Butnaru Oana-Maria, Tatarciuc Monica, Luchian Ionut, Tudorici Teona, Balcos Carina, Budala Dana Gabriela, Sirghe Ana, Virvescu Dragos Ioan, Haba Danisia
Department of Biophysics, Faculty of Dental Medicine, "Grigore T. Popa" University of Medicine and Phamacy, 700115 Iasi, Romania.
Department of Prosthodontics, Faculty of Dental Medicine, "Grigore T. Popa" University of Medicine and Pharmacy, 700115 Iasi, Romania.
Medicina (Kaunas). 2025 Mar 23;61(4):572. doi: 10.3390/medicina61040572.
Artificial intelligence (AI) is increasingly used in healthcare, including dental and periodontal diagnostics, due to its ability to analyze complex datasets with speed and precision. This study aimed to evaluate the reliability of AI-assisted dental-periodontal diagnoses compared to diagnoses made by senior specialists, specialists, and general dentists. A comparative study was conducted involving 60 practitioners divided into three groups-general dentists, specialists, and senior specialists-along with an AI diagnostic system (Planmeca Romexis 6.4.7.software). Participants evaluated six high-quality panoramic radiographic images representing various dental and periodontal conditions. Diagnoses were compared against a reference "gold standard" validated by a dental imaging expert and senior clinician. A statistical analysis was performed using SPSS 26.0, applying chi-square tests, ANOVA, and Bonferroni correction to ensure robust results. AI's consistency in identifying subtle conditions was comparable to that of senior specialists, while general dentists showed greater variability in their evaluations. The key findings revealed that AI and senior specialists consistently demonstrated the highest performance in detecting attachment loss and alveolar bone loss, with AI achieving a mean score of 6.12 in identifying teeth with attachment loss, compared to 5.43 for senior specialists, 4.58 for specialists, and 3.65 for general dentists. The ANOVA highlighted statistically significant differences between groups, particularly in the detection of attachment loss on the maxillary arch (F = 3.820, = 0.014). Additionally, AI showed high consistency in detecting alveolar bone loss, with comparable performance to senior specialists. AI systems exhibit significant potential as reliable tools for dental and periodontal assessment, complementing the expertise of human practitioners. However, further validation in clinical settings is necessary to address limitations such as algorithmic bias and atypical cases. AI integration in dentistry can enhance diagnostic precision and patient outcomes while reducing variability in clinical assessments.
由于人工智能（AI）能够快速、精确地分析复杂数据集，它在医疗保健领域的应用越来越广泛，包括牙科和牙周诊断。本研究旨在评估与高级专家、专科医生和普通牙医所做诊断相比，人工智能辅助牙科-牙周诊断的可靠性。进行了一项比较研究，60名从业者分为三组——普通牙医、专科医生和高级专家——并使用了一个人工智能诊断系统（普兰梅卡Romexis 6.4.7软件）。参与者评估了六张高质量的全景放射影像，这些影像代表了各种牙科和牙周状况。将诊断结果与由牙科影像专家和资深临床医生验证的参考“金标准”进行比较。使用SPSS 26.0进行统计分析，应用卡方检验、方差分析和Bonferroni校正以确保结果的稳健性。人工智能在识别细微状况方面的一致性与高级专家相当，而普通牙医的评估结果显示出更大的变异性。关键发现表明，人工智能和高级专家在检测附着丧失和牙槽骨丧失方面始终表现出最高的性能，人工智能在识别有附着丧失的牙齿时平均得分为6.12，相比之下，高级专家为5.43，专科医生为4.58，普通牙医为3.65。方差分析突出了各组之间的统计学显著差异，特别是在上颌弓附着丧失的检测方面（F = 3.820，P = 0.014）。此外，人工智能在检测牙槽骨丧失方面表现出高度一致性，其性能与高级专家相当。人工智能系统作为牙科和牙周评估的可靠工具具有巨大潜力，可补充人类从业者的专业知识。然而，有必要在临床环境中进行进一步验证，以解决算法偏差和非典型病例等局限性。牙科领域的人工智能整合可以提高诊断精度和患者治疗效果，同时减少临床评估中的变异性。
