonstrated significant correlations between age and volume and mean attenuation for a variety of organ groups (eg, age and aortic volume [rs = 0.64; P < .001]; age and mean attenuation of the autochthonous dorsal musculature [rs = −0.74; P < .001]).
Conclusion: The developed model enables robust and accurate segmentation of 104 anatomic structures. The annotated dataset (https:// doi.org/10.5281/zenodo.6802613) and toolkit (https://www.github.com/wasserth/TotalSegmentator) are publicly available.
TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images
Jakob Wasserthal, et al.
Radiology: Artificial Intelligence 2023; 5(5):e230024 - Key Points
■ The proposed model was trained on a diverse dataset of 1204 CT examinations randomly sampled from routine clinical studies; the dataset contained segmentations of 104 anatomic structures (27 organs, 59 bones, 10 muscles, and eight vessels) that are relevant for use cases such as organ volumetry, disease characterization, and surgical or radiation therapy planning.
■ The model achieved a high Dice similarity coefficient (0.943; 95% CI: 0.938, 0.947) on the test set encompassing a wide range of clinical data, including major abnormalities, and outperformed other publicly available segmentation models on a separate dataset (Dice score, 0.932 vs 0.871; P < .001).
■ Both the training dataset (https://doi.org/10.5281/zenodo.6802613) and developed model (https://www.github.com/wasserth/ TotalSegmentator) are publicly available.
TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images
Jakob Wasserthal, et al.
Radiology: Artificial Intelligence 2023; 5(5):e230024
TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images
Jakob Wasserthal, et al.
Radiology: Artificial Intelligence 2023; 5(5):e230024
- “Radiologists have experienced rapid and far-reaching technological changes in their practices. Whether it is computerized tomography, where the conversation focused on number of detectors (64, 128, 256, 512), or transformation of numbers of X-ray tubes in a scanner (1 or 2) into a whole new technology with photon scanning, or magnetic resonance where the arguments have progressed from 1.5 to 3 Tesla scanners to 7 Tesla units, the technology is transformed quickly. Radiologists and their patients have benefited from change and celebrate it. But those changes took years, and the end-users of those technological changes generally watched its development from afar. Even the integration of Artificial Intelligence (AI) into clinical practice has developed gradually, from the theoretical use of AI to help with image interpretation or workflow management into frequent, integrated use in both private practice and academic practice has become standard clinical practice, whether for pulmonary embolism detection, analysis of mammograms, or triage of an emergency room reading list.”
Watching Innovation in Real Time: The Story of ChatGPT and Radiology.
Fishman EK, Weeks WB, Lavista Ferres JM, Chu LC.
Can Assoc Radiol J. 2023 May 3:8465371231174817. doi: 10.1177/08465371231174817. Epub ahead of print. PMID: 37138372. - “Some of the explanation for these improvements in performance may be explained by the fact that, to date, ChatGPT has been trained only on publicly available information and not specifically on medicine. As versions of large language models are trained on medical books, articles, journals, and guidelines, the power of large language models in medicine will become even stronger. That ChatGPT can pass medical licensing exams is especially impressive when one realizes that the systems were not specifically trained for these tasks and that the available information was often limited as, for example, they could not look at material behind a paywall. We are beginning to realize that if properly guided, the power of this new technology will redefine medicine.3,4 The key then will be to use this technology to strengthen the doctor-patient relationship alive rather than replacing it. Clinicians need to guide the technology into practice.”
Watching Innovation in Real Time: The Story of ChatGPT and Radiology.
Fishman EK, Weeks WB, Lavista Ferres JM, Chu LC.
Can Assoc Radiol J. 2023 May 3:8465371231174817. doi: 10.1177/08465371231174817. Epub ahead of print. PMID: 37138372. - “The speed of change in the world of technology has never been as rapid as it is today. For clinicians - as well as clinical researchers - this can translate to both the best of times and the worst of times. While clinicians and clinical researchers become excited by the development of a new computer chip, CT scanner, or vaccine, they know that time will pass before the new technology impacts our daily workflow. That is, until the introduction of large language models like ChatGPT, an offering that allows us to use the technology as it is being developed. This technology – which has been in development for decades - has the chance to impact everything we do at work or at home. And advancements in this technology, fostered by incredible computing power and amounts of data available for learning, occur so quickly that before you can write and publish an article, your descriptions may be one to two versions behind. The story ChatGPT is not one only of imagining what could be but also seeing the change in near real time.”
Watching Innovation in Real Time: The Story of ChatGPT and Radiology.
Fishman EK, Weeks WB, Lavista Ferres JM, Chu LC.
Can Assoc Radiol J. 2023 May 3:8465371231174817. doi: 10.1177/08465371231174817. Epub ahead of print. PMID: 37138372. - “In a recent editorial, Microsoft's president Brad Smith highlights that, like any other technology in history, AI will be used both as a tool and a weapon. In the fields of medicine and research, large language models can be intentionally misused to create scientific-sounding papers containing false information, which can easily spread on the internet. As some individuals are advocating for a shift away from traditional peer-review processes in medical journals, we must emphasize that the human peer-review process is more vital and critical than ever. Further, in what may be an arms race, reviewers and editors must make use of tools such as ChatGPT to uncover unethical activities on the part of authors.”
Beyond chatting: The opportunities and challenges of ChatGPT in medicine and radiology
Juan M. Lavista Ferres, William B. Weeks, Linda C. Chu, Steven P. Rowe, Elliot K. Fishman
Diagnostic and Interventional Imaging, Volume 104, Issue 6,2023, Pages 263-264, - “AI will undoubtedly have a profound impact across all aspects of civilization. While we should embrace its benefits, we must also be aware of its potential drawbacks and limitations. With every step forward, there are bound to be missteps, and the road ahead will likely be full of surprises. Technology will continue to challenge, motivate, and occasionally even frighten us, but ultimately it has the potential to enhance our lives in countless ways.”
Beyond chatting: The opportunities and challenges of ChatGPT in medicine and radiology
Juan M. Lavista Ferres, William B. Weeks, Linda C. Chu, Steven P. Rowe, Elliot K. Fishman
Diagnostic and Interventional Imaging, Volume 104, Issue 6,2023, Pages 263-264, - “Looking toward the future, AI systems will undoubtedly continue to advance and evolve [3]. As with any tool, it is essential to understand where it can contribute and in what scenarios it will not. When Tim Berners-Lee created the World Wide Web (WWW) in the 1990s, people, for the first time, had access to information that was not previously available. As we have come to learn, the web also is full of inaccuracies and misinformation. As a society, we are still working to help humans deal with this, but this was not a reason to throw out the WWW.”
Beyond chatting: The opportunities and challenges of ChatGPT in medicine and radiology
Juan M. Lavista Ferres, William B. Weeks, Linda C. Chu, Steven P. Rowe, Elliot K. Fishman
Diagnostic and Interventional Imaging, Volume 104, Issue 6,2023, Pages 263-264,
- Background : Existing (artificial intelligence [AI]) tools in radiology are modeled without necessarily considering the expectations and experience of the end user—the radiologist. The literature is scarce on the tangible parameters that AI capabilities need to meet for radiologists to consider them useful tools.
Objective : The purpose of this study is to explore radiologists' attitudes toward AI tools in pancreatic cancer imaging and to quantitatively assess their expectations of these tools.
Methods : A link to the survey was posted on the www.ctisus.com website, advertised in the www.ctisus.com email newsletter, and publicized on LinkedIn, Facebook, and Twitter accounts. This survey asked participants about their demographics, practice, and current attitudes toward AI. They were also asked about their expectations of what constitutes a clinically useful AI tool. The survey consisted of 17 questions, which included 9 multiple choice questions, 2 Likert scale questions, 4 binary (yes/no) questions, 1 rank order question, and 1 free text question.
Radiologists' Expectations of Artificial Intelligence in Pancreatic Cancer Imaging: How Good Is Good Enough?.
Chu, Linda C. MD*; Ahmed, Taha MBBS*; Blanco, Alejandra MD*; Javed, Ammar MD†; Weisberg, Edmund M. MS, MBE*; Kawamoto, Satomi MD*; Hruban, Ralph H. MD‡; Kinzler, Kenneth W. PhD§; Vogelstein, Bert MD§; Fishman, Elliot K. MD*.
Journal of Computer Assisted Tomography July 28, 2023. (in press) - Results: A total of 161 respondents completed the survey, yielding a response rate of 46.3% of the total 348 clicks on the survey link. The minimum acceptable sensitivity of an AI program for the detection of pancreatic cancer chosen by most respondents was either 90% or 95% at a specificity of 95%. The minimum size of pancreatic cancer that most respondents would find an AI useful at detecting was 5 mm. Respondents preferred AI tools that demonstrated greater sensitivity over those with greater specificity. Over half of respondents anticipated incorporating AI tools into their clinical practice within the next 5 years.
Conclusion: Radiologists are open to the idea of integrating AI-based tools and have high expectations regarding the performance of these tools. Consideration of radiologists' input is important to contextualize expectations and optimize clinical adoption of existing and future AI tools.
Radiologists' Expectations of Artificial Intelligence in Pancreatic Cancer Imaging: How Good Is Good Enough?.
Chu, Linda C. MD*; Ahmed, Taha MBBS*; Blanco, Alejandra MD*; Javed, Ammar MD†; Weisberg, Edmund M. MS, MBE*; Kawamoto, Satomi MD*; Hruban, Ralph H. MD‡; Kinzler, Kenneth W. PhD§; Vogelstein, Bert MD§; Fishman, Elliot K. MD*.
Journal of Computer Assisted Tomography July 28, 2023. (in press) - “The minimum acceptable sensitivity of an AI program for the detection of pancreatic cancer chosen by most respondents was either 90% or 95% at a specificity of 95%. The minimum size of pancreatic cancer that most respondents would find an AI useful at detecting was 5 mm. Respondents preferred AI tools that demonstrated greater sensitivity over those with greater specificity. Over half of respondents anticipated incorporating AI tools into their clinical practice within the next 5 years.”
Radiologists' Expectations of Artificial Intelligence in Pancreatic Cancer Imaging: How Good Is Good Enough?.
Chu, Linda C. MD*; Ahmed, Taha MBBS*; Blanco, Alejandra MD*; Javed, Ammar MD†; Weisberg, Edmund M. MS, MBE*; Kawamoto, Satomi MD*; Hruban, Ralph H. MD‡; Kinzler, Kenneth W. PhD§; Vogelstein, Bert MD§; Fishman, Elliot K. MD*.
Journal of Computer Assisted Tomography July 28, 2023. (in press)
- Beyond the new data inputs is the potential to extract far more information from routinely obtained medical images. X-rays, mammograms, and CT images contain a wealth of data, much of which is beyond human perception. Analysing these sources of health data with new AI models presents an opportunity to improve risk stratification and make early disease detection strategies more accurate and efficient. For example, although early trials of lung cancer screening with chest x-rays did not show a mortality benefit, new AI models are now able to predict an individual’s risk of lung cancer from chest x-rays alone, and AI analysis of chest CTs can predict an individual’s 6-year risk of lung cancer, which could be used to guide personalised screening intervals.
Rebooting cancer screening with artificial intelligence.
Adams SJ, Topol EJ.
Lancet. 2023 Aug 5;402(10400):440. - “New data inputs and the ability for machine eyes to see what is not perceptible to humans points towards a potential transformation for cancer screening. AI analysis of multimodal data sources could give rise to a statistical biopsy, offering a comprehensive, personalised approach to screening and early cancer detection. Yet continued development of AI models to efficiently integrate an increasing number of data sources and the validation of AI models in diverse populations, including in randomised controlled trials, will be needed. Looking ahead, health-care systems could harness a shift towards more informative screening to improve efficiency and cost-effectiveness—with improved accuracy and outcomes at the individual and population levels.”
Rebooting cancer screening with artificial intelligence.
Adams SJ, Topol EJ.
Lancet. 2023 Aug 5;402(10400):440.
- Background: Existing (artificial intelligence [AI]) tools in radiology are modeled without necessarily considering the expectations and experience of the end user—the radiologist. The literature is scarce on the tangible parameters that AI capabilities need to meet for radiologists to consider them useful tools.
Objective: The purpose of this study is to explore radiologists' attitudes toward AI tools in pancreatic cancer imaging and to quantitatively assess their expectations of these tools.
Conclusion: Radiologists are open to the idea of integrating AI-based tools and have high expectations regarding the performance of these tools. Consideration of radiologists' input is important to contextualize expectations and optimize clinical adoption of existing and future AI tools. .
Radiologists' Expectations of Artificial Intelligence in Pancreatic Cancer Imaging: How Good Is Good Enough?
Chu LC, Ahmed T, Blanco A, Javed A, Weisberg EM, Kawamoto S, Hruban RH, Kinzler KW, Vogelstein B, Fishman EK.
J Comput Assist Tomogr. 2023 Jul 28. doi: 10.1097/RCT.0000000000001503. Epub ahead of print. - Results: A total of 161 respondents completed the survey, yielding a response rate of 46.3% of the total 348 clicks on the survey link. The minimum acceptable sensitivity of an AI program for the detection of pancreatic cancer chosen by most respondents was either 90% or 95% at a specificity of 95%. The minimum size of pancreatic cancer that most respondents would find an AI useful at detecting was 5 mm. Respondents preferred AI tools that demonstrated greater sensitivity over those with greater specificity. Over half of respondents anticipated incorporating AI tools into their clinical practice within the next 5 years. .
Radiologists' Expectations of Artificial Intelligence in Pancreatic Cancer Imaging: How Good Is Good Enough?
Chu LC, Ahmed T, Blanco A, Javed A, Weisberg EM, Kawamoto S, Hruban RH, Kinzler KW, Vogelstein B, Fishman EK.
J Comput Assist Tomogr. 2023 Jul 28. doi: 10.1097/RCT.0000000000001503. Epub ahead of print. - “In conclusion, this study demonstrates that radiologists are open to the idea of integrating AI-based tools as long as they meet high though probably attainable performance criteria. We believe that, based on continuing progress in the technical capabilities of AI aswell as instrumentation, and the results of this survey, the clinical implementation of AI technology for the detection of pancreatic cancer is aworthy and feasible goal. Future studies should transition toward investigating the preliminary experiences of current radiology AI users to guide further development of AI and to encourage AI adoption among practices not currently using AI tools.” .
Radiologists' Expectations of Artificial Intelligence in Pancreatic Cancer Imaging: How Good Is Good Enough?
Chu LC, Ahmed T, Blanco A, Javed A, Weisberg EM, Kawamoto S, Hruban RH, Kinzler KW, Vogelstein B, Fishman EK.
J Comput Assist Tomogr. 2023 Jul 28. doi: 10.1097/RCT.0000000000001503. Epub ahead of print.
- “Furthermore, in our experience, the environment in which some health care organizations operate often leads these organizations to focus on near-term financial results at the cost of investment in longer-term, innovative forms of technology such as AI. Health care organizations that prioritize innovation link investment decisions to “total mission value,” which includes both financial and nonfinancial factors such as quality improvement, patient safety, patient experience, clinician satisfaction, and increased access to care.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58. - “We think that the need for AI to help improve health care delivery should no longer be questioned, for many reasons. Take the case of the exponential increase in the collective body of medical knowledge required to treat a patient. In 1980, this knowledge doubled every 7 years; in 2010, the doubling period was fewer than 75 days.1 Today, what medical students learn in their first 3 years would be only 6 percent of known medical information at the time of their graduation. Their knowledge could still be relevant but might not always be complete, and some of what they were taught will be outdated. AI has the potential to supplement a clinical team’s knowledge in order to ensure that patients everywhere receive the best care possible. Bringing that potential to reality has not been easy, but thereare some successes.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58. - “AI is broadly defined as a machine or computing platform that is capable of making intelligent decisions. Two types of AI have generally been pursued in health care delivery: machine learning, which involves computational techniques that learn from examples instead of operating from predefined rules, and natural language processing, which is the ability of a computer to transform human language and unstructured text into machine-readable structured data that reliably reflect the intent of the language.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58. - “AI is also being used in prior authorization, a process that involves substantial manual labor, with only 21% of prior authorizations automated.9 The process can be costly because it requires doctors and registered nurses to review requests for authorization. From the payer’s perspective, the objective is to ensure that patients are receiving clinically appropriate treatment. Therefore, prior authorization is meant to be a check on what the provider has ordered.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58. - “Another use of AI in clinical operations is tackling clinician burnout. Physicians now spend more than 50% of their time updating electronic health records (EHRs), and this use of time is a documented contributor to burnout. Multiple providers are piloting natural language processing to reduce this burden. If these efforts are successful, natural language processing could turn unstructured data such as clinicians’ notes into the structured data needed for the EHR as well as for other uses, such as documenting quality metrics or filling in appropriate Current Procedural Terminology codes. This application of AI would give clinicians more time to spend with patients and on tasks that require human judgment.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58. - “For example, starting with a strategic vision, one of the greatest challenges is properly defining the costs and benefits of deploying AI. Historically, the decision to invest in AI has been based on financial return. This calculation should be expanded to include nonfinancial factors as well. Otherwise, AI adoption could continue to lag in certain domains in which a large portion of its effect is nonfinancial, such as quality and safety.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58.
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58.- “I adoption in health care delivery has lagged behind adoption in other business sectors, but the past few years have shown the potential and promise of AI, which has already begun to shape the operations of payers and providers in some areas. If the promise of AI is realized, the quality of and access to health care delivery will be improved. The promise remains, but realizing it in practice has not been easy.”
Artificial Intelligence in U.S. Health Care Delivery
Nikhil R. Sahni and Brandon Carrus
N Engl J Med 2023;389:348-58.
- IMPORTANCE Consumers are increasingly using artificial intelligence (AI) chatbots as a source of information. However, the quality of the cancer information generated by these chatbots has not yet been evaluated using validated instruments.
OBJECTIVE To characterize the quality of information and presence of misinformation about skin, lung, breast, colorectal, and prostate cancers generated by 4 AI chatbots.
CONCLUSIONS AND RELEVANCE Findings of this cross-sectional study suggest that AI chatbots generally produce accurate information for the top cancer-related search queries, but the responses are not readily actionable and are written at a college reading level. These limitations suggest that AI chatbots should be used supplementarily and not as a primary source for medical information.
Assessment of Artificial Intelligence Chatbot Responses to Top Searched Queries About Cancer
Alexander Pan et al.
JAMA Oncol. doi:10.1001/jamaoncol.2023.2947 - “The primary outcomeswere the quality of consumer health information based on the validated DISCERN instrument (scores from 1 [low] to 5 [high] for quality of information) and the understandability and actionability of this information based on the understandability and actionability domains of the Patient Education Materials Assessment Tool (PEMAT) (scores of 0%-100%, with higher scores indicating a higher level of understandability and actionability). Secondary outcomes included misinformation scored using a 5-item Likert scale (scores from 1 [no misinformation] to 5 [high misinformation]) and readability assessed using the Flesch-Kincaid Grade Level readability score.”
Assessment of Artificial Intelligence Chatbot Responses to Top Searched Queries About Cancer
Alexander Pan et al.
JAMA Oncol. doi:10.1001/jamaoncol.2023.2947 - Key Points
Question What is the quality of cancer-related health information outputted by artificial intelligence (AI) chatbots?
Findings In this cross-sectional study, the responses of 4 AI chatbots to the top search queries related to the 5 most prevalent cancers were high quality but were written at a college reading level and had poor actionability.
Meaning Findings of this study suggest that AI chatbots are an accurate and reliable supplementary resource for medical information but are limited in their readability and should not replace health care professionals for individualized health care questions.
Assessment of Artificial Intelligence Chatbot Responses to Top Searched Queries About Cancer
Alexander Pan et al.
JAMA Oncol. doi:10.1001/jamaoncol.2023.2947 - ”Artificial intelligence chatbots are becoming a major source of medical information for consumers. Findings of this crosssectional study suggest that they generally produce reliable and accurate medical information about lung, breast, colorectal, skin, and prostate cancers. However, the usefulness of the information is limited by its poor readability and lack of visual aids. These limitations suggest that AI chatbots should be used supplementarily and not as a primary source for medical information. To this end, AI chatbots typically encourage users to seek medical attention relating to cancer symptoms and treatment.”
Assessment of Artificial Intelligence Chatbot Responses to Top Searched Queries About Cancer
Alexander Pan et al.
JAMA Oncol. doi:10.1001/jamaoncol.2023.2947 - Large language models (LLMs) can respond to free-text queries without being specifically trained in the task in question, causing excitement and concern about their use in healthcare settings. ChatGPT is a generative artificial intelligence (AI) chatbot produced through sophisticated fine-tuning of an LLM, and other tools are emerging through similar developmental processes. Here we outline how LLM applications such as ChatGPT are developed, and we discuss how they are being leveraged in clinical settings. We consider the strengths and limitations of LLMs and their potential to improve the efficiency and effectiveness of clinical, educational and research work in medicine.
Large language models in medicine
Arun James Thirunavukarasu et al.
Nature Medicine https://doi.org/10.1038/s41591-023-02448-8 - ChatGPT (OpenAI) is an LLM chatbot: a generative AI application that now produces text in response to multimodal input (having previously accepted only text input). Its backend LLM is Generative Pretrained Transformer 3.5 or 4 (GPT-3.5 or GPT-4), described below. ChatGPT’s impact stems from its conversational interactivity and near-human-level or equal-to-human-level performance in cognitive tasks across fields, including medicine. ChatGPT has attained passing-level performance in United States Medical Licensing Examinations, and there have been suggestions that LLM applications may be ready for use in clinical, educational or research settings.
Large language models in medicine
Arun James Thirunavukarasu et al.
Nature Medicine https://doi.org/10.1038/s41591-023-02448-8 - Deep learning: a variant of machine learning involving neural networks with multiple layers of processing ‘perceptrons’ (nodes), which together facilitate extraction of higher features of unstructured input data (for example, images, video and text).
Generative artificial intelligence: computational systems capable of producing content, such as text, images or sound, on demand.
Large language model: a type of AI model using deep neural networks to learn the relationships between words in natural language, using large datasets of text to train.
Machine learning: a field of AI featuring models that enable computers to learn and make predictions based on input data, learning from experience.
Natural language processing: a field of AI research focusing on the interaction between computers and human language.
Neural network: computing systems inspired by biological neural networks, comprising ‘perceptrons’ (nodes), usually arranged in layers, communicating with one another and performing transformations upon input data.
- ”AI-supported mammography screening resulted in a similar cancer detection rate compared with standard double reading, with a substantially lower screen-reading workload, indicating that the use of AI in mammography screening is safe. The trial was thus not halted and the primary endpoint of interval cancer rate will be assessed in 100 000 enrolled participants after 2-years of follow up.”
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44 - “The results from this randomised trial support the findings of earlier retrospective studies, indicating a general potential of AI to improve screening efficacy and reduce workload. The clinical safety analysis concludes that the AI-supported screen-reading procedure can be considered safe. Implementation of AI in clinical practice to reduce the screen-reading workload could therefore be considered to help address workforce shortages. The assessment of the primary endpoint of interval cancer rate, together with a characterisation of detected cancers in the entire study population, will provide further insight into the efficacy of screening, possible side-effects such as overdiagnosis, and the prognostic implications of using AI in mammography screening, taking cost-effectiveness into account.”
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44 - To our knowledge, this is the first randomised controlled trial investigating the use of AI in mammography screening. In this first report, the objective was to assess the safety of an AI-supported screen-reading procedure, involving triage and detection support. AI-supported screening resulted in 20% more cancers being detected and exceeded the lowest acceptable limit for safety compared with standard double reading without AI, without affecting the false positive rate. The AI supported screen-reading procedure enabled a 44·3% reduction in the screen-reading workload. The results indicate that the proposed screening strategy is safe.
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44 - “In summary, this clinical safety analysis of the MASAI trial, in which an AI system was used to triage screening examinations to single or double reading and as detection support, showed that AI-supported mammography screening can be considered safe, since it resulted in a similar rate of screen-detected cancer—exceeding the lowest acceptable limit for safety—without increasing rates of recalls, false positives, or consensus meetings, and while substantially reducing the screen-reading workload compared with screening by means of standard double reading.”
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trial (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
Kristina Lång et al.
Lancet Oncol 2023; 24: 936–44
- “AI for Good, a multimillion dollar philanthropic initiative from Microsoft, highlights the potential of AI and aims to help and empower those working around the world to solve issues related to five pillars: Earth, Accessibility, Humanitarian Action, Cultural Heritage, and Health. AI for Health is a $40 million investment made over 5 years with the goal of empowering researchers and organizations to use AI to advance the health of people and communities around the world. We also dedicated $20 million to help those on the front lines of research of COVID-19 through the AI for Health program.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “ For some world problems, relying on AI is the only option we have. Diabetic retinopathy (DR) is the leading cause of vision loss globally, and out of the 451 million people worldwide with diabetes, a third of these people will develop DR. If treated, blindness due to DR is completely preventable, but the problem is that for many individuals, a proper diagnosis of DR is impossible given the fact that there are only 200,000 ophthalmologists in the world. Thankfully, AI models have >97% accuracy (on par with ophthalmologists) in detecting DR [7], compensating for the lack of physicians capable of making this diagnosis. However, despite AI’s promising potential, it comes with challenges and limitations of its own.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “AI expertise alone cannot solve these problems; we need to collaborate with subject matter experts. Machine learning excels at prediction and correlation, but not at identifying causation. It does not know the direction of the causality.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “We can be fooled by bias. In 1991, a study published in the New England Journal of Medicine found that left-handed people died 9 years younger than righthanded people. However, the study failed to consider that there used to be bias against lefthanded people and many of them were forced to become right-handed. Researchers had assumed that the percentage of left-handed people is stable over time; the population, although random, is biased against lefthanded people. Most data we collect have biases, and if we do not understand them and take them into account, our data models will not be correct.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “We forget that correlation or predictive power does not imply causation; the fact that two variables are correlated does not imply that one causes the other. In a Gallup poll several years ago, surveyors asked participants if correlation implied causation, and 64% of Americans answered yes. This occurs because humans learn from correlation, but we cannot observe causality. We have to understand that most people do not know the difference.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Models are very good at cheating; if there is anything the model can use to cheat, it will learn it. An AI model was trained to distinguish between skin cancer and benign lesions and was thought to achieve dermatologist-level performance. However, many of the positive cases had a ruler in the picture but the negative cases did not. The model learned that if there was a ruler present in the image, there was a much higher chance of the patient having cancer.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Access to data is one of the biggest challenges we face. There is a significant amount of data that cannot be open to researchers because of privacy issues, especially in the medical world. In most scenarios, data anonymization just does not work because even if we anonymize the data, there is always the risk of someone attempting to deanonymize it. As a result, Microsoft has invested in the Differential Privacy Platform, which provides a way for researchers to ascertain insights from the data without violating the privacy of individuals. Privacy-preserving synthetic images can also generate realistic synthetic data, including synthetic medical images, after training on a real data set, without affecting the privacy of individuals.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Other technology companies such as Google Health, NVIDIA, and Amazon have also invested in strategic partnerships with health care organizations to combine their expertise in AI with our health care domain knowledge to solve impactful clinical problems. These collaborations leverage the power of big data analytics and have immense potential to improve cancer detection, predict patient outcomes, and reduce health equity by improving patient access.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Contrary to popular media speculation, AI alone is not enough to overcome the problems that society seeks to resolve; rather, machine learning depends on subject-matter experts to find solutions. AI provides us with numerous opportunities for advancement in the field of radiology: improved diagnostic certainty, suspicious case identification for early review, better patient prognosis, and a quicker turnaround. Machine learning depends on radiologists and our expertise, and the convergence of radiologists and AI will bring forth the best outcomes for patients.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print.
- “Artificial Intelligence (AI) is set to transform medical imaging by leveraging the vast data contained in medical images. Deep learning and radiomics are the two main AI methods currently being applied within radiology. Deep learning uses a layered set of self-correcting algorithms to develop a mathematical model that best fits the data. Radiomics converts imaging data into mineable features such as signal intensity, shape, texture, and higher-order features. Both methods have the potential to improve disease detection, characterization, and prognostication.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - ”The number of features extracted during this image analysis process can vary widely, depending on the software package and filters used. However, a high number of features and a low number of cases in a group for a classification task can result in overfitting of the model. To mitigate this risk, it is essential to perform feature selection or dimension reduction to reduce the number of features and increase the validity and generalizability of the results. Once appropriate features have been selected, they are subsequently analyzed with advanced machine learning algorithms, such as random forest or support vector machine, to perform specific classification tasks that can be used to help answer clinical questions.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “While most prior studies had applied radiomics as a "second reader" to catch a diagnosis that may be missed due to human error, several more recent studies have reported that radiomics models may be able to detect PDAC before it is even discernable to the human eye on imaging . During the development of PDAC, the pancreas undergoes various morphological changes. PDACs may arise from detectable precancerous lesions such as intraductal papillary mucinous neoplasms (IPMN), and the pancreatic parenchyma upstream from a subtle cancer may show focal parenchymal atrophy and changes of chronic pancreatitis. Each of these can gradually increase the heterogeneity of the pancreatic tissue and result in detectable morphological and textural changes. These alterations may be difficult to interrogate on visual assessment, making AI the ideal tool to analyze them.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “AI has also made advancements in the detection of a variety of solid and cystic pancreatic neoplasms aside from PDAC. In a recent study, a deep learning model was able to detect PDAC, pancreatic neuroendocrine tumors (pNET), solid pseudopapillary neoplasms (SPN), mucinous cystic neoplasms (MCN), serous cystic neoplasms, and IPMNs with a sensitivity of 98%−100% for solid lesions and 92%–93% for cystic lesions larger than 1.0 cm across two test sets consisting of 1192 patients. The performance of this model was not significantly different from that of radiologists (95–100% for solid lesions and 93–98% for cystic lesions > 1.0 cm). Similar prior deep learning studies have reported the sensitivity of detecting PDAC, pNET and pancreatic cystic lesions ranging from 78.8% to 87.6%.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “Studies on classification of cystic pancreatic tumors have also reported excellent results. These studies have predominantly employed three different strategies in cyst classification: 1), A multi-class method to distinguish each category of pancreatic cyst. 2), A binary approach to separate benign cysts from those with malignant potential. 3), A binary classification of mucin-producing cysts into high-grade or low-grade dysplasia. A recent multiclass study consisting of 214 patients reported a radiomics model to perform on par with experienced academic radiologists at classifying various cystic tumors (IPMN, MCN, serous cystadenoma, SPN, PNET) with an AUC of 0.940 for the radiomics model compared to an AUC of 0.895 for the radiologists.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “The third category of radiomics-based cyst classification studies have attempted to fill this gap through applying radiomics to risk-stratify patients with IPMNs. Studies have found that radiomics features extracted from CT, MRI and ultrasound have superior ability in identifying high-risk vs low-risk disease compared to clinical features and/or Fukuoka criteria. The most recent of such studies enrolled 66 patients and compared both MRI and CT radiomics models. In this study, the MRI model outperformed the CT radiomics model and achieved an AUC of 0.940 in preoperatively predicting malignant potential of IPMNs . Numerous similar prior studies with MRI or CT radiomics model have been conducted, and despite methodological variation, these studies have reported comparably strong results (AUC range, 0.71–0.96). Prior studies have also integrated clinical models based on the Fukuoka guidelines with radiomics models and have demonstrated superior performance of the combined models . Future integration of additional multidimensional data, such as novel radiogenomic features of cyst fluid DNA, into machine learning models has potential to further improve upon the performance of existing models.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “A unique application of radiomics beyond diagnostics has been to predict preoperatively a patient's survival should they undergo surgical resection of PDAC. Survival prediction of PDAC currently relies mainly on postoperative features such as the TNM stage and margins. This precludes preoperative survival prediction which could separate patients who will benefit from surgery from those who will not. Although surgical resection remains the only cure for PDAC, it is associated with significant complications and carries a small mortality risk. The development of an accurate preoperative survival prediction model could allow for a quantitative risk-benefit analysis prior to pancreatic resection and allow for individualized triaging of patients for surgery based on overall anticipated benefit from resection. In the current literature, four studies directly compared the prognostic performance of radiomics models with the clinical TNM staging criteria, with all four studies reporting that the radiomics models outperformed clinical criteria in predicting overall survival.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “Neoadjuvant therapy for PDAC is associated with lower rates of post-operative nodal involvement and perineural invasion, and higher rates of negative margin resection. However, determining the response to neoadjuvant treatment and resectability can be difficult. Radiomics have demonstrated the potential to identify a rapid response to chemotherapy and early down-staging to a surgically resectable tumor by evaluating the longitudinal evolution of radiomic features over chemoradiation cycles, termed delta radiomic features. Nasief et al. notably showed that CT delta radiomics features, particularly skewness and kurtosis (a measure of the shape of the distribution), could differentiate good responders from poor responders of chemoradiation therapy in a validation cohort of 40 patients with an AUC of 0.94 .”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “When discussing the quality and generalizability of current evidence, perhaps the biggest challenge that needs to be overcome is the consolidation of large public annotated datasets. Radiomics and deep learning need to be trained on large datasets, and model performance and generalizability are critically dependent on the quality and size of these datasets. While the existence of such datasets had been previously limited, efforts to grow them are underway, with a notable dataset such as Imagenet and the National Cancer Institute's The Cancer Imaging Archive already being used by one included study to externally validate their model. In addition, efforts to synthetically augment datasets through deep learning methods such as neural style transfer and generative adversarial networks, which garnered public interest due to its application in "deepfake" media, have also demonstrated potential, but their current utility is uncertain. Although efforts to develop these databases and augment datasets are ongoing, their utilization in the development and validation of recent radiomics studies remains limited, as highlighted by the low overall RQS for validation amongst included studies. ”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - ”Beyond quality and generalizability issues, several practical barriers also exist, with one being the determination of the added clinical value of these models. To truly gauge a model's clinical utility, it must be compared against existing gold standards. For studies reporting models that autonomously detect and classify pancreatic lesions, the gold standard remains the radiologists' reads, while among studies reporting models that predict patient survival, the gold standard is TNM staging. Amongst included studies, only 31.5% reported a comparison with these gold standards making interpretation of the net clinical benefit of most current models questionable. While the vast majority (98.1%) of reviewed studies discussed potential clinical applications, simply suggesting prospective methods in which AI models may be of value is no longer adequate and future studies should report objective metrics such as incremental value over gold standards or decision curve analyses alongside their models.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “The final and arguably most formidable barrier to AI adoption is that of the legal hurdles associated with AI use in healthcare. The current legal framework for computer aided detection tools and for AI in triaging (e.g., AIdoc) is cloudy, and how this approach translates to the next generation of increasingly autonomous and diagnostic AI is uncertain. Mistakes are inevitable and consideration should be given to the difficult questions that will arise when these mistakes happen. Who is responsible for the accuracy of an autonomous AI system when it makes an error? How do we factor in the radiologist's liability when using AI tools? What is the liability of the health system that purchases an AI product? It has been suggested that analysis of how lawsuits involving autonomous cars, which share certain similarities with medical AI tools, have been handled by the courts could be instructive in providing a legal framework for medical AI. ”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - AI has made significant progress in the detection, classification, and prognostication of pancreatic lesions, through techniques such as radiomics and deep learning. Despite this promise, the quality of existing literature is far from robust. While we acknowledge that the potential value of existing literature may extend beyond what may be formally evaluated through the RQS tool, to fully realize the benefit of these advancements, current results need to be validated through higher quality studies and multicenter trials that include the full spectrum of normal and abnormal. Fundamental questions still need addressing before clinical adoption, and efforts to establish sound evidence for future studies is warranted. Given the rate of discovery of AI in abdominal imaging however, we optimistically believe that these challenges will inevitably be overcome and that a future in which synergy between radiologists and machines will become the norm is not a matter of ‘if’ but only a matter of ‘when’.
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - Radiomics and Pancreatic Cancer
1. early lesion detection
2. classification of pancreatic tumors
3. risk stratification of masses (IPMN)
4. prediction of tumor grade
5. survival prediction
6. treatment response and surgical resectability - “Stopping surveillance after stable cyst imaging for 5 years is a recommendation of the AGA guidelines, and after 10 years of stability in low-risk cysts, or sooner if the patients reaches 80 years of age after stability, or until the patients is no longer a surgical candidate is the recommendations of the ACR. This issue of ceasing surveillance isnot addressed in the other PCN clinical guidelines.The ACG guidelines recommend that patients fit for surgery should continue surveillance until they are no longer surgical candidates, and that patients older than 75 years should undergo surveillance imaging only after discussion with the multidisciplinary team.”
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640 - A greater understanding of the biology and natural history of progression of pancreatic cysts is needed to improve our PCN surveillance strategies. This in turn may permit the development and validation of a blood-based approach for pancreatic cyst diagnosis and stratification, as well as refining pancreatic cyst fluid biomarkers for prediction of natural history purposes. Further studies on the role of chemoprevention andeven PCN ablation to alter the natural history of PCNs may impact on how we survey these patients Additional prospective studies such as the ACRIN-ECOG 2185 which is a prospective randomized controlled trial comparing a high-intensity surveillance program with a low-intensity testing program will provide very valuable, needed, and detailed clinical outcome information on pancreatic cysts surveillance, which will allow for a more reasoned discussions about the intensity of surveillance, use of valuable resources, and when to consider stopping surveillance.
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640 - “It is increasingly appreciated that new-onset diabetes, especially type 2 diabetes mellitus (DM), is a risk factor for the development of PDAC. Data including a recent large meta-analysis demonstrate the association between the development of DM and both the morphologic progression of pancreatic cysts as well as the development of cancer. Some clinical guidelines have incorporated new-onset DM as a WF necessitating closer imaging and surveillance.”
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640 - “A combination of significant family history of pancreatic cancer (defined as greater than 2 affected family members with PDAC) and certain inherited germline mutations is known to be associated an increased risk of PDAC. The exact interplay between family history, germline genetics, and PCNs still remains to be clarified. Some studies suggest that a family history of pancreatic cancer and germline mutations are associated with a higher risk of morphologic progression and cancer risk in pancreatic cysts, hence justifying closer and prolonged surveillance. However,other studies have not been able to demonstrate a strong correlation to justify a change in cyst surveillance based on a limited family history of PDAC.”
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640
Esophagus
- “Gastrointestinal (GI) lymphoma accounts for 5–20% of extranodal lymphomas: the stomach is the most common site, followed by small intestine (ileum (60–65%), jejunum (20%−25%), and duodenum (6%–8%) and then colorectal lymphomas (6–12%)). GI lymphomas most commonly occur around the sixth decade of life and, although rare in childhood, they are the most common GI tumours in this age.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Primary esophageal lymphomas account for less than 1% in all primary GI lymphomas, while usually result from lymph node metastasis of the lymphomas from the cervical or mediastinal region. Both findings on barium studies, as irregular filling defects, and on CT, as thickened esophageal wall with narrowed lumen, are nonspecific and mimic esophageal adenocarcinoma. However, CT may be useful to differentiate primary esophageal lymphoma from lymph node involvements in the cervical or mediastinal regions, in staging of the disease and in evaluating response to therapy.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143
Liver
- “An inflammatory pseudotumor of the liver is an uncommon, benign, tumor-like lesion that sometimes mimics a malignant tumor, particularly metastatic disease or cholangiocarcinoma.9 In the majority of cases, including the case presented by Rosa and associates, these inflammatory pseudotumors are most likely inflammatory or infectious in origin.1 The lesions often appear to develop from a healing abscess or an inflammatory condition resulting from rupture of the bile duct and extravasation of bile into the tissue, which provokes a xanthogranulomatous inflammatory response that heals with scarring.”
Inflammatory pseudotumor of the liver: a rare but distinct tumor-like lesion.
Balabaud C, Bioulac-Sage P, Goodman ZD, Makhlouf HR.
Gastroenterol Hepatol (N Y). 2012 Sep;8(9):633-4. - “Imaging studies of inflammatory pseudotumors have revealed hypoechoic, as well as hyperechoic, masses on ultrasound. Unenhanced computed tomography images show the tumor’s hypoattenuating liver parenchyma, and contrast administration reveals variable patterns of enhancement. On magnetic resonance imaging, the lesions are typically hypointense on T1-weighted images and hyperintense on T2-weighted images, with variable enhancement patterns”
Inflammatory pseudotumor of the liver: a rare but distinct tumor-like lesion.
Balabaud C, Bioulac-Sage P, Goodman ZD, Makhlouf HR.
Gastroenterol Hepatol (N Y). 2012 Sep;8(9):633-4.
Pancreas
- “Artificial Intelligence (AI) is set to transform medical imaging by leveraging the vast data contained in medical images. Deep learning and radiomics are the two main AI methods currently being applied within radiology. Deep learning uses a layered set of self-correcting algorithms to develop a mathematical model that best fits the data. Radiomics converts imaging data into mineable features such as signal intensity, shape, texture, and higher-order features. Both methods have the potential to improve disease detection, characterization, and prognostication.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - ”The number of features extracted during this image analysis process can vary widely, depending on the software package and filters used. However, a high number of features and a low number of cases in a group for a classification task can result in overfitting of the model. To mitigate this risk, it is essential to perform feature selection or dimension reduction to reduce the number of features and increase the validity and generalizability of the results. Once appropriate features have been selected, they are subsequently analyzed with advanced machine learning algorithms, such as random forest or support vector machine, to perform specific classification tasks that can be used to help answer clinical questions.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “While most prior studies had applied radiomics as a "second reader" to catch a diagnosis that may be missed due to human error, several more recent studies have reported that radiomics models may be able to detect PDAC before it is even discernable to the human eye on imaging . During the development of PDAC, the pancreas undergoes various morphological changes. PDACs may arise from detectable precancerous lesions such as intraductal papillary mucinous neoplasms (IPMN), and the pancreatic parenchyma upstream from a subtle cancer may show focal parenchymal atrophy and changes of chronic pancreatitis. Each of these can gradually increase the heterogeneity of the pancreatic tissue and result in detectable morphological and textural changes. These alterations may be difficult to interrogate on visual assessment, making AI the ideal tool to analyze them.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “AI has also made advancements in the detection of a variety of solid and cystic pancreatic neoplasms aside from PDAC. In a recent study, a deep learning model was able to detect PDAC, pancreatic neuroendocrine tumors (pNET), solid pseudopapillary neoplasms (SPN), mucinous cystic neoplasms (MCN), serous cystic neoplasms, and IPMNs with a sensitivity of 98%−100% for solid lesions and 92%–93% for cystic lesions larger than 1.0 cm across two test sets consisting of 1192 patients. The performance of this model was not significantly different from that of radiologists (95–100% for solid lesions and 93–98% for cystic lesions > 1.0 cm). Similar prior deep learning studies have reported the sensitivity of detecting PDAC, pNET and pancreatic cystic lesions ranging from 78.8% to 87.6%.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “Studies on classification of cystic pancreatic tumors have also reported excellent results. These studies have predominantly employed three different strategies in cyst classification: 1), A multi-class method to distinguish each category of pancreatic cyst. 2), A binary approach to separate benign cysts from those with malignant potential. 3), A binary classification of mucin-producing cysts into high-grade or low-grade dysplasia. A recent multiclass study consisting of 214 patients reported a radiomics model to perform on par with experienced academic radiologists at classifying various cystic tumors (IPMN, MCN, serous cystadenoma, SPN, PNET) with an AUC of 0.940 for the radiomics model compared to an AUC of 0.895 for the radiologists.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “The third category of radiomics-based cyst classification studies have attempted to fill this gap through applying radiomics to risk-stratify patients with IPMNs. Studies have found that radiomics features extracted from CT, MRI and ultrasound have superior ability in identifying high-risk vs low-risk disease compared to clinical features and/or Fukuoka criteria. The most recent of such studies enrolled 66 patients and compared both MRI and CT radiomics models. In this study, the MRI model outperformed the CT radiomics model and achieved an AUC of 0.940 in preoperatively predicting malignant potential of IPMNs . Numerous similar prior studies with MRI or CT radiomics model have been conducted, and despite methodological variation, these studies have reported comparably strong results (AUC range, 0.71–0.96). Prior studies have also integrated clinical models based on the Fukuoka guidelines with radiomics models and have demonstrated superior performance of the combined models . Future integration of additional multidimensional data, such as novel radiogenomic features of cyst fluid DNA, into machine learning models has potential to further improve upon the performance of existing models.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “A unique application of radiomics beyond diagnostics has been to predict preoperatively a patient's survival should they undergo surgical resection of PDAC. Survival prediction of PDAC currently relies mainly on postoperative features such as the TNM stage and margins. This precludes preoperative survival prediction which could separate patients who will benefit from surgery from those who will not. Although surgical resection remains the only cure for PDAC, it is associated with significant complications and carries a small mortality risk. The development of an accurate preoperative survival prediction model could allow for a quantitative risk-benefit analysis prior to pancreatic resection and allow for individualized triaging of patients for surgery based on overall anticipated benefit from resection. In the current literature, four studies directly compared the prognostic performance of radiomics models with the clinical TNM staging criteria, with all four studies reporting that the radiomics models outperformed clinical criteria in predicting overall survival.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “Neoadjuvant therapy for PDAC is associated with lower rates of post-operative nodal involvement and perineural invasion, and higher rates of negative margin resection. However, determining the response to neoadjuvant treatment and resectability can be difficult. Radiomics have demonstrated the potential to identify a rapid response to chemotherapy and early down-staging to a surgically resectable tumor by evaluating the longitudinal evolution of radiomic features over chemoradiation cycles, termed delta radiomic features. Nasief et al. notably showed that CT delta radiomics features, particularly skewness and kurtosis (a measure of the shape of the distribution), could differentiate good responders from poor responders of chemoradiation therapy in a validation cohort of 40 patients with an AUC of 0.94 .”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “When discussing the quality and generalizability of current evidence, perhaps the biggest challenge that needs to be overcome is the consolidation of large public annotated datasets. Radiomics and deep learning need to be trained on large datasets, and model performance and generalizability are critically dependent on the quality and size of these datasets. While the existence of such datasets had been previously limited, efforts to grow them are underway, with a notable dataset such as Imagenet and the National Cancer Institute's The Cancer Imaging Archive already being used by one included study to externally validate their model. In addition, efforts to synthetically augment datasets through deep learning methods such as neural style transfer and generative adversarial networks, which garnered public interest due to its application in "deepfake" media, have also demonstrated potential, but their current utility is uncertain. Although efforts to develop these databases and augment datasets are ongoing, their utilization in the development and validation of recent radiomics studies remains limited, as highlighted by the low overall RQS for validation amongst included studies. ”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - ”Beyond quality and generalizability issues, several practical barriers also exist, with one being the determination of the added clinical value of these models. To truly gauge a model's clinical utility, it must be compared against existing gold standards. For studies reporting models that autonomously detect and classify pancreatic lesions, the gold standard remains the radiologists' reads, while among studies reporting models that predict patient survival, the gold standard is TNM staging. Amongst included studies, only 31.5% reported a comparison with these gold standards making interpretation of the net clinical benefit of most current models questionable. While the vast majority (98.1%) of reviewed studies discussed potential clinical applications, simply suggesting prospective methods in which AI models may be of value is no longer adequate and future studies should report objective metrics such as incremental value over gold standards or decision curve analyses alongside their models.”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - “The final and arguably most formidable barrier to AI adoption is that of the legal hurdles associated with AI use in healthcare. The current legal framework for computer aided detection tools and for AI in triaging (e.g., AIdoc) is cloudy, and how this approach translates to the next generation of increasingly autonomous and diagnostic AI is uncertain. Mistakes are inevitable and consideration should be given to the difficult questions that will arise when these mistakes happen. Who is responsible for the accuracy of an autonomous AI system when it makes an error? How do we factor in the radiologist's liability when using AI tools? What is the liability of the health system that purchases an AI product? It has been suggested that analysis of how lawsuits involving autonomous cars, which share certain similarities with medical AI tools, have been handled by the courts could be instructive in providing a legal framework for medical AI. ”
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - AI has made significant progress in the detection, classification, and prognostication of pancreatic lesions, through techniques such as radiomics and deep learning. Despite this promise, the quality of existing literature is far from robust. While we acknowledge that the potential value of existing literature may extend beyond what may be formally evaluated through the RQS tool, to fully realize the benefit of these advancements, current results need to be validated through higher quality studies and multicenter trials that include the full spectrum of normal and abnormal. Fundamental questions still need addressing before clinical adoption, and efforts to establish sound evidence for future studies is warranted. Given the rate of discovery of AI in abdominal imaging however, we optimistically believe that these challenges will inevitably be overcome and that a future in which synergy between radiologists and machines will become the norm is not a matter of ‘if’ but only a matter of ‘when’.
A primer on artificial intelligence in pancreatic imaging.
Ahmed TM, Kawamoto S, Hruban RH, Fishman EK, Soyer P, Chu LC.
Diagn Interv Imaging. 2023 Mar 24:S2211-5684(23)00050-5. - Radiomics and Pancreatic Cancer
1. early lesion detection
2. classification of pancreatic tumors
3. risk stratification of masses (IPMN)
4. prediction of tumor grade
5. survival prediction
6. treatment response and surgical resectability - “Stopping surveillance after stable cyst imaging for 5 years is a recommendation of the AGA guidelines, and after 10 years of stability in low-risk cysts, or sooner if the patients reaches 80 years of age after stability, or until the patients is no longer a surgical candidate is the recommendations of the ACR. This issue of ceasing surveillance isnot addressed in the other PCN clinical guidelines.The ACG guidelines recommend that patients fit for surgery should continue surveillance until they are no longer surgical candidates, and that patients older than 75 years should undergo surveillance imaging only after discussion with the multidisciplinary team.”
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640 - A greater understanding of the biology and natural history of progression of pancreatic cysts is needed to improve our PCN surveillance strategies. This in turn may permit the development and validation of a blood-based approach for pancreatic cyst diagnosis and stratification, as well as refining pancreatic cyst fluid biomarkers for prediction of natural history purposes. Further studies on the role of chemoprevention andeven PCN ablation to alter the natural history of PCNs may impact on how we survey these patients Additional prospective studies such as the ACRIN-ECOG 2185 which is a prospective randomized controlled trial comparing a high-intensity surveillance program with a low-intensity testing program will provide very valuable, needed, and detailed clinical outcome information on pancreatic cysts surveillance, which will allow for a more reasoned discussions about the intensity of surveillance, use of valuable resources, and when to consider stopping surveillance.
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640 - “It is increasingly appreciated that new-onset diabetes, especially type 2 diabetes mellitus (DM), is a risk factor for the development of PDAC. Data including a recent large meta-analysis demonstrate the association between the development of DM and both the morphologic progression of pancreatic cysts as well as the development of cancer. Some clinical guidelines have incorporated new-onset DM as a WF necessitating closer imaging and surveillance.”
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640 - “A combination of significant family history of pancreatic cancer (defined as greater than 2 affected family members with PDAC) and certain inherited germline mutations is known to be associated an increased risk of PDAC. The exact interplay between family history, germline genetics, and PCNs still remains to be clarified. Some studies suggest that a family history of pancreatic cancer and germline mutations are associated with a higher risk of morphologic progression and cancer risk in pancreatic cysts, hence justifying closer and prolonged surveillance. However,other studies have not been able to demonstrate a strong correlation to justify a change in cyst surveillance based on a limited family history of PDAC.”
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640
- In general, all suspected MCNs exceeding 4 cm in size and any suspected MCN which has a mural nodule or is symptomatic should undergo resection. Surveillance can be pursued for presumed MCNs less than 3 cm in size in the absence of WFs or HRS that characterize indications for resection in IPMN disease. In 3- to 4-cm lesions suspected to be MCN, the risks of empiric resection with a low likelihood of finding invasive disease should be balanced with the need for long-term surveillance. As the natural history of MCN growth is not well understood, adherence to a surveillance regimen is important; this can be quite prolonged as a diagnosis typically occurs in the fourth or fifth decade of life.
Surveillance of Pancreatic Cystic Neoplasms
Ankit Chhoda, Julie Schmidt, James J. Farrell
Gastrointest Endoscopy Clin N Am 33 (2023) 613–640
Practice Management
- “Successful program building and leadership are not easy, and, perhaps, at times, Dr. Cameron was a bit rougher than he needed to be. However, to make diamonds, you need to apply pressure. His guiding lights were the reflections of great leaders: “Always do the right thing,” “Good judgment comes from experience and experience comes from bad judgment”, and, “If you pick a profession you love, you will never work a day in your life”. Leadership is a responsibility and an honor. It was important for him to treat my faculty as though he was working for them, rather than them working for him. Each of us has unique personalities, and when you look at world leaders such as Churchill, Franklin Roosevelt, or Truman, their leadership was partly based on their personalities and partly influenced by situation.”
An Approach to Leadership in Academic Medicine: Lessons Learned From the Experience of Dr. John L. Cameron,
Steven P. Rowe, Elliot K. Fishman, Linda C. Chu, Pamela T. Johnson, John L. Cameron
Current Problems in Diagnostic Radiology, Volume 52, Issue 5, 2023,Pages 313-314 - “In the era of film radiography, clinical teams would routinely come to the reading room at various points during the day to review their patients’ cases. Those interactions enabled us to build meaningful relationships with referring clinicians and play an active role in patient care through discussions about the details of the patients’ conditions, which were critical in guiding our interpretations. Digital technology eliminated daily trips to radiology for many providers.”
An Approach to Leadership in Academic Medicine: Lessons Learned From the Experience of Dr. John L. Cameron,
Steven P. Rowe, Elliot K. Fishman, Linda C. Chu, Pamela T. Johnson, John L. Cameron
Current Problems in Diagnostic Radiology, Volume 52, Issue 5, 2023,Pages 313-314 - “In the modern era, we can avail ourselves of technology such as Zoom or shared augmented reality platforms in order to expand our reach. However, we should also remember that virtual meetings and augmented reality are imperfect substitutions for the face-to-face presence of an engaged radiologist. Dr Cameron always emphasized the need to sit next to the radiologist and review key aspects of the imaging. When clinical teams are present in the reading room with at least 1 radiologist who acknowledges the importance of their questions and believes in a multidisciplinary approach, a number of important things can happen. First, we should avail ourselves of the opportunity to talk through therapeutic approaches to rare diseases and correlate those with imaging patterns. The constructive feedback between the clinical and imaging teams can be decidedly important, but is also not reimbursed and must be supported by the administration in the individual departments.”
An Approach to Leadership in Academic Medicine: Lessons Learned From the Experience of Dr. John L. Cameron,
Steven P. Rowe, Elliot K. Fishman, Linda C. Chu, Pamela T. Johnson, John L. Cameron
Current Problems in Diagnostic Radiology, Volume 52, Issue 5, 2023,Pages 313-314 - “Make diamonds.” The national focus on physician burnout has fueled a resurgence of programs to improve joy in medicine. However, the greatest joy can come from providing clinical excellence and contributing to new discoveries through research. High expectations should be set for students, trainees, and junior faculty. Indeed, it is perhaps lowering expectations that lead to dissatisfaction and burn-out. Expecting excellence may be the lightning rod many of us need.”
An Approach to Leadership in Academic Medicine: Lessons Learned From the Experience of Dr. John L. Cameron,
Steven P. Rowe, Elliot K. Fishman, Linda C. Chu, Pamela T. Johnson, John L. Cameron
Current Problems in Diagnostic Radiology, Volume 52, Issue 5, 2023,Pages 313-314 - “Departments of radiology can help avoid burn-out and dissatisfaction, while at the same time leveraging significant academic contributions from their faculty. Early-career physician-scientist incubator programs and mid-to-late-career bridge funding can either start or maintain research programs that may contribute substantively to a department's long-term grant funding and discovery-based output. Talented investigators should not be left struggling for funding between grant cycles, but must also be accountable for utilizing “soft money” from the department in a productive way. Further, junior faculty should be encouraged to take on certain organs and diseases as their “own,” subsequently embedding themselves in tumor boards and multidisciplinary conferences where they can be valuable assets; however, departments must also be ready to absorb the cost to protect their faculties’ time to engage in those activities. Additionally, senior investigators can help bring along their junior colleagues, partially by helping them navigate departmental administrative hurdles, but also by making sure that they remain academically productive in ways that advance their careers and the field.”
An Approach to Leadership in Academic Medicine: Lessons Learned From the Experience of Dr. John L. Cameron,
Steven P. Rowe, Elliot K. Fishman, Linda C. Chu, Pamela T. Johnson, John L. Cameron
Current Problems in Diagnostic Radiology, Volume 52, Issue 5, 2023,Pages 313-314 - “AI for Good, a multimillion dollar philanthropic initiative from Microsoft, highlights the potential of AI and aims to help and empower those working around the world to solve issues related to five pillars: Earth, Accessibility, Humanitarian Action, Cultural Heritage, and Health. AI for Health is a $40 million investment made over 5 years with the goal of empowering researchers and organizations to use AI to advance the health of people and communities around the world. We also dedicated $20 million to help those on the front lines of research of COVID-19 through the AI for Health program.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “ For some world problems, relying on AI is the only option we have. Diabetic retinopathy (DR) is the leading cause of vision loss globally, and out of the 451 million people worldwide with diabetes, a third of these people will develop DR. If treated, blindness due to DR is completely preventable, but the problem is that for many individuals, a proper diagnosis of DR is impossible given the fact that there are only 200,000 ophthalmologists in the world. Thankfully, AI models have >97% accuracy (on par with ophthalmologists) in detecting DR [7], compensating for the lack of physicians capable of making this diagnosis. However, despite AI’s promising potential, it comes with challenges and limitations of its own.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “AI expertise alone cannot solve these problems; we need to collaborate with subject matter experts. Machine learning excels at prediction and correlation, but not at identifying causation. It does not know the direction of the causality.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “We can be fooled by bias. In 1991, a study published in the New England Journal of Medicine found that left-handed people died 9 years younger than righthanded people. However, the study failed to consider that there used to be bias against lefthanded people and many of them were forced to become right-handed. Researchers had assumed that the percentage of left-handed people is stable over time; the population, although random, is biased against lefthanded people. Most data we collect have biases, and if we do not understand them and take them into account, our data models will not be correct.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “We forget that correlation or predictive power does not imply causation; the fact that two variables are correlated does not imply that one causes the other. In a Gallup poll several years ago, surveyors asked participants if correlation implied causation, and 64% of Americans answered yes. This occurs because humans learn from correlation, but we cannot observe causality. We have to understand that most people do not know the difference.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Models are very good at cheating; if there is anything the model can use to cheat, it will learn it. An AI model was trained to distinguish between skin cancer and benign lesions and was thought to achieve dermatologist-level performance. However, many of the positive cases had a ruler in the picture but the negative cases did not. The model learned that if there was a ruler present in the image, there was a much higher chance of the patient having cancer.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Access to data is one of the biggest challenges we face. There is a significant amount of data that cannot be open to researchers because of privacy issues, especially in the medical world. In most scenarios, data anonymization just does not work because even if we anonymize the data, there is always the risk of someone attempting to deanonymize it. As a result, Microsoft has invested in the Differential Privacy Platform, which provides a way for researchers to ascertain insights from the data without violating the privacy of individuals. Privacy-preserving synthetic images can also generate realistic synthetic data, including synthetic medical images, after training on a real data set, without affecting the privacy of individuals.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Other technology companies such as Google Health, NVIDIA, and Amazon have also invested in strategic partnerships with health care organizations to combine their expertise in AI with our health care domain knowledge to solve impactful clinical problems. These collaborations leverage the power of big data analytics and have immense potential to improve cancer detection, predict patient outcomes, and reduce health equity by improving patient access.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print. - “Contrary to popular media speculation, AI alone is not enough to overcome the problems that society seeks to resolve; rather, machine learning depends on subject-matter experts to find solutions. AI provides us with numerous opportunities for advancement in the field of radiology: improved diagnostic certainty, suspicious case identification for early review, better patient prognosis, and a quicker turnaround. Machine learning depends on radiologists and our expertise, and the convergence of radiologists and AI will bring forth the best outcomes for patients.”
Artificial Intelligence as a Public Service.
Lavista Ferres JM, Fishman EK, Rowe SP, Chu LC, Lugo-Fagundo E.
J Am Coll Radiol. 2023 Mar 30:S1546-1440(23)00265-X. doi: 10.1016/j.jacr.2023.01.013. Epub ahead of print.
- Sickle-Cell Trait/Disease: Some have suggested that contrast medium exposure to patients with sickle cell trait or sickle cell disease might increase the risk of an acute sickle crisis; however, there is no evidence this occurs with modern iodinated or gadolinium-based contrast medium. Therefore, restricting contrast medium use or premedicating solely on the basis of sickle cell trait or sickle cell disease is not recommended.
ACR Manual on Contrast Media 2023 - Patients premedicated for a prior contrast reaction have a breakthrough reaction rate (2.1%) that is 3-4 times the ordinary reaction rate in the general population, while patients premedicated for other indications have a breakthrough reaction rate close to 0%. In most cases (~81%), breakthrough reaction severity is similar to index reaction severity. Patients with a mild index reaction have a very low risk (<1%) of developing a severe breakthrough reaction.
ACR Manual on Contrast Media 2023 - Elective Premedication (12- or 13-hour oral premedication)
1. Prednisone-based: 50 mg prednisone by mouth at 13 hours, 7 hours, and 1 hour before contrast medium administration, plus 50 mg diphenhydramine intravenously, intramuscularly, or by mouth 1 hour before contrast medium administration [22]. Or
2. Methylprednisolone-based: 32 mg methylprednisolone by mouth 12 hours and 2 hours before contrast medium administration. 50 mg diphenhydramine may be added as in option 1 [39].
Although never formally compared, both regimens are considered similarly effective. The presence of diphenhydramine in regimen 1 and not in regimen 2 is historical and not evidence-based. Therefore, diphenhydramine may be considered optional.
ACR Manual on Contrast Media 2023 - “In patients with a prior allergic-like or unknown-type contrast reaction to a known contrast medium, changing contrast media within the same class (e.g., one iodinated medium for another) may help reduce the likelihood of a subsequent contrast reaction. Some studies have shown that the effect size of switching contrast media actually may be greater than that of premedication alone, but combining premedication with a change in agent seems to have the greatest effect. Unfortunately, many patients do not know which specific agent they have reacted to in the past; they simply remember they had a reaction. In the future, through improved electronic medical records, routine linking of reactions to specific contrast media is likely to add value. In the current state, investigating which agent was responsible for one or more prior reactions often is not possible. ”
ACR Manual on Contrast Media 2023 - Extravasations and severe extravasation injuries are more common in patients who 1) are uncommunicative, 2) have altered circulation in the injected extremity, 3) have had radiation of the injected extremity, or 4) are injected in the hand, foot, or ankle.
• Extravasations are also more common in patients injected with more viscous contrast material [6, 8, 35].
• The risk of extravasation can be minimized by 1) using angiocatheters rather than butterfly needles, 2) performing meticulous intravenous catheter insertion technique (confirming intravenous location by aspirating blood through an inserted catheter and flushing the inserted catheter with a test injection), 3) and carefully securing an inserted catheter.
ACR Manual on Contrast Media 2023 - Outpatients who have suffered contrast media extravasation should be released from the radiology department only after an initial period of observation, provided the radiologist is satisfied that any signs and symptoms that were present initially have improved or that new symptoms have not developed during the observation period. Clear instructions should be given to the patient to seek additional medical care for severe pain, progressive pain, numbness or tingling, diminished range of motion (active or passive), skin ulceration, or other neurologic or circulatory symptoms. This is because initial symptoms of a serious compartment syndrome may be absent or relatively mild (such as limited to the development of focal paresthesia).
ACR Manual on Contrast Media 2023 - The incidence of delayed allergic-like reactions has been reported to range from 0.5% to 14% . A prospective study of 258 individuals receiving intravenous iohexol demonstrated a delayed reaction rate of 14.3% compared to 2.5% in a control group undergoing imaging without intravascular contrast material. In that same study, 26 of 37 delayed adverse reactions were cutaneous in nature. For several reasons (lack of awareness of such adverse events, usual practice patterns, relatively low frequency of serious outcomes), such reactions are often not brought to the attention of the radiologist. Delayed reactions are more common in patients treated with interleukin-2 (IL-2) therapy . There is some evidence that the iso-osmolar dimer iodixanol may have a slightly higher rate of delayed cutaneous adverse events when compared to other LOCM . A prospective study by Schild et al demonstrated an increased frequency of delayed cutaneous adverse events to nonionic dimeric contrast material compared to nonionic monomeric contrast material.
ACR Manual on Contrast Media 2023
Small Bowel
- “Castleman disease (CD) is a group of rare and complex lymphoproliferative disorders that can manifest in two general forms: unicentric CD (UCD) and multicentric CD (MCD). These two forms differ in clinical manifestation, imaging appearances, treatment options, and prognosis. UCD typically manifests as a solitary enlarging mass that is discovered incidentally or after development of compression-type symptoms. MCD usually manifests acutely with systemic symptoms including fever and weight loss. As a whole, CD involves lymph nodes throughout the chest, neck, abdomen, pelvis, and axilla and can have a wide variety of imaging appearances.”
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021 - “Most commonly, lymph nodes or masses in UCD occur in the chest, classically with well-defined borders, hyperenhancement, and possible characteristic patterns of calcification and/or feeding vessels. Lymph nodes affected by MCD, while also hyperenhancing, tend to involve multiple nodal chains and manifest alongside anasarca or hepatosplenomegaly. The polyneuropathy, organomegaly, endocrinopathy, monoclonal plasma cell disorder, and skin changes (POEMS) subtype of MCD may demonstrate lytic or sclerotic osseous lesions in addition to features typical of MCD. Since a diagnosis of CD based solely on imaging findings is often not possible, pathologic confirmation with core needle biopsy and/or surgical excision is necessary. Nevertheless, imaging plays a crucial role in supporting the diagnosis of CD, guiding appropriate regions for biopsy, and excluding other potential causes or mimics of disease.”
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021 - - Castleman disease (CD), also known as angiofollicular lymph node hyperplasia, is a complex lymphoproliferative disorder broadly divided into two forms:unicentric CD (UCD) and multicentric CD (MCD). The disease likely encompasses at least four different entities with overlapping histopathologic features but distinct causes, clinical manifestations, radiologic features, treatment approaches, and outcomes.
- Most commonly, UCD manifests as a solitary enlarging mass, which can be discovered incidentally at imaging performed for other reasons, be palpated at physical examination, or be discovered owing to compression-type symptoms on adjacent structures.
- CD most commonly manifests as unicentric disease with various sites of involvement. The chest is the most common site (30%–70%), followed by the neck (10%–40%), abdomen and pelvis (12%–39%), and axilla (4%–5%).
- Multiple modalities may demonstrate prominent feeding vessels to the lymph node or mass, which can affect the surgical approach.
- Lymphoma is a common mimic that is often a diagnostic consideration for presumed CD in different parts of the body. Lymphoma can be confused with CD in the mediastinum, MCD, or nonspecific masses in other parts of the body.
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021 - “Lymph nodes in MCD tend to be hypervascular, similar to those in UCD, although some reports suggest that nodal masses associated with HHV8-MCD may not show substantial enhancement. MCD in the chest typically manifests as cervical, hilar, mediastinal, or axillary lymphadenopathy, which affects multiple nodal chains. Pleural or pericardial effusions are common. Lung involvement may appear as centrilobular nodules and nodular opacities and less frequently as ground-glass attenuation, airspace consolidation, and bronchiectasis. Although unusual, when direct pleural involvement occurs, it can manifest as a well-defined pleural-based mass with or without associated pleural effusion.”
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021 - “Lymphoma is a common mimic that is often a diagnostic consideration for presumed CD in different parts of the body .Lymphoma can be confused with CD in the mediastinum, MCD, or nonspecific masses in other parts of the body .At imaging, increasing adenopathy, ascites, or new splenomegaly may indicate a diagnosis of lymphoma in the setting of known CD . Nodal masses with higher SUVmax are alsoconcerning for either non-Hodgkin lymphoma (NHL) or Hodgkin lymphoma, as the median SUV of lymph nodes involved by CD tends to be slightly lower.”
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021 - “In the absence of complications or development of lymphoma, UCD typically has no significant effect on life expectancy. Available data on outcomes and prognosis in MCD are more limited. The clinical course of patients with MCD is variable: some patients may present with indolent and slowly progressive disease, others may experience a relapsing-remitting course, and still others may present with fulminant disease that can be fatal within the course of weeks. In one study, 35% of patients diagnosed with iMCD were reported to have died within 5 years of diagnosis and 60% were reported to have died within 10 years. Patients also have a threefold increased prevalence of malignancy.”
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021 - “CD is a complex and heterogeneous group of lymphoproliferative disorders that can manifest with varying clinical, imaging, and histopathologic features. UCD typically manifests as an incidental solitary enlarging mass, whereas MCD manifests acutely with systemic manifestations. A multidisciplinary approach to CD including clinical information, laboratory evaluation, and imaging workup is necessary to select appropriate regions for biopsy, exclude potential mimics, differentiate UCD from MCD, arrive at an accurate diagnosis, and ultimately manage CD. Imaging can be used to monitor response to treatment and detect development of any complications such as lymphoma. Surgical excision is typically curative for UCD, while MCD requires systemic therapy with a more guarded prognosis.”
Imaging of Castleman Disease
Marika A. Pitot et al.
RadioGraphics 2023; 43(8):e22021
- The most common CT/MR patterns of small bowel lymphoma are:
(i) Polypoid/nodular pattern.
(ii) Infiltrative pattern.
(iii) Aneurismal pattern.
(iv) Exophytic mass.
(v) Stenosing mass (rare).
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “The polypoid pattern is characterized by the presence of a solid nodule, with a homogeneous signal density/intensity, that develops in the submucosa and protrudes into the lumen appearing as a polypoid mass. There is no wall thickening and/or lymph adenopathy and the mucosa is intact. This mass may cause intussusception. The infiltrative form is characterized by segmental symmetrical or slightly asymmetrical infiltrating lesions with a medium diameter of 1.5 cm and 2 cm, associated with mild circumferential thickening of the small bowel wall. Usually, the infiltrative lesions show ill-defined margins and a homogeneous contrast enhancement; the latter may rarely be inhomogeneous because of the presence of hypodense/ hypointense areas due to development of necrosis and/or ischemia in the context of the lesion. These lesions may extend to the whole bowel thickness, from the endoluminal mucosa to the tunica serosa.The length of the thickened small bowel segment is variable.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “The aneurismal pattern (diameter of dilatation of the lumen over 4 cm), firstly diagnosed byCupps et al. in 1969, represents 31% of small bowel lymphomas. It usually coexists with the infiltrative form since it can represent its natural evolution. Several factors are responsible for the aneurismal dilation secondary to infiltrative growth of neoplastic lesion, as a progressive destruction of myenteric plexus, destruction of muscle layers with stretching of the muscle fibers, and loss of contractile cells; on the other hand, the infiltration of arterial and lymphatic vessels determines anoxia and necrosis within the lesion. According to some authors, this tumour necrosis could lead to cavitation and be also responsible for the aneurismal dilatation.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “Differential diagnosis includes all inflammatory, neoplastic, and metastatic lesions involving the small bowel. Primary carcinoma, metastases (especially those from melanoma and renal cancer), and the intestinal leiomyosarcoma are characterized by large necrotic/colliquative cavitations. In rare cases, inflammatory conditions, such as Crohn’s disease and intestinal tuberculosis, have to be differentiated: the significant thickening of the bowel wall (greater than 2 cm), the presence of lymphomatous nodules, and the coexistence of perivisceral multiple lymph nodes are CT features that are suggestive for a lymphoproliferative process.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143
Spleen
- “Splenic angiosarcoma is exceedingly rare, but it is the most common primary nonhematolymphoid malignant neoplasm of the spleen. It is a highly aggressive malignancy with a poor prognosis. The majority of patients present with abdominal pain or a palpable abdominal mass. Occasionally, widespread metastases or splenic rupture will be the presenting manifestation. The clinicopathologic features of splenic angiosarcoma are widely reported throughout the medical literature .”
Angiosarcoma of the Spleen: Imaging Characteristics in 12 Patients
William M. Thompson, et al
Radiology 2005; 235:106–115 - “Similar to our results, previous CT reports have demonstrated solitary or multiple nodular masses of heterogeneous low attenuation in an enlarged spleen. Some of these masses show peripheral enhancement, and the margins of the lesions are often irregular or poorly marginated. On precontrast CT scans, the tumors may appear hyperattenuating, which corresponds to acute hemorrhage. On dynamic contrast-enhanced CT scans, the lesions may exhibit substantial peripheral contrast enhancement similar to that of hepatic hemangiomas.”
Angiosarcoma of the Spleen: Imaging Characteristics in 12 Patients
William M. Thompson, et al
Radiology 2005; 235:106–115
Stomach
- “The most common CT patterns of gastric lymphoma are the presence of diffuse or segmental wall thickening of 2–5 cm with low contrast enhancement and extensive lateral extension of the tumour due to submucosal spread; moreover, CT can assess the presence of lymphadenopathies. Less commonly, gastric lymphoma may present on CT as a polypoidal mass, an ulcerative lesion, or a mucosal nodularity. Considering the CT features of lymphoma, in low-grade ones there is less severe gastric wall thickening than in highgrade lymphoma, and abdominal lymphadenopathy is less common.The absence of abnormality or the presence of just minimal gastric wall thickening or a shallow lesion at CT suggests low-grade MALT lymphoma; yet, CT is of limited value in its diagnosis. A greater thickening may indicate transformation to a higher grade lymphoma.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143 - “CT is particularly useful both for staging and in the follow-up after surgery or chemoradiotherapy.Nowadays,CT allows the evaluation of wall thickness, mesenteric vasculature, and any associated extramural findings. Small bowel CT, or entero-CT, performed through a multislice CT scanner has led to considerable advances in the detection and staging of intestinal diseases.The advantage of this technique lies in its panoramic view, which allows the evaluation of the intestinal wall thickness, the degree of bowel distension, and the circular folds. Yet, ileal loops and also those of the deep pelvis, the mesentery, the surrounding adipose tissue, and other abdominal organs are studied.”
Radiological Features of Gastrointestinal Lymphoma
Giuseppe Lo Re et al.
Gastroenterology Research and Practice Volume 2016, Article ID 2498143
- “Carney-Stratakis syndrome, first described in 2002, describes the dyad of familial paraganglioma and gastrointestinal stromal tumor. It is a unique entity, separate from the previously described Carney triad, which noted the association between paraganglioma, gastrointestinal stromal tumor (GIST), and pulmonary chondroma. Carney-Stratakis syndrome is associated with germline mutations in the succinate dehydrogenase genes SDHB, SDHC, and SDHD, with autosomal dominant inheritance and incomplete penetrance.”
Carney-Stratakis syndrome: A dyad of familial paraganglioma and gastrointestinal stromal tumor,
Hannah S. Recht, Elliot K. Fishman
Radiology Case Reports, Vol 15, Issue 11,2020, Pages 2071-2075, - “Carney-Stratakis syndrome is one of the familial GIST syndromes, which also includes primary familial GIST syndrome and neurofibromatosis type 1. However, while the familial GIST syndrome is associated with KIT and PDGFRA mutations, Carney-Stratakis syndrome is associated with germline mutations of the succinate dehydrogenase subunits B, C, and D.”
Carney-Stratakis syndrome: A dyad of familial paraganglioma and gastrointestinal stromal tumor,
Hannah S. Recht, Elliot K. Fishman
Radiology Case Reports, Vol 15, Issue 11,2020, Pages 2071-2075,
Vascular
- Median arcuate ligament syndrome (MALS) is one of the abdominal vascular compression syndromes diagnosed by Harjola in 1963. It is more commonly known as celiac artery compression syndrome, also known as Dunbar syndrome, named after the radiologist JD Dunbar. It typically affects young women in the ratio of 2:1 to 3:1 and age group of 20–40 years of age with a reported incidence of 2 per 100,000 patients. Usually, the median arcuate ligament crosses at L1 above the origin of the celiac artery. In 10%–24% of the general population, MAL can have a low insertion; however, even a smaller subset will develop symptoms due to celiac artery compression. Typical physical examination findings include abdominal bruit in a mid-epigastric region that varies with respiration.
- “Median arcuate ligament syndrome (MALS) is one of the abdominal vascular compression syndromes diagnosed by Harjola in 1963. It is more commonly known as celiac artery compression syndrome, also known as Dunbar syndrome, named after the radiologist JD Dunbar. It typically affects young women in the ratio of 2:1 to 3:1 and age group of 20–40 years of age with a reported incidence of 2 per 100,000 patients. Usually, the median arcuate ligament crosses at L1 above the origin of the celiac artery. In 10%–24% of the general population, MAL can have a low insertion; however, even a smaller subset will develop symptoms due to celiac artery compression.”
Median arcuate ligament syndrome diagnosis on Computed Tomography: what a radiologist needs to know.
Narwani P et al..
Radiol Case Rep. 2021 Sep 16;16(11):3614-3617 - “Median arcuate ligament syndrome or celiac artery compression syndrome is one of the abdominal vascular compression syndromes due to compression of proximal celiac artery by the median arcuate ligament. The median arcuate ligament unites diaphragmatic crura on either side at the level of aortic hiatus. The ligament has a low insertion causing compression of the celiac artery resulting in clinical symptoms of postprandial pain and weight loss. It is a rare syndrome, detected incidentally on routine Computed Tomography abdomen and pelvis studies.”
Median arcuate ligament syndrome diagnosis on Computed Tomography: what a radiologist needs to know.
Narwani P et al..
Radiol Case Rep. 2021 Sep 16;16(11):3614-3617 - “In MALS, the pathophysiology is vascular due to compression of the celiac artery causing foregut ischemia, vascular steal phenomenon causing midgut ischemia with splanchnic vasoconstriction and ischemia. Some authors think the aetiology is neurogenic due to the compression of celiac plexus and ganglion.”
Median arcuate ligament syndrome diagnosis on Computed Tomography: what a radiologist needs to know.
Narwani P et al..
Radiol Case Rep. 2021 Sep 16;16(11):3614-3617 - ‘Also, note that CT studies are best evaluated in the end-inspiratory phase. Since MAL is attached to the diaphragm, movement occurs with respiration, and true compression can be evaluated in the end-inspiratory phase. Isolated compression of the celiac axis in expiration can be observed in 13%–50% of healthy individuals and can be clinically insignificant. Few of these patients would have clinical symptoms due to hemodynamic compromise. In a retrospective study, Heo et al. showed that 87% of patients with classical imaging findings of MALS incidentally detected on CT had no symptoms.”
Median arcuate ligament syndrome diagnosis on Computed Tomography: what a radiologist needs to know.
Narwani P et al..
Radiol Case Rep. 2021 Sep 16;16(11):3614-3617 - --Median arcuate ligament syndrome should be diagnosed carefully in combination with clinical and imaging findings. Low insertion of the median arcuate ligament is an important anatomical variant and can be seen in asymptomatic individuals.
--Classical imaging findings include compression of proximal celiac artery by the ligament with associated post stenotic dilatation resulting in the characteristic hooked shaped configuration.
--MALS can be easily missed on routine CT, so the radiologist should carefully assess the multiplanar images to establish the diagnosis.
Median arcuate ligament syndrome diagnosis on Computed Tomography: what a radiologist needs to know.
Narwani P et al..
Radiol Case Rep. 2021 Sep 16;16(11):3614-3617 - Dynamic CT examination may also be performed in both deep inspiration and expiration in order to evaluate the dynamic modifications in celiac artery diameter. CT imaging should include the early arterial phase acquired in deep expiration in order to increase the proximal celiac trunk compression by the median arcuate ligament, followed by the portal venous phase in deep inspiration. Sagittal and coronal images should be included for optimal visualization of the celiac artery. The proximal narrowing of the celiac trunk can be better depicted on sagittal CT reconstructions, demonstrating a focal indentation on the superior surface of the vessel with a typical “hooked appearance”, in the absence of atherosclerotic plaques or other causes of extrinsic compression.
CT imaging findings of abdominopelvic vascular compression syndromes: what the radiologist needs to know.
Gozzo, C., Giambelluca, D., Cannella, R. et al.
Insights Imaging 11, 48 (2020) - “The classic clinical manifestations of MALS include chronic postprandial epigastric pain, nausea, and loss of weight due to dynamic compression of the celiac artery. However, this anatomical anomaly is asymptomatic in up to 85% of patients and may be incidentally encountered on CT examinations performed for unrelated reasons. The mechanism of pain is still debated. During expiration, the abdominal aorta and its branches are displaced superiorly and the median arcuate ligament compression of the celiac artery increases. This may be responsible for a steal phenomenon with blood flow diverted away from the superior mesenteric artery to the celiac artery branches trough the collateral pathway of pancreaticoduodenal arcades.”
CT imaging findings of abdominopelvic vascular compression syndromes: what the radiologist needs to know.
Gozzo, C., Giambelluca, D., Cannella, R. et al.
Insights Imaging 11, 48 (2020)