{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,4]],"date-time":"2026-08-04T08:51:40Z","timestamp":1785833500544,"version":"3.56.0"},"reference-count":39,"publisher":"Oxford University Press (OUP)","issue":"5","license":[{"start":{"date-parts":[[2026,3,24]],"date-time":"2026-03-24T00:00:00Z","timestamp":1774310400000},"content-version":"vor","delay-in-days":1,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,7,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Objectives<\/jats:title>\n                    <jats:p>To develop and evaluate JADE, a proof-of-concept retrieval-augmented generation (RAG) diagnostic assistive system, designed to enhance large language model (LLM) reasoning for jawbone lesion assessment. This study examined whether RAG improves diagnostic accuracy and stability compared with standalone LLMs and ORAD, a supervised learning-based system.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>JADE was developed as a cloud-based RAG system integrating an expert-curated oral radiology database embedded using text-embedding-3-large and indexed in Qdrant Cloud. Structured clinical inputs were encoded as prioritized vector queries. Hybrid semantic and keyword-based retrieval appended relevant evidence to prompts for differential diagnosis generation across LLM backbones. Performance was evaluated in 25 validation cases and compared with GPT-5, Claude Sonnet 4.5, DeepSeek-R1, Gemini 2.5 Flash, their RAG configurations, and ORAD. Diagnostic accuracy was analyzed using Cochran\u2019s Q test, with post-hoc McNemar\u2019s tests and Bonferroni correction. Intra-model stability and response time were assessed.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>RAG-GPT-5 achieved the highest diagnostic accuracy (20\/25), followed by RAG-Claude Sonnet 4.5 (18\/25), RAG-DeepSeek R1 (17\/25), and RAG-Gemini 2.5 Flash (15\/25). Standalone models achieved 9\u201313\/25 correct diagnoses, ORAD achieved 17\/25. No significant differences were observed among standalone models or RAG-based models and ORAD. A significant improvement was observed for GPT-5 when integrated with RAG (P\u2009=\u2009.002). RAG configurations showed higher intra-model stability, with RAG-GPT-5 achieving mean stability of 0.90\u2009\u00b1\u20090.11. Mean response times ranged from 3 to 10\u2009s.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>JADE improved diagnostic accuracy and stability compared with standalone LLMs, underscoring value of RAG reasoning in jawbone lesion assessment and marking the first RAG application in dentomaxillofacial radiology.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/dmfr\/twag017","type":"journal-article","created":{"date-parts":[[2026,3,24]],"date-time":"2026-03-24T02:42:32Z","timestamp":1774320152000},"page":"497-507","source":"Crossref","is-referenced-by-count":2,"title":["JADE: jawbone lesion diagnosis and decision supporting system"],"prefix":"10.1093","volume":"55","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2092-6076","authenticated-orcid":false,"given":"Soroush","family":"Baseri Saadi","sequence":"first","affiliation":[{"name":"OMFS-IMPATH Research Group, Department of Imaging and Pathology, Catholic University Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Department of Oral and Maxillofacial Surgery, University Hospitals Leuven , Leuven 3000,","place":["Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8879-6052","authenticated-orcid":false,"given":"Jonas","family":"Ver Berne","sequence":"additional","affiliation":[{"name":"OMFS-IMPATH Research Group, Department of Imaging and Pathology, Catholic University Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Department of Pathology, University Hospitals Leuven , Leuven 3000,","place":["Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6426-9768","authenticated-orcid":false,"given":"Rocharles","family":"Cavalcante Fontenele","sequence":"additional","affiliation":[{"name":"OMFS-IMPATH Research Group, Department of Imaging and Pathology, Catholic University Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Division of Oral Radiology, Department of Stomatology, Public Health, and Forensic Dentistry, School of Dentistry of Ribeirao Preto, University of Sao Paulo (USP) , Ribeirao Preto, Sao Paulo 14040-904,","place":["Brazil"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9489-9819","authenticated-orcid":false,"given":"Peter","family":"Claes","sequence":"additional","affiliation":[{"name":"Department of Human Genetics, KU Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Medical Imaging Research Center, UZ Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Department of Electrical Engineering, Processing of Speech and Images (ESAT-PSI), KU Leuven , Leuven 3000,","place":["Belgium"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3461-0363","authenticated-orcid":false,"given":"Reinhilde","family":"Jacobs","sequence":"additional","affiliation":[{"name":"OMFS-IMPATH Research Group, Department of Imaging and Pathology, Catholic University Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Department of Oral and Maxillofacial Surgery, University Hospitals Leuven , Leuven 3000,","place":["Belgium"]},{"name":"Department of Dental Medicine, Karolinska Institutet , Stockholm SE-141 04,","place":["Sweden"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,3,23]]},"reference":[{"key":"2026063015133096300_twag017-B1","doi-asserted-by":"publisher","first-page":"392","DOI":"10.1111\/j.1600-0714.2006.00451.x","article-title":"An analysis of oral and maxillofacial pathology found in adults over a 30-year period","volume":"35","author":"Jones","year":"2006","journal-title":"J Oral Pathol Med"},{"key":"2026063015133096300_twag017-B2","doi-asserted-by":"publisher","first-page":"693","DOI":"10.1016\/j.path.2017.04.007","article-title":"Bone-related lesions of the jaws","volume":"10","author":"Baumhoer","year":"2017","journal-title":"Surg Pathol Clin"},{"key":"2026063015133096300_twag017-B3","volume-title":"Oral Radiology: Principles and Interpretation","author":"White","year":"2014","edition":"7th ed"},{"key":"2026063015133096300_twag017-B4","doi-asserted-by":"publisher","first-page":"104581","DOI":"10.1016\/j.jdent.2023.104581","article-title":"A deep learning approach for radiological detection and classification of radicular cysts and periapical granulomas","volume":"135","author":"Ver Berne","year":"2023","journal-title":"J Dent"},{"key":"2026063015133096300_twag017-B5","doi-asserted-by":"publisher","first-page":"105688","DOI":"10.1016\/j.jdent.2025.105688","article-title":"Automated classification of panoramic radiographs with inflammatory periapical lesions using a CNN-LSTM architecture","volume":"156","author":"Ver Berne","year":"2025","journal-title":"J Dent"},{"key":"2026063015133096300_twag017-B6","doi-asserted-by":"publisher","first-page":"100225","DOI":"10.1016\/j.oor.2024.100225","article-title":"A review: exploring the role of ChatGPT in the diagnosis and treatment of oral pathologies","volume":"10","author":"Panwar","year":"2024","journal-title":"Oral Oncol Rep"},{"key":"2026063015133096300_twag017-B7","doi-asserted-by":"publisher","first-page":"613","DOI":"10.1093\/dmfr\/twaf060","article-title":"The performance of large language models in dentomaxillofacial radiology: a systematic review","volume":"54","author":"Liu","year":"2025","journal-title":"Dentomaxillofac Radiol"},{"key":"2026063015133096300_twag017-B8","doi-asserted-by":"publisher","first-page":"57","DOI":"10.1016\/j.jdsr.2026.01.002","article-title":"The role of large language models in dental diagnosis, decision-making, and communication: a systematic review","volume":"62","author":"Kaplan","year":"2026","journal-title":"Jpn Dent Sci Rev"},{"key":"2026063015133096300_twag017-B9","doi-asserted-by":"publisher","first-page":"638","DOI":"10.1186\/s12903-025-06034-x","article-title":"Can Deepseek and ChatGPT be used in the diagnosis of oral pathologies?","volume":"25","author":"Kaygisiz","year":"2025","journal-title":"BMC Oral Health"},{"key":"2026063015133096300_twag017-B10","doi-asserted-by":"publisher","first-page":"1557","DOI":"10.1007\/s10006-024-01284-5","article-title":"Concordance of clinician, Chat-GPT4, and ORAD diagnoses against histopathology in odontogenic keratocysts and tumours: a 15-year New Zealand retrospective study","volume":"28","author":"Kim","year":"2024","journal-title":"Oral Maxillofac Surg"},{"key":"2026063015133096300_twag017-B11","doi-asserted-by":"publisher","first-page":"3361","DOI":"10.1111\/odi.70007","article-title":"Diagnostic performance of ChatGPT-4o and DeepSeek-3 differential diagnosis of complex oral lesions: a multimodal imaging and case difficulty analysis","volume":"31","author":"Hassanein","year":"2025","journal-title":"Oral Dis"},{"key":"2026063015133096300_twag017-B12","doi-asserted-by":"publisher","first-page":"573","DOI":"10.1186\/s12903-025-05926-2","article-title":"Artificial intelligence performance in answering multiple-choice oral pathology questions: a comparative analysis","volume":"25","author":"Yilmaz","year":"2025","journal-title":"BMC Oral Health"},{"key":"2026063015133096300_twag017-B13","doi-asserted-by":"publisher","first-page":"2328","DOI":"10.1016\/j.prosdent.2024.10.020","article-title":"Large language models in periodontology: assessing their performance in clinically relevant questions","volume":"134","author":"Chatzopoulos","year":"2025","journal-title":"J Prosthet Dent"},{"key":"2026063015133096300_twag017-B14","first-page":"4504","author":"Ji","year":"2023"},{"key":"2026063015133096300_twag017-B15","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3571730","article-title":"Survey of hallucination in natural language generation","volume":"55","author":"Ji","year":"2023","journal-title":"ACM Comput Surv"},{"key":"2026063015133096300_twag017-B16","doi-asserted-by":"publisher","author":"He","year":"2022","DOI":"10.48550\/arXiv.2301.00303,"},{"key":"2026063015133096300_twag017-B17","first-page":"706","author":"Wang","year":"2024"},{"key":"2026063015133096300_twag017-B18","first-page":"9459","article-title":"Retrieval-augmented generation for knowledge-intensive NLP tasks","volume":"33","author":"Lewis","year":"2020","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026063015133096300_twag017-B19","doi-asserted-by":"publisher","first-page":"551","DOI":"10.1007\/s12599-025-00945-3","article-title":"Retrieval-augmented generation (RAG)","volume":"67","author":"Klesel","year":"2025","journal-title":"Bus Inf Syst Eng"},{"key":"2026063015133096300_twag017-B20","first-page":"3929","article-title":"Retrieval augmented language model pre-training","volume":"119","author":"Guu","year":"2020","journal-title":"Proc Int Conf Mach Learn"},{"key":"2026063015133096300_twag017-B21","first-page":"2206","article-title":"Improving language models by retrieving from trillions of tokens","volume":"162","author":"Borgeaud","year":"2022","journal-title":"Proc Int Conf Mach Learn"},{"key":"2026063015133096300_twag017-B22","doi-asserted-by":"publisher","first-page":"1760990","DOI":"10.3389\/fdmed.2025.1760990","article-title":"Retrieval augmented generation in dentistry: potentials, applications, and future directions","volume":"6","author":"Firoozi","year":"2026","journal-title":"Front Dent Med"},{"key":"2026063015133096300_twag017-B23","doi-asserted-by":"publisher","first-page":"106006","DOI":"10.1016\/j.ijmedinf.2025.106006","article-title":"Development and evaluation of a retrieval-augmented large language model framework for enhancing endodontic education","volume":"203","author":"Xu","year":"2025","journal-title":"Int J Med Inform"},{"key":"2026063015133096300_twag017-B24","doi-asserted-by":"publisher","first-page":"226","DOI":"10.3390\/ai6090226","article-title":"Retrieval-augmented generation (RAG) in healthcare: a comprehensive review","volume":"6","author":"Neha","year":"2025","journal-title":"AI"},{"key":"2026063015133096300_twag017-B25","doi-asserted-by":"publisher","first-page":"53","DOI":"10.1259\/dmfr.18.2.2699592","article-title":"Computer-aided differential diagnosis of oral radiographic lesions","volume":"18","author":"White","year":"1989","journal-title":"Dentomaxillofac Radiol"},{"key":"2026063015133096300_twag017-B26","doi-asserted-by":"crossref","DOI":"10.1002\/9781119218739","volume-title":"Oral and Maxillofacial Radiology: A Diagnostic Approach","author":"MacDonald","year":"2019","edition":"2nd ed"},{"key":"2026063015133096300_twag017-B27","volume-title":"Diagnostic Imaging: Oral and Maxillofacial: Diagnostic Imaging: Oral and Maxillofacial E-Book","author":"Koenig","year":"2017","edition":"3rd ed"},{"key":"2026063015133096300_twag017-B28","volume-title":"Cysts of the Oral and Maxillofacial Regions","author":"Shear","year":"2008","edition":"4th ed"},{"key":"2026063015133096300_twag017-B29","volume-title":"Essentials of Oral Pathology and Oral Medicine","author":"Cawson","year":"1998","edition":"6th ed"},{"key":"2026063015133096300_twag017-B30","volume-title":"Diagnostic Surgical Pathology of the Head and Neck E-Book: Expert Consult-Online and Print","author":"Gnepp","year":"2009","edition":"2nd ed"},{"key":"2026063015133096300_twag017-B31","author":"Qdrant"},{"key":"2026063015133096300_twag017-B32","first-page":"5","author":"Robertson","year":"2025"},{"key":"2026063015133096300_twag017-B33","author":"Streamlit"},{"key":"2026063015133096300_twag017-B34","first-page":"2","article-title":"Docker: lightweight Linux containers for consistent development and deployment","volume":"2014","author":"Merkel","year":"2014","journal-title":"Linux j"},{"key":"2026063015133096300_twag017-B35","author":"Google Cloud"},{"key":"2026063015133096300_twag017-B36","doi-asserted-by":"publisher","author":"Seabold","DOI":"10.25080\/Majora-92bf1922-011"},{"key":"2026063015133096300_twag017-B37","author":"OpenAI"},{"key":"2026063015133096300_twag017-B38","doi-asserted-by":"publisher","first-page":"111495","DOI":"10.1016\/j.isci.2024.111495","article-title":"Evaluation of the ability of large language models to self-diagnose oral diseases","volume":"27","author":"Zhuang","year":"2024","journal-title":"iScience"},{"key":"2026063015133096300_twag017-B39","doi-asserted-by":"publisher","first-page":"e224","DOI":"10.4317\/medoral.26824","article-title":"Accuracy of ChatGPT 3.5, 4.0, 4o and Gemini in diagnosing oral potentially malignant lesions based on clinical case reports and image recognition","volume":"30","author":"Pradhan","year":"2025","journal-title":"Med Oral Patol Oral Cir Bucal"}],"container-title":["Dentomaxillofacial Radiology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/dmfr\/advance-article-pdf\/doi\/10.1093\/dmfr\/twag017\/67486023\/twag017.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/dmfr\/article-pdf\/55\/5\/497\/67486023\/twag017.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/dmfr\/article-pdf\/55\/5\/497\/67486023\/twag017.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,1]],"date-time":"2026-07-01T08:35:20Z","timestamp":1782894920000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/dmfr\/article\/55\/5\/497\/8538006"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,23]]},"references-count":39,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2026,3,23]]},"published-print":{"date-parts":[[2026,7,1]]}},"URL":"https:\/\/doi.org\/10.1093\/dmfr\/twag017","relation":{},"ISSN":["0250-832X","1476-542X"],"issn-type":[{"value":"0250-832X","type":"print"},{"value":"1476-542X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,7]]},"published":{"date-parts":[[2026,3,23]]}}}