Study On Data Processing Algorithms And Applications In Brain Functional Magnetic Resonance Imaging

The functional magnetic resonance imaging (fMRI), with high spatial resolution, is an important nondestructive examination tool of detecting human brain and has drawn great attention in related fields. The fMRI method of acquiring brain function information is pivotal in integrating fMRI with brain cognition, neuro-science, and clinical application. This technology still is not quite mature now and still needs improving.The present study focuses on such issues in brain science as the fMRI data processing algorithms of the locating and clustering of the brain functional activation areas, the brain functional asymmetry and so on, and proposes adopting the frequency information of fMRI time series, phase spectrum (PS) information, convolution power spectrum (CPS), and affinity propagation clustering (APC) to detect brain functional activity, with improvement to the time-frequency fMRI signal analysis method which has strong anti-interference capability. The study further raises the idea of exploring functional asymmetry in the motor cortex through quantitative analysis of the changes of power spectrum. Analysis of the simulation and the real data demonstrate that these methods and technical innovations are effective and applicable. The details are shown as follows:1) The fMRI data processing method for temporal information often neglects the frequency information of the time series. In this paper, a new method is proposed to detect the functional activation regions in brain by using the frequency information of fMRI time series in the Hilbert space. The main idea is that the frequency entropy information (FEI) difference of fMRI data between task and control states is specified as brain activation index. Simulation is conducted to confirm the validity of the proposed approach. The comparison of receiver operating characteristic (ROC) curves acquired respectively from the proposed scheme, the statistical parametric mapping (SPM), and the support vector machine (SVM) methods of fMRI data analysis indicates an obvious superiority of the frequency information method. The in vivo fMRI studies too reveal that this method can enable the effective detection of brain functional activation.2) A phase spectrum method is presented to identify brain functional activation areas via using the phase information of fMRI time series. The basic idea is that the phase at the characteristic frequency of fMRI signal is specified as brain activation index. The developed phase approach is tested and confirmed by the result from both simulation and in vivo fMRI data.3) The conventional fMRI data-processing method aims at modeling the blood oxygen level-dependent (BOLD) response of voxels as a function of time. But the theory of power spectrum analysis focuses completely on the understanding the dynamic energy change of interacting systems. This study therefore proposes a new CPS analysis of fMRI data, based on the theory of prior image signal, to detect brain functional activation for fMRI data. First, convolution signals are computed between the measured fMRI signals and the image signal of prior experimental pattern to suppress noise in the fMRI data. Then, the power spectrum density (PSD) analysis of the convolution signal is specified as the quantitative analysis energy index of BOLD signal change. The data from simulation studies and in vivo fMRI studies, including block-design experiments, reveal that the CPS method enables a more effective detection of some aspects of brain functional activation, as compared with the canonical power spectrum, SPM and SVM methods. Our results demonstrate that the CPS method with strong anti-noise capability is useful as a complementary analysis in revealing brain functional information regarding the complex nature of fMRI time series.4) Clustering analysis is a quite good data-driven method for the analysis of fMRI time series. The huge computation load, however, makes it difficult for practical use. In view of this, proposal is made to use APC, an efficient and fast new clustering algorithm especially for large data sets to detect brain functional activation from fMRI. It considers all data points as possible exemplars through the minimization of an energy function and message passing architecture, and obtains the optimal set of exemplars and their corresponding clusters instead of randomly choosing initial exemplars. Four simulation studies and three in vivo fMRI data sets containing both block-design and event-related experiments reveal that brain functional activation can be effectively detected and that different response patterns can be distinguished using this method. The performance measures in the average squared error show that APC is clearly superior to the general k-centers cluster methods in block-design and event-related experiments. The results of research demonstrate that through fMRI data, APC algorithm can effectively detect and cluster brain function areas.5) This paper proposes analyzing quantitatively power changes in BOLD signals to investigate functional asymmetry of cortical activity in motor areas. Six right-handed subjects are included in the fMRI experiments. Both bi-handed and single-handed movements are analyzed. The power spectrum method demonstrated that right-handed subjects exhibited a larger power difference in BOLD signals between task and rest states in the right motor area than in the left motor area. These results show that more nerve cells are evoked in the right motor area. In addition, in contrast with the signal magnitude analysis, reasnablness analysis is conducted of the power spectrum method involved in the detection of functional asymmetry. The power spectrum method is confirmed to be a valid quantitative-analysis method for brain asymmetry analyses.