Raman Spectroscopy of Cells for Cancer Classification Through Machine Learning.

The term cancer indicates a pathological condition characterized by the uncontrolled proliferation of cells that have the ability to infiltrate the normal organs and tissues of the body, altering their structure and functioning. Therefore, since cancer is caused by DNA mutations within cells, Raman spectroscopy can be a valuable tool for gathering information about their composition. With this technique, a sample is illuminated by a beam of monochromatic light and the interaction between them produces an effect that allows to obtain information on the sample examined. This study aims to combine Raman spectroscopy with artificial intelligence to develop a model capable of distinguishing cancerous cells from healthy ones. In this regard, the experiments were conducted on a data set provided by the Center for Nanophotonics and Optoelectronics for Human Health (CNOS), which analyzed the cells of a patient suffering from liver cancer. Specifically, the dataset was created through a lengthy data collection process, which involved first analyzing the cells with spectroscopy and then training several machine learning, tree-based, and boosting classifiers to distinguish cancer cells from healthy ones. The main contribution of the work consists in using genetic algorithms to select the most significant frequencies. The best results are obtained using Extra Tree Classifier reaching a value of F-score up to 91%.