Discovering knowledge through machine learning model explainability:
How do we estimate age and sex from panoramic dental X-ray images
Abstract
Explainable AI (XAI) methods are crucial for understanding and validating
deep learning models, especially in high-stakes domains like medical imaging.
However, interpreting the complex visual outputs of XAI techniques remains challenging.
This paper introduces a novel XAI tool that combines YOLO object
detection with SHAP analysis to automatically identify anatomical regions influencing model decisions
in panoramic dental X-rays. We apply our tool to age and sex estimation models trained on a 943
Thai patient X-ray dataset. By aggregating results across images, we learn which dental structures are most
salient for different predictions. We validate our findings against existing dental research,
confirming known age and sex indicators and identifying potentially novel biomarkers.
Our approach enables more interpretable and actionable XAI for medical imaging,
bridging the gap between model explanations and domain expertise.
We demonstrate how XAI can explain individual predictions and discover generalizable insights to advance scientific
understanding when applied systematically across datasets.
Awards
Gold prize
Young Rising Star of Science 2024 (YRSS)
Silver medal
Higher Education Innovation Awards 2024
BibTeX
title = {DeepTooth: Estimating Age and Gender with Panoramic Radiograph Image},
author = {Somdej, Wanita and Thamvongsa, Athitiya and Hirunchavarod, Natthanich and Sributsayakarn, Natnicha and Pornprasertsuk-Damrongsri, Suchaya and Jirarattanasopha, Varangkanar and Intharah, Thanapong},
booktitle = {2023 International Technical Conference on Circuits/Systems, Computers, and Communications (ITC-CSCC)},
pages = {1--6},
year = {2023},
organization={IEEE}
}
title = {Deeptoothduo: Multi-task age-sex estimation and understanding via panoramic radiograph},
author = {Hirunchavarod, Natthanich and Phuphatham, Pornnakanok and Sributsayakarn, Natnicha and Prathansap, Narawit and Pornprasertsuk-Damrongsri, Suchaya and Jirarattanasopha, Varangkanar and Intharah, Thanapong},
booktitle = {2024 IEEE International Symposium on Biomedical Imaging (ISBI)},
pages = {1--5},
year = {2024},
organization={IEEE}
}
title = {OPG-SHAP: A Dental AI Tool for Explaining Learned Orthopantomogram Image Recognition},
author = {Hirunchavarod, Natthanich and Dangsungnoen, Lapatrada and Thongprasant, Kwansawan and Phuphatham, Pornnakanok and Prathansap, Narawit and Sributsayakarn, Natnicha and Pornprasertsuk-Damrongsri, Suchaya and Jirarattanasopha, Varangkanar and Intharah, Thanapong},
booktitle = {2024 International Technical Conference on Circuits/Systems, Computers, and Communications (ITC-CSCC)},
pages = {1--6},
year = {2024},
organization={IEEE}
}
The website template was borrowed from Michaël Gharbi.