Date of Award
Spring 4-22-2026
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Dentistry
First Advisor
Stephanie Sidow
Second Advisor
Manal Hamdan
Third Advisor
Zaid Badr
Abstract
Objective: To evaluate the effect of low-code no-code (LCNC) artificial intelligence (AI) as a diagnostic aid for dental students in detecting periapical radiolucencies (PR) on periapical radiographs (PAX), compared with experienced clinicians. Methods: Nineteen participants, ten fourth-year dental students and nine experienced clinicians (endodontists and endodontic residents), assessed 140 teeth on 50 PAXs in two sessions separated by a three-month washout period. CBCT volumes and radiology reports served as the reference standard for confirming the presence or absence of periapical radiolucent lesions. In session 1, images were evaluated without AI assistance. During session 2, participants re-evaluated the same images, followed by interpretation with AI assistance. Diagnostic performance metrics including sensitivity, specificity, accuracy, precision, and F1 score were calculated, and paired t-tests were used to assess the impact of AI on diagnostic performance. Results: Among dental students, all diagnostic performance metrics improved significantly with AI assistance: sensitivity (69.0% vs. 87.2%), specificity (87.4% vs. 96.7%), accuracy (83.4% vs. 94.7%), precision (65.6% vs. 87.9%), and F1 score: (65.5% vs. 87.1%) (p< 0.05). AI assistance significantly reduced diagnostic time across all groups. Conclusion: AI-assisted interpretation of PAXs significantly enhanced diagnostic accuracy, sensitivity, and efficiency for both students and clinicians. These findings support AI’s potential as an educational tool for improving diagnostic accuracy and consistency among dental students and as a valuable clinical adjunct to reduce variability and support clinical decision-making for experienced clinicians.