C.E. Credit. Artificial Intelligence Applications for the Radiographic Detection of Periodontal Disease: A Scoping Review

• Main Authors: Anne Miller (Author), Chunbo Huang (Author), Erica R. Brody (Author), Rafael Siqueira (Author)
• Format: Book
• Published: Taylor & Francis Group, 2023-12-01T00:00:00Z.
• Subjects: article
• Online Access: Connect to this object online.
• Notes: 10.1080/19424396.2023.2206301; 1942-4396
• Keywords: Periodontitis; artificial intelligence; deep learning; diagnostic imaging; computer-assisted radiographic image interpretation; Dentistry; RK1-715
• Source: Journal of the California Dental Association, Vol 51, Iss 1 (2023)

ABSTRACTBackground Calculating radiographic bone loss (RBL) can be time-consuming, labor-intensive, and examiner dependent. Artificial intelligence (AI) models have been developed to automate the detection of RBL and the risk of developing periodontal disease and tooth loss. The aim of this scoping review was to identify and evaluate the performance of AI models in diagnosing periodontal disease by detecting RBL.Types of Studies Reviewed Medline, Embase, Web of Science, Dental and Oral Sciences Source, Cumulated Index to Nursing and Allied Health Literature, ProQuest Dissertations & Theses, ClinicalTrials.gov, and MedRxiv were searched to collect evidence about AI models utilized to detect RBL and diagnosis of periodontal disease. Qualitative analysis was conducted to describe the performance and limitations of AI models for panoramic radiographs, periapical images, and cone beam computed tomography (CBCT).Results Nineteen studies met the inclusion criteria out of 322 articles identified. Mean accuracy for panoramic radiographic AI models ranged from 63% to 94%. Automated models based on periapical radiographs yielded results ranging from precision as low as 25% for detecting mild disease to a high accuracy of 99% in radiographic bone level staging performance. Periodontal bone loss sensitivity on CBCT ranged from 45% to 72%, while periodontal bone loss specificity was 81% to 83%.Practical Implications AI systems can be a helpful initial approach to screening radiographic images for periodontal disease, given that research shows these diagnostic methods might help provide a more accurate evaluation than examiner review alone. AI models developed for the diagnosis of periodontal disease using RBL require further development to accurately and reliably assess RBL and determine risk of periodontal disease independent of clinician assessment.Continuing Education Credit Available: The practice worksheet is available online in the supplemental material tab for this article.A CDA Continuing Education quiz is online for this article: https://www.cdapresents360.com/learn/catalog/view/20.