Neuropsychiatric disorders classification using a video based pupil detection system

Eye tracking systems have become increasingly popular in computer vision applications. In this study, a video-based pupil detection system and neuropsychological tests for the system are developed for diagnosing of neuropsychiatric disorders. An Active Contour Snake model with ellipse fitting algorithm is proposed to extract the position (x, y) and radius (r) information of pupils in video sequences. The neuropsychological tests in this research are applied to three different groups, namely; bipolar disorder group, control group and schizophrenia group. In order to conduct the neuropsychological tests, a number of selected words are shown to the people on a computer screen. The pupil detection system is used to determine the duration and the coordinates of the screen where the eye looking at while reading the test words in real time. The information acquired from the tests is used as a feature vector in Support Vector Machines (SVM) for classifying the neuropsychiatric disorder (bipolar) and normal (control) groups. The information acquired from schizophrenia group is also presented for the same tests.