Page 93 - AIH-1-3

Artificial Intelligence in Health Dental cavity prediction with computer vision

redefines the way we diagnose and assess oral health. The Ethics approval and consent to participate
YOLO model ensemble achieved a notable mAP of 0.732, Not applicable.
an accuracy of 0.789, and a recall of 0.701. Considering
that this method identifies cavities directly from standard Consent for publication
device camera photographs, this accuracy is remarkable. Not applicable.
Our approach, comprising precise localization, accurate
quantification, and nuanced visualization, demonstrates Availability of data
its potential to improve dental health assessments to
unprecedented levels of accuracy and efficiency. Through The image data can be obtained from Kaggle (https://
meticulous augmentation and annotation of a colored www.kaggle.com/datasets/salmansajid05/oral-
dental image dataset, we harnessed the power of the diseases?resource=download-directory).
YOLOv5 model for dental cavity localization, providing References
bounding box predictions with remarkable accuracy. The
introduction of the SAM brought the ability to focus with 1. Pruthi S. A Window to Your Overall Health, Oral Health.
surgical precision on dental cavities, enriching our analysis United States: Mayo Clinic; 2024.
and empowering dental professionals with detailed 2. Global Burden of Disease (GBD). Institute for Health
visualizations for diagnosis. Our innovative quantification Metrics and Evaluation, IHME; 2019.
methodology, which extracts cavity areas from bounding 3. Kim SM, Jang WM, Ahn HA, Park HJ, Ahn HS. Korean
box coordinates, offers a quantitative edge, enhancing National Health Insurance value incentive program:
diagnostic insights. The potential of our research is Achievements and future directions. J Prev Med Public
promising. We foresee the continued evolution of dental Health. 2012;45:148-155.
health assessments through the fusion of technology and
health care. In the future, our framework could pave the doi: 10.3961/jpmph.2012.45.3.148
way for automated dental check-ups, reducing the burden 4. Maru AM, Narendran S. Epidemiology of dental caries
on both patients and health-care professionals. Moreover, among adults in a rural area in India. J Contemp Dent Pract.
as technology advances, we envision our methods 2012;13(3):382-388.
becoming even more accurate and efficient. Remote dental doi: 10.5005/jp-journals-10024-1155
diagnostics, tele-dentistry, and improved oral health-care 5. Shankar V, Narang U. Emerging market innovations:
access may become more widespread, particularly in Unique and differential drivers, practitioner implications,
underserved areas. and research agenda. J Acad Mark Sci. 2020;48:1030-1052.
Acknowledgments doi: 10.1007/s11747-019-00685-3

None. 6. Silvertown JD, Wong BP, Abrams SH, Sivagurunathan KS,
Mathews SM, Amaechi BT. Comparison of the canary
Funding system and DIAGNOdent for the in vitro detection of caries
under opaque dental sealants. J Investig Clin Dent. 2017;8(4).
None.
doi: 10.1111/jicd.12239
Conflict of interest 7. Almalki YE, Imam Din A, Ramzan M, et al. Deep

The authors declare that they have no competing interests. learning models for classification of dental diseases using
orthopantomography X-ray OPG images. Sensors (Basel).
Author contributions 2022;22(19):7370.
Conceptualization: Mohammad Aqeel, Payam doi: 10.3390/s22197370
Norouzzadeh, Abbas Maazallahi, Eli Snir, Bahareh 8. Retrouvey JM, Conley RS. Decoding deep learning
Rahmani applications for diagnosis and treatment planning. Dent
Investigation: Mohammad Aqeel, Golnesa Rouie Miab, Press J Orthod. 2023;27.
Laila Al Dehailan, David Stoeckel, Bahareh Rahmani doi: 10.1590/2177-6709.27.5.e22spe5
Methodology: Mohammad Aqeel, Payam Norouzzadeh, 9. Kumar S, Kumar H. Analysis of image segmentation
Salih Tutun, Eli Snir, Bahareh Rahmani techniques for dental radiography. Element Educ Online.
Writing – original draft: Mohammad Aqeel 2021;20(4):3868-3875.
Writing – review & editing: Payam Norouzzadeh, Abbas
Maazallahi, Salih Tutun, David Stoeckel, Eli Snir, doi: 10.17051/ilkonline.2021.04.422
Bahareh Rahmani 10. Tareq A, Faisal MI, Islam S, et al. Visual diagnostics of

Volume 1 Issue 3 (2024) 87 doi: 10.36922/aih.3184