Systematic Literature Review on Early Diagnosis of Oral Squamous Cell Carcinoma by Deep Learning Techniques

Oral Squamous Cell Carcinoma (OSCC) is the seventh most prevalent type of cancer in the neck and head. The prognosis and survival rate of the patient is significantly improved by early identification of OSCC. Due to tumor heterogeneity, such a diagnosis requires much time and a high-efficiency human experience. As a result, artificial intelligence systems assist professionals and physicians in making precise diagnoses. Recent advances in computer vision-based techniques and Computational Intelligence (CI) improve accuracy in medical images. This study aims to develop hybrid methodologies based on fused features to produce excellent outcomes for the early detection of OSCC. This systematic review aims to estimate deep learning (DL) based algorithms for early diagnosis OSCC to assist clinicians in oral cancer diagnosis and screening. The terms "squamous cell carcinoma," "early diagnosis," "oral cavity," "histopathological image," "biomarker," "Optical Coherence Tomography (OCT) image," and "deep learning" were used in a Google Scholar Cochrane and MEDLINE (PubMed), Embase and WoS databases (January 2018 to July 2023) to find relevant articles. The inclusion criteria were the use of deep learning approaches for early diagnosis of OSCC, articles older than 5 years, and publications written in English. Case reports and studies written in foreign languages met the exclusion criteria. 70 publications were chosen to be included in the systematic review out of the 194 studies that were initially found through the search. Deep learning techniques based on hybrid features are examined in evaluating performance metrics and superior results of the present systems employing biomedical images for OSCC diagnosis. It has been established that the deep learning-based early identification method for biomedical images has the ability to offer decision support for efficient oral cancer diagnosis and screening.