AI-Driven Quantitative Dental Imaging: A Clinical Framework for Assessing Root Resorption Across Treatment Modalities

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Orthodontically induced external root resorption (ERR) can compromise long-term tooth stability.
• Accurate assessment of root length changes is crucial for evaluating treatment outcomes.
• Existing methods for quantifying root resorption can be time-consuming and subjective.

Purpose of the Study:

• To develop and validate an AI-based framework for automatic segmentation and quantitative evaluation of root resorption.
• To compare root resorption between fixed appliance therapy and clear aligner therapy using AI analysis.
• To assess the reliability and reproducibility of AI-derived measurements against expert annotations.

Main Methods:

• A dataset of 100 anonymised orthopantomographic (OPG) radiographs (50 fixed appliance, 50 clear aligner) was used.
• A U-Net model was trained for automatic tooth segmentation and root length assessment.
• AI-predicted measurements were compared with dentist-annotated masks using pixel-based Python analysis.

Main Results:

• The AI segmentation model achieved high performance (IoU up to 89%, Dice up to 95%).
• Significant root length reduction was observed in both treatment groups (p < 0.0001).
• Clear aligner therapy group showed significantly less root resorption (AI: 2.81%, Human: 4.98%) compared to the fixed appliance group (AI: 8.95%, Human: 10.16%).

Conclusions:

• AI analysis of OPGs offers a reliable method for quantifying orthodontically induced root resorption.
• Clear aligner therapy appears associated with reduced root length reduction compared to fixed appliances.
• Further multi-center studies and external validation are necessary for clinical implementation.