iTRAQ-based comparison of proteins derived from oral cells collected by brush biopsy. Protein abundance levels compared between oral pre-malignant cells, oral cancer cells and healthy normal cells, all collected from human patients. Two separate iTRAQ labeled biological replicate analyses were conducted.

Cancers of the oral cavity are one of the most prevalent cancers worldwide. With highest incidence in developing countries in Asia, they have shown huge variation in geographical distribution. In Taiwan, oral cancer is the 6th most common cancer overall and the 4th most common cancer among men.

Oral cavity cancer often results in impaired functions and esthetics like difficulty in breathing, swallowing, mastication, speech along with disfigurement of the face. Advanced stage oral cancer is associated with high mortality rates and poses a serious threat to public health. Around 85% of these malignant lesions are diagnosed as oral squamous cell carcinoma (OSCC) and this diagnosis is confirmed through tissue biopsy. Although currently considered as gold standard, histopathological classification and tissue biopsies (surgical biopsy, punch biopsy, lymph node biopsy, brush biopsy, and needle aspiration biopsy) are invasive, time consuming and cannot be repeated frequently. Further, prediction of dysplastic oral lesions and OSCC only through visual inspection is challenging and demands significant training and expertise. It is therefore crucial to explore alternative diagnostic/screening tools like Artificial Intelligence (AI) and Machine Learning (ML) tools for image classification.

Currently, deep learning (DL) is a dominant machine learning (ML) technique establishing new capability standards for image classification in a variety of domains including medicine. DL has shown promising results in several image-based medical screening and diagnostic applications. It is important to learn the progress and development of work in the literature on using ML especially DL techniques for automatic oral cancer image analysis/classification. To help understand what has been done, identify the challenges and opportunities, address its research gap, and improve its status of the art, we propose to conduct a scoping review on this topic.

Study Objective: This scoping review aims at mapping the studies and synthesizing the evidence on the artificial intelligence and/or machine learning techniques available for the classification of intraoral images to identify oral cancers or precancers. Furthermore, the review will also explore and compare the performance of the models/techniques when used for classification of oral cancer and/or precancer and how they aid in diagnosis made visually. The research gap will be identified and will be used to provide direction for future studies in this area.

Sensitivity and specificity of the automated CellScope vs. cytology.

Sensitivity and specificity of the automated CellScope vs. histology.

Clinical and pathological diagnosis of patients.

Sensitivity and specificity of conventional cytology vs. histology.

Determination of kappa value between the two pathologists.

Early detection of oral cancer necessitates a minimally invasive, tissue-specific diagnostic tool that facilitates screening/surveillance. Brush biopsy, though minimally invasive, demands skilled cyto-pathologist expertise. In this study, we explored the clinical utility/efficacy of a tele-cytology system in combination with Artificial Neural Network (ANN) based risk-stratification model for early detection of oral potentially malignant (OPML)/malignant lesion. A portable, automated tablet-based tele-cytology platform capable of digitization of cytology slides was evaluated for its efficacy in the detection of OPML/malignant lesions (n = 82) in comparison with conventional cytology and histology. Then, an image pre-processing algorithm was established to segregate cells, ANN was trained with images (n = 11,981) and a risk-stratification model developed. The specificity, sensitivity and accuracy of platform/ stratification model were computed, and agreement was examined using Kappa statistics. The tele-cytology platform, Cellscope, showed an overall accuracy of 84–86% with no difference between tele-cytology and conventional cytology in detection of oral lesions (kappa, 0.67–0.72). However, OPML could be detected with low sensitivity (18%) in accordance with the limitations of conventional cytology. The integration of image processing and development of an ANN-based risk stratification model improved the detection sensitivity of malignant lesions (93%) and high grade OPML (73%), thereby increasing the overall accuracy by 30%. Tele-cytology integrated with the risk stratification model, a novel strategy established in this study, can be an invaluable Point-of-Care (PoC) tool for early detection/screening in oral cancer. This study hence establishes the applicability of tele-cytology for accurate, remote diagnosis and use of automated ANN-based analysis in improving its efficacy.