The use of artificial intelligence in the diagnosis of oral carcinoma
Authors
Francesco Gianfreda, Andrea Danieli, Maria Scarpati Cioffari di Castiglione , Marco Gargari, Patrizio Bollero, Mirko Martelli
Abstract
Introduction: Artificial intelligence (AI) is poised to transform the medical field significantly. Machine learning (ML) and deep learning (DL) techniques are increasingly utilized to improve the diagnosis, management, and treatment of oral lesions, including oral squamous cell carcinoma (OSCC). The high incidence and mortality associated with OSCC underscore the critical need for early and accurate diagnostic tools. AI offers advanced methodologies for enhanced diagnostic precision and operational efficiency, addressing the limitations of traditional techniques. Materials and Methods: This scoping review systematically explores current AI applications focused on oral lesions. Studies included addressed AI in diagnosing and managing OSCC, published in English or Italian. A comprehensive search was conducted across electronic databases (PubMed, Scopus, and Web of Science). The data regarding study characteristics, AI types, methodologies, and outcomes were extracted. Discussion: AI shows immense potential in optimizing oral carcinoma diagnosis and treatment processes, as evidenced by various studies. For instance, deep learning techniques have demonstrated high accuracy in identifying oral carcinoma and analyzing histological data. AI systems can streamline diagnosis, dramatically reducing waiting times and enhancing reliability. Additionally, AI applications extend to prognosis and staging, with methodologies showcasing promising accuracies for the clinical management of neoplasms. Results: Integrating AI in the diagnosis of oral lesions has resulted in accuracy rates exceeding 99% and significantly reduced diagnostic waiting times. Machine learning models effectively detect subtle anomalies and offer diagnostic performance that matches or surpasses expert clinical pathologists. The use of automatic classifiers has improved early diagnosis and operational efficiency in clinical practice, addressing traditional challenges and reducing diagnostic errors. Conclusions: Incorporating AI in diagnosing and managing oral lesions presents significant opportunities for improving clinical outcomes with high accuracy and speed. Continuous research and development are needed to refine these systems and ensure their effective implementation in clinical practice. Collaboration among interdisciplinary teams is crucial for ensuring data security and meeting the needs of healthcare professionals in future AI developments.