Estimation and modelling of fMRI BOLD response

One of the current topics of research in neuroimaging techniques is related to explaining and modelling the Blood Oxygen Level Dependent (BOLD) responses. BOLD responses are estimated by processing functional Magnetic Resonance Imaging (fMRI) data. BOLD responses are caused by hemodynamic responses to neural activity which alter the levels of blood oxygenation at local brain regions. The main aims of the current thesis were to i) develop and examine methods regarding BOLD response estimation from the visual cortex and the frontal cortex of human brain and to ii) develop a model in order to explain the physiological mechanisms which cause the estimated BOLD responses. In order to satisfy the main aims, fMRI data were provided by the Center of Medical Imaging and Visualization (CMIV). The provided fMRI data consist of fMRI brain measurements of twelve healthy human subjects who were subjected to visual stimulation. By processing the fMRI data, Regions Of Interest (ROIs) were extracted at the anatomical sites of the visual cortex and the frontal cortex. Afterwards, the fMRI data were manipulated in order to extract BOLD responses from the visual cortex and the frontal cortex. Various methods were developed and compared in terms of which technique provided well representative BOLD responses. Subsequently, a model was developed by using software Wolfram Mathematica 9 in order to explain the physiological mechanisms of the estimated BOLD responses at the visual and the frontal cortex. The model aimed to solve for oxygen concentration in blood plasma as blood flows from the arterial part to the venous part of the blood circulation system through a capillary. Oxygen outward diffusion through the capillary wall and oxygen concentration at the extravascular environment were modelled as well. Blood plasma oxygen concentration was turned into hemoglobin oxygen saturation (Saí µí±‚ 2) through hemoglobin oxygen dissociation curve and Henry's law for gases. As a result, the Saí µí±‚ 2 was estimated through modelling for oxygen concentration in blood plasma. Finally, the developed model ended to a system with input the fractional change of Cerebral Blood Flow (CBF) velocity and Cerebral Metabolic Rate of Oxygen (CMRí µí±‚ 2) and as output a proportional signal to the BOLD response. By simulating for different scenarios of fractional changes of CBF velocity and CMRí µí±‚ 2 and by comparing the resulted BOLD responses to the estimated ones, it was attempted to explain for the physiological mechanisms which caused the BOLD responses at …