Quantile graphs in EEG signal analysis for neurological disease classification

The electroencephalogram (EEG) is a non-invasive instrumental test that allows us to examine the functioning of the brain by analyzing and recording its electrical activity. It is carried out by placing electrodes on the head corresponding to specific areas of the brain. The electrodes detect the electrical impulses and transmit them to a machine capable of translating them into a trace that is then printed on paper or transferred to an electronic medium. Typically, once the EEG signal has been acquired, a pre-processing phase is carried out to prepare and organize the data before the machine learning process begins. Next, feature extraction/selection, used together or alone, serves to create/select the most informative features to train the classifier in the final phase of the method. In this seminar, the quantile graph (QG) map method will be presented, which is used to convert a time series into a complex network by dividing the EEG signal into quantiles. In this way it is possible to represent on the network the signal changes caused by the disease compared to healthy subjects (HC). This method has shown promising results in differentiating patients from HC from a complex network perspective, achieving high levels of accuracy in dementia datasets