The Analysis Of Resting-state FMRI Data On Sub-acute Stroke

Most of functional magnetic resonance imaging (fMRI) detects the reaction region in brain using the blood oxygen-level dependent method. Comparing with computer tomography technology, fMRI has no radioactivity and is more; adaptive to clinic application. fMRI can also avoid the risk of using contrast agent, such as nephrogenic systemic fibrosis compared with dynamic susceptibility contrast technology. People have begun applying nondestructive fMRI to imaging detection of stroke patients in recent years internationally. However, the reliability and repeatability of these methods need to be verified.We studied the subacute ischemic stroke patients data in this paper. The BOLD data were preprocessed using SPM software, and then outliers detection, linear fitting and low frequency filtering were also done. The regions of interest in abnormal area and normal area were chosen respectively, and we analyzed the signal of ROIs in the following three aspects:1. The time domain signal were transformed to frequency domain using Fourier transformation to get the power spectrum, the amplitude of low frequency fluctuation (ALFF) is defined as the mean square value of the power spectrum in the low frequency (0.01-0.08Hz) stage. Then the ALFF for each ROIs were calculated.2. The correlation analysis was done to get the correlation intensity of abnormal and normal ROIs’mean signals, that is the functional connectivity intensity.3. The mean signal in normal ROI was shifted-6TR-6TR, then the obtained 13 time series were done correlation analysis with the mean signal in normal ROI, the time shift value corresponding to the maximum correlation intensity is the abnormal area’s time delay value related to the normal area. In the same way, we can gain the time delay values of abnormal and normal areas with respect to the whole brain mean value respectively.After all the process above, no consistent differences were founded between normal and abnormal regions for the three rs-fMRI indexes of the sub-acute ischemic stroke patients and the result was different with previous studies. In addition, there were also no significant differences between lesion and health areas of the three indicatiors for the 14 patients after paired t test.We thought this could possibly due to the complexity of fMRI signal in stroke area, the stroke lesion location, severty of illness, lesion size and the time from onset to treatment will all affect our ultimate result. Therefore, further investigation of new analysis approaches is still needed when applying rs-fMRI to stroke research.