Application Of Independent Component Analysis In Visual FMRI On Sedated Children

PrefaceIndependent component analysis (ICA) is a data - driven method. It attempts to disaggregate multidimensional data to several independent components. In fMRI data analysis, it can separate independent components without any assumptions about the paradigm and hemodynamic response function. Statistical parametric mapping ( SPM) is method based on General linear model ( GLM) , it is a powerful tool for analysis of functional magnetic resonance data. SPM need the assumption of hemodynamic response function and paradigm series. However the hemodynamic response function between sedated children and normal healthy adult is different in visual paradigm. Analysis with the fixed dynamics response function (HRF) curve (based on normal adult's dynamics response function) may cause the different result in visual paradigm between sedated children and normal adult. Such differences may be due to fixed HRF not well reflected sedated children's hemodynamic response function. Independent component analysis have not the assumption on the shape of hemodynamic response function and therefore the data analysis for this situation using independent component analysis may have some advantages.PurposeIn the research we use independent component analysis and general linear model for visual paradigm fMRI data analysis in sedated children. Through these two methods analysis results to evaluate the ability of independent components a-nalysis in visual fMRI on sedated children.Materials and MethodThe subject include 10 sedated children. The paradigm is a block design that contains 4 times task, each task is a 30 s visual stimulation with black and white reserving checkbox with 30s control state, the frequency is 8Hz. Scans performed in Philips intera 3. 0T MRI scanner. The preprocessing steps include: movement correction, time slicing, coregistration between functional image and structure image. Next we use SPM and infomaxICA algorithms implemented on the GIFT software packages to process functional images. In SPM analysis attest was perform with uncorrected P<0.001. During inference state, we define control state as a baseline and stimulation state is under the baseline. The design matrix of SPM consists of two synthetic gamma function convolved with the visual paradigm time series, The results of SPM analysis given in the form of active maps. The first step of ICA analysis estimates of the number of independent elements, and then using PCA to reduce data dimension before performing an independent component analysis. A correlation analysis was performed between ICs and reference HRF curve contained in SPM s design matrix. Then we use correlation coefficient to sort ICs.ResultIn SPM analysis, we can detect visual cortex activation in all 10 subject, because we defined the stimulation state under the control state which is defined as baselines, activation map means that the signal of areas is decreased. In ICA analysis, all of 10 patients are detected tasks related independent components (ICs) , theirs position is similar to SPMs result in some degree. The time course of task related ICs is not similar to reference HRF curve;it seems time courses have a delay and decrease of signal obviously. The correlation coefficient between reference HRF curve and task related ICs is -0.2 to -0.4.ConclusionIndependent component analysis can detect tasks related components of sedated children in visual paradigm, namely the visual cortex active map.The result of Independent component analysis in visual fMRI suggests sedated children' s shapes of HRF is not similar to conventional hemodynamics response function.