Classification of Healthy Condition and Depression using functional Near Infrared Spectroscopy (fNIRS)

Depression is a serious social issue where the World Health Organisation (2017) has mentioned that there are estimated over 300 million people whom are suffering from depression that corresponded to 4.4 of the world’s population [1]. Depression also is one of the biggest contribution to suicide death. Therefore, we need to detect early depression so that we can prevent it. In this study, fNIRS has been chosen as a neuroimaging modality to establish a new biomarker for Brain-Computer Interface (BCI). The aim of this project is to investigate if fNIRS showing the haemodynamic of brain activity can help distinguish the condition of the individual either healthy condition and depression. A total of 30 participants have been asked to do the Verbal Fluency Task (VFT) within 100s and their fNIRS signal was collected concurrently. A few classifier algorithms were used to find out the relationship between the oxygenation with the condition of the participants which are support vector machine, k-nearest neighbour and naïve Bayes classifier. All these classifiers has been validate using leave one out cross validation (LOOCV) and the accuracy, specificity and sensitivity of the classes has been calculated. Finding showed that, the highest accuracy was performed by support vector machine (SVM) which is 96.43% with specificity and sensitivity: 93.75% and 100%. Based results, this study stipulates that fNIRS has a bright potential as an application for the establishment of BCIs.