fMRI SİNYALLERİNDE DEĞİŞİM NOKTALARININ TAHMİNLENMESİ
Project no. 190 · Status: Kapanmış · 2017
Summary
Today, functional magnetic resonance imaging (fMRI) studies are frequently used in both clinical and research fields. Functional imaging aims to associate a given stimulus with the activation regions that related to the stimulus. However, since the brain is constantly working, in other words being the areas that are constantly active, it is a very difficult problem to be able to associate the activation with the stimulus. Therefore, fMRI experiments are used. The experiments include various tasks that required to make by the subjects. In order to determine the activation, the fMRI experiments consist of two stages with and without the function to be sought. Afterward, these instances are compared with each other in a statistical manner and determined the activated regions. It is also important to find the instances the activations trigger and occur as much as finding the activation areas. These instances cannot be determined by the commonly used tools for fMRI analysis. Furthermore, it is difficult to determine the location of the activation time in a time series due to the differences during the fMRI scan as well as the interpersonal differences. In this project, it is studied to determine the instances through change point analyzing methods. In this context, a regression based and a statistical based point detection method is proposed to estimate the activation instances by handling as change points. The methods are applied a data set that commonly used in change point detection studies as well as three different fMRI data sets. fMRI data sets consist of motor, visual and linguistic tasks. The analyses show that the activation instances are with high accuracy and activation regions are in accordance with the methods in the literature. Keywords: fmri, medical image analysis, change point detection problem, regression