Routine Diagnosis of Dental Caries on RVG Images with Perplexity AI

Abstract

Background: Artificial intelligence (AI)–based digital interpretation tools are increasingly incorporated into dental practice to enhance efficiency and support clinical decision-making. While these systems provide rapid outputs, their diagnostic accuracy in comparison with conventional manual interpretation remains a subject of debate. AI-based digital interpretation helps in reducing the time required for image analysis and improves workflow efficiency in dental clinics. Manual interpretation allows detailed visual assessment and clinical correlation, which contributes to higher diagnostic accuracy. Digital interpretation provides rapid results, making it useful in busy clinical settings. Accuracy of digital tools may vary depending on image quality, software algorithms, and calibration. Manual interpretation remains the gold standard due to the examiner’s experience and ability to detect subtle findings.  AI tools may sometimes miss fine details or produce incorrect interpretations in complex cases. Digital interpretation can be helpful as a supporting tool for screening and preliminary diagnosis.  Combining AI-based interpretation with clinician expertise may improve overall diagnostic performance. Proper training and validation of AI software are essential before routine clinical use.  Further studies with larger sample sizes are required to improve the reliability of AI-based systems.  Continuous updates and improvements in AI algorithms may enhance diagnostic accuracy in the future.

Aim: To compare manual interpretation with digital (AI-based) interpretation in dental diagnosis with respect to interpretation time and diagnostic accuracy.

Materials and Methods: A comparative analytical study was conducted using 100 dental samples. Each sample was evaluated using two approaches: manual interpretation performed by a trained dental examiner and digital interpretation using an AI-based tool (Perplexity AI). The time required for interpretation and diagnostic accuracy were recorded for both methods. Results were analyzed descriptively and comparatively.

Results: Manual interpretation demonstrated a diagnostic accuracy of 100%, with an average interpretation time of approximately 1 minute per sample. Digital interpretation significantly reduced interpretation time to approximately 0.5 seconds per sample; however, diagnostic accuracy was limited to 80%. Manual interpretation showed superior accuracy, whereas digital interpretation showed superior time efficiency.

Conclusion: Although AI-based digital interpretation offers a substantial reduction in interpretation time, it demonstrates lower diagnostic accuracy compared to manual interpretation. Manual assessment remains the gold standard, particularly for the detection of dental caries. Digital AI tools should be considered adjunctive aids rather than replacements for clinician-based diagnosis.