The current medical standard for setting an oral cancer (OC) diagnosis is histological examination of a tissue sample taken from the oral cavity. This process is time-consuming and more invasive than an alternative approach of acquiring a brush sample followed by cytological analysis. Using a microscope, skilled cytotechnologists are able to detect changes due to malignancy; however, introducing this approach into clinical routine is associated with challenges such as a lack of resources and experts.
Automated analysis of cytological data can increase the efficiency of the process to a level that opens possibilities for population-wide screening, towards early cancer detection. Deep learning (DL) based image classification methods have shown the ability to detect differences between malignant versus healthy samples, without the need for very difficult and time-consuming labeling of each individual cell. Systems based on such methods for oral cancer (OC) detection could assist cytotechnologists, given that the method is reliable enough. As an assisting technology to an expert, the DL-based methods should detect all questionable slides (bags) not to miss malignancy. A medical expert would then resolve challenging cases and provide a final decision. The goal of this project is to approach a relatively sensitive patient classification and detection of suspicious cells.