Feasibility of Differential Diagnosis Based on Imaging Patterns Using a Large Language Model

HomeRadiologyVol. 308, No. 1 PreviousNext Original ResearchComputer ApplicationsFeasibility of Differential Diagnosis Based on Imaging Patterns Using a Large Language ModelJonathan Kottlorsa, Grischa Bratkea, Philip Rauena, Christoph Kabbascha, Thorsten Persigehla, Marc Schlamanna, Simon Lennartza Jonathan Kottlorsa, Grischa Bratkea, Philip Rauena, Christoph Kabbascha, Thorsten Persigehla, Marc Schlamanna, Simon Lennartza Author AffiliationsInstitute for Diagnostic and Interventional Radiology, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, GermanyCorresponding Author Simon Lennartz, MD Institute for Diagnostic and Interventional Radiology, University Hospital of Cologne, Kerpener Straße 62, 50937 Cologne, Germany. Email address: [email protected]Jonathan KottlorsaGrischa BratkeaPhilip RauenaChristoph KabbaschaThorsten PersigehlaMarc SchlamannaSimon Lennartza Published Online:Jul 5 2023https://doi.org/10.1148/radiol.231167MoreSectionsFull textPDF ToolsImage ViewerAdd to favoritesCiteTrack CitationsPermissionsReprints ShareShare onFacebookTwitterLinked In AbstractGPT-4 attained a concordance of 68.8% (55 of 80) with expert consensus at determining top differential diagnoses based on imaging patterns, and 93.8 % (75 of 80) of differential diagnoses proposed by GTP-4 were deemed acceptable alternatives.References1. Procedural and Declarative Knowledge: An Evolutionary Perspective. Ten Berge, T. & Van Hezewijk, R., Theory Psychol 9, (1999). Crossref, Google Scholar2. Radiology as personal knowledge. Gray, B. R., Mark Mutz, J. & Gunderman, R. B., American Journal of Roentgenology 214, (2020). Crossref, Medline, Google Scholar3. Evidence-based radiology: Review and dissemination. Medina, L. S. & Blackmore, C. C., Radiology vol. 244 Preprint (2007). Link, Google Scholar4. OpenAI. GPT-4 Technical Report. 1–100 (2023). Google Scholar5. Leveraging GPT-4 for Post Hoc Transformation of Free-Text Radiology Reports into Structured Reporting: A Multilingual Feasibility Study. Adams, L. C. et al. Radiology (2023) doi: https://doi.org/10.1148/radiol.230725. Link, Google Scholar6. ChatGPT and Other Large Language Models Are Double-edged Swords. Shen Y, Heacock L, Elias J et al. Radiology 2023; 307:2. doi: https://doi.org/10.1148/radiol.230163 Link, Google ScholarArticle HistoryPublished online: July 05 2023 FiguresReferencesRelatedDetailsAccompanying This ArticleChatGPT in RadiologyAug 1 2023Default Digital Object SeriesRecommended Articles RSNA Education Exhibits RSNA Case Collection Vol. 308, No. 1 PodcastAbbreviations Abbreviations: (GPT-4) Generative Pre-trained Transformer 4 (LLM) Large language models (y) Years Metrics Altmetric Score PDF download