A Comprehensive Study on Artificial Intelligence Techniques for Oral Cancer Diagnosis: Challenges and Opportunities
R. Sathishkumar, M. Govindarajan
Abstract
The malignant nature of oral cancer is made evident by its low average chance of survival. Oncology and surgical treatment modality advancements have only modestly improved outcomes. The rapid progression and metastasis of this type of cancer, the lack of primary symptoms, the large number of risk factors, and the general lack of public knowledge of oral malignancy are all contributing causes to its high fatality rate. Recently, it has been suggested to use artificial intelligence (AI) methods for screening for oral cancer. By using AI techniques for identifying oral carcinoma, this study collects reviewed data from possibly accessible databases and different oral cancer diagnosis techniques, evaluates them against established techniques, and gives technical information with major conclusions. We also go through some difficulties, inefficiencies, and potential future research areas for the various medical image analysis methods. A subfield of AI called ML and DL has arisen during the past ten years, and its applications in the healthcare sector have produced remarkable outcomes at lower costs and higher efficiency. For the period of 2020 to 2023, this study investigates several datasets, algorithms, modalities, and key findings used in ML and DL methodologies for the identification of oral cancer. Deep learning beats traditional machine learning for identifying oral carcinoma in large datasets, according to results from prior studies. Research gaps identified in current studies suggest that practical and scientific research is urgently required for future healthcare improvement.