Valerie Builoff BS , Aakash Shanbhag MSc , Robert JH. Miller MD , Damini Dey PhD , Joanna X. Liang MPH , Kathleen Flood BS , Jamieson M. Bourque MD , Panithaya Chareonthaitawee MD , Lawrence M. Phillips MD , Piotr J. Slomka PhD
{"title":"Evaluating AI proficiency in nuclear cardiology: Large language models take on the board preparation exam","authors":"Valerie Builoff BS , Aakash Shanbhag MSc , Robert JH. Miller MD , Damini Dey PhD , Joanna X. Liang MPH , Kathleen Flood BS , Jamieson M. Bourque MD , Panithaya Chareonthaitawee MD , Lawrence M. Phillips MD , Piotr J. Slomka PhD","doi":"10.1016/j.nuclcard.2024.102089","DOIUrl":"10.1016/j.nuclcard.2024.102089","url":null,"abstract":"<div><h3>Background</h3><div>Previous studies evaluated the ability of large language models (LLMs) in medical disciplines; however, few have focused on image analysis, and none specifically on cardiovascular imaging or nuclear cardiology. This study assesses four LLMs—GPT-4, GPT-4 Turbo, GPT-4omni (GPT-4o) (Open AI), and Gemini (Google Inc.)—in responding to questions from the 2023 American Society of Nuclear Cardiology Board Preparation Exam, reflecting the scope of the Certification Board of Nuclear Cardiology (CBNC) examination.</div></div><div><h3>Methods</h3><div>We used 168 questions: 141 text-only and 27 image-based, categorized into four sections mirroring the CBNC exam. Each LLM was presented with the same standardized prompt and applied to each section 30 times to account for stochasticity. Performance over six weeks was assessed for all models except GPT-4o. McNemar's test compared correct response proportions.</div></div><div><h3>Results</h3><div>GPT-4, Gemini, GPT-4 Turbo, and GPT-4o correctly answered median percentages of 56.8% (95% confidence interval 55.4% - 58.0%), 40.5% (39.9% - 42.9%), 60.7% (59.5% - 61.3%), and 63.1% (62.5%–64.3%) of questions, respectively. GPT-4o significantly outperformed other models (<em>P</em> = .007 vs GPT-4 Turbo, <em>P</em> < .001 vs GPT-4 and Gemini). GPT-4o excelled on text-only questions compared to GPT-4, Gemini, and GPT-4 Turbo (<em>P</em> < .001, <em>P</em> < .001, and <em>P</em> = .001), while Gemini performed worse on image-based questions (<em>P</em> < .001 for all).</div></div><div><h3>Conclusion</h3><div>GPT-4o demonstrated superior performance among the four LLMs, achieving scores likely within or just outside the range required to pass a test akin to the CBNC examination. Although improvements in medical image interpretation are needed, GPT-4o shows potential to support physicians in answering text-based clinical questions.</div></div>","PeriodicalId":16476,"journal":{"name":"Journal of Nuclear Cardiology","volume":"45 ","pages":"Article 102089"},"PeriodicalIF":3.0,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142769756","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Diagnostic accuracy of low-dose myocardial perfusion imaging in a real-world setting","authors":"Mathieu Perrin MD , Marine Claudin MD , Karim Djaballah MD , Caroline Boursier MD , Antoine Verger MD, PhD , Laetitia Imbert PhD , Véronique Roch MSc , Matthieu Doyen PhD , Loïc Marie MSc , Gilles Karcher MD, PhD , Batric Popovic MD, PhD , Zohra Lamiral MSc , Edoardo Camenzind MD, PhD , Pierre-Yves Marie MD, PhD","doi":"10.1016/j.nuclcard.2025.102140","DOIUrl":"10.1016/j.nuclcard.2025.102140","url":null,"abstract":"<div><h3>Background</h3><div>This large-scale study analyzes factors affecting the diagnostic accuracy of low-dose myocardial perfusion imaging and correlation with coronary angiography in a real-world practice.</div></div><div><h3>Methods</h3><div>We compared data extracted from routine reports of (i) low-dose [<sup>99m</sup>Tc]sestamibi stress-MPI performed with no attenuation correction and predominantly exercise stress testing and (ii) the corresponding coronary angiography.</div></div><div><h3>Results</h3><div>We considered 1070 pairs of coronary angiography/stress-MPI results reported by 11 physicians. Mean MPI effective dose was 4.5 ± 2.1 mSv. The extent of MPI-ischemia was predictive of >70% but not 50%–70% coronary stenoses. A positive test was associated with a sensitivity of 74.7% (413/553) and a specificity of 53.2% (275/517) for >70% stenosis detection. Positive predictive values were lower in patients with left bundle branch block or pacemakers (LBBB/PM) (45.6% vs 64.7%, <em>P</em> = .006) and markedly higher for patients with MPI-ischemia ≥3 segments or associated with ST-segment depression (75.0% (165/220)) as compared to those with <3 segments MPI-ischemia, MPI-infarction or isolated ST-segment depression (57% (248.0/435), <em>P</em> < .001). Negative predictive values were lower for patients with previous coronary artery disease (CAD) history (58.3%), male (61.0%), and elderly patients (59.6%) (vs 72.1%, 79.2%, and 72.4%, respectively, all <em>P</em> < .05).</div></div><div><h3>Conclusions</h3><div>Routine results from low-dose stress-MPI, predominantly associated with exercise stress testing and uncorrected for attenuation, correlate with real-world coronary angiography results. However, this correlation is lower than that achieved with conventional study designs and affected by the definition of significant CAD and context variables (LBBB/PM, CAD history, sex, and age). Better consideration of these interacting factors could improve patient monitoring.</div></div>","PeriodicalId":16476,"journal":{"name":"Journal of Nuclear Cardiology","volume":"45 ","pages":"Article 102140"},"PeriodicalIF":3.0,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142950268","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Serial assessment of diffusion-weighted whole-body magnetic resonance imaging with background body signal suppression and <sup>18</sup>F-fluorodeoxyglcose positron emission tomography/computed tomography after corticosteroid therapy in Takayasu arteritis.","authors":"Kazunori Omote, Motoo Oi, Tadao Aikawa","doi":"10.1016/j.nuclcard.2025.102171","DOIUrl":"10.1016/j.nuclcard.2025.102171","url":null,"abstract":"","PeriodicalId":16476,"journal":{"name":"Journal of Nuclear Cardiology","volume":" ","pages":"102171"},"PeriodicalIF":3.0,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143537149","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Huidong Xie, Alaa Alashi, Stephanie L Thorn, Xiongchao Chen, Bo Zhou, Albert J Sinusas, Chi Liu
{"title":"Increasing angular sampling for dedicated cardiac single photon emission computed tomography scanner: Implementation with deep learning and validation with human data.","authors":"Huidong Xie, Alaa Alashi, Stephanie L Thorn, Xiongchao Chen, Bo Zhou, Albert J Sinusas, Chi Liu","doi":"10.1016/j.nuclcard.2025.102168","DOIUrl":"10.1016/j.nuclcard.2025.102168","url":null,"abstract":"","PeriodicalId":16476,"journal":{"name":"Journal of Nuclear Cardiology","volume":" ","pages":"102168"},"PeriodicalIF":3.0,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143476581","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"‘TIDing’ over the uncertainty: Prognostic value of Transient Ischemic Dilation in Rubidium-82 PET myocardial perfusion imaging","authors":"Christoph Rischpler","doi":"10.1016/j.nuclcard.2025.102142","DOIUrl":"10.1016/j.nuclcard.2025.102142","url":null,"abstract":"","PeriodicalId":16476,"journal":{"name":"Journal of Nuclear Cardiology","volume":"44 ","pages":"Article 102142"},"PeriodicalIF":3.0,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143403502","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}