Early Diagnosis of Oral Cancer Using Image Processing and Artificial Intelligence

Eman Shawky Mira^*1, Ahmed M. Saaduddin Sapri ², Rowaa F. Aljehanı³, Bayan S. Jambı⁴, Taseer Bashir⁵,  El-Sayed M. El-Kenawy⁶, Mohamed Saber⁷

¹ Diagnosis & Oral Radiology Department, Faculty of Dentistry , Mansoura University,

Mansoura 35516, Egypt

² Oral and Maxillofacial Surgery Department, Faculty of Dentistry, Mansoura University,

Mansoura 35516, Egypt  

^3,4 Dentistry Program, Batterjee Medical College, Jeddah, Saudi Arabia

⁵ Department of Oral medicine and radiology, Batterjee Dental College, Jeddah,

 Saudi Arabia

⁶ Department of Communications and Electronics, Delta Higher Institute of Engineering and Technology, Mansoura 35111, Egypt

⁷ Electronics and Communications Engineering Department, Faculty of Engineering, Delta University for Science and Technology, Gamasa City 11152, Egypt

Emails: emanshawky@mans.edu.eg; ahmedsaaduddin@mans.edu.eg; Ro2a_aljehani@hotmail.com; bayanja11@gmail.com; taseermds2013@gmail.com; skenawy@ieee.org; Mohamed.Saber@deltauniv.edu.eg

Abstract

There has yet to be a comprehensive investigation on enhancing the diagnostic accuracy of oral disease using handheld smartphone photographic photos. To overcome the difficulties associated with the automatic detection of oral illnesses, we describe an approach based on smartphone image diagnosis powered by a deep learning algorithm. The centered rule method of image capture was offered as a quick and easy way to get high-quality pictures of the mouth. A resampling method was proposed to mitigate the influence of image variability from handheld smartphone cameras, and a medium-sized oral dataset with five types of disorders was developed based on this approach. Finally, we introduce a recently developed deep-learning network to assess oral cancer diagnosis. On 455 test images, the proposed technique showed an impressive 83.0% sensitivity, 96.6% specificity, 84.3% accuracy, and 83.6% F1. The proposed "center positioning" method was about 8% higher than a simulated "random positioning" method; the resampling process had an additional 6% performance improvement. The performance of a deep learning algorithm for detecting oral cancer can be enhanced by capturing oral photos centered on the lesion. Primary oral cancer diagnosis using smartphone-based images with deep learning offers promising potential.

Keywords: Deep learning; Smartphone-based imaging; Image collection; Oral cancer diagnosis; Oral potentially malignant disorders.