Research On Functional Magnetic Resonance Imaging Processing

Functional magnetic resonance imaging (fMRI) is a new technology which is developed to explore the functional activities of brain in recent years. It takes three elements into consideration such as function, image, and anatomy. It is an effective approach to orientate each functional region of the living body in human brain. fMRI is a kind of non-invasive and non-radialized technology and it has high temporal and spatial resolution and can be operated repeatedly as well. Therefore, fMRI has already become an important method today to do researches on brain and life sciences.With the development of fMRI technology, a lot of fMRI data processing techniques have been presented and lots of achievements have been reached. At present, fMRI data processing techniques include two types: model-driven and data-driven. This paper mainly studies to extract the brain active regions of the fMRI on the basis of the data-driven processing method.Some in-depth studies on independent component analysis (ICA) and time cluster analysis (TCA) based on data-driven method are discussed in this paper. Simulation data of fMRI are designed to simulate and verify Infomax and FastICA algorithms of ICA and the experimental results are satisfactory. These two algorithms are further to be verified by activating data with vision and the results are good comparing with those obtained by processing using SPM software. Additionally, Infomax algorithm and FastICA algorithm are compared and evaluated in the aspects such as time accuracy and spatial accuracy.Exhaustive researches are undertaken on the original time clustering analysis (OTCA) and modified time clustering analysis (MTCA) of TCA technology, and a new method of the neighborhood-local-correlation TCA is put forward. This new method can overcome the shortcoming of being so sensitive to noises about existed TCA technologies, and can reduce the effects produced by noises on exploring functional active regions effectively. In the paper, OTCA, MTCA and the neighborhood-local-correlation TCA algorithms are simulated using emulated fMRI data and actual fMRI data. And the compared experimental results show that the neighborhood-local-correlation TCA algorithm is much more effective than the other two algorithms.