Background The onsets of needling sensation introduced by acupuncture stimulus can vary widely from subject to subject. This should be explicitly accounted for by the model blood oxygenation-level dependent （BOLD） t...Background The onsets of needling sensation introduced by acupuncture stimulus can vary widely from subject to subject. This should be explicitly accounted for by the model blood oxygenation-level dependent （BOLD） time course used in general linear model （GLM） analysis to obtain more consistent across-subject group results. However, in standard GLM analysis, the model BOLD time course obtained by convolving a canonical hemodynamic response function with an experimental paradigm time course is assumed identical across subjects. Although some added-on properties to the model BOLD time course, such as temporal and dispersion derivatives, may be used to account for different BOLD response onsets, they can only account for the BOLD onset deviations to the extent of less than one repetition time （TR）. Methods In this study, we explicitly manipulated the onsets of model BOLD time course by shifting it with -2, -1, or 1 TR and used these temporally shifted BOLD model to analyze the functional magnetic resonance imaging （fMRI） data obtained from three acupuncture fMRI experiments with GLM analysis. One involved acupuncture stimulus on left ST42 acupoint and the other two on left GB40 and left BL64 acupoints. Results The model BOLD time course with temporal shifts, in addition to temporal and dispersion derivatives, could result in better statistical power of the data analysis in terms of the average correlation coefficients between the used BOLD models and extracted BOLD responses from individual subject data and the T-values of the activation clusters in the grouped random effects. Conclusions The GLM analysis with ordinary BOLD model failed to catch the large variability of the onsets of the BOLD responses associated with the acupuncture needling sensation. Shifts in time with more than a TR on model BOLD time course might be required to better extract the acupuncture stimulus-induced BOLD activities from individual fMRI data.展开更多
文摘Background The onsets of needling sensation introduced by acupuncture stimulus can vary widely from subject to subject. This should be explicitly accounted for by the model blood oxygenation-level dependent （BOLD） time course used in general linear model （GLM） analysis to obtain more consistent across-subject group results. However, in standard GLM analysis, the model BOLD time course obtained by convolving a canonical hemodynamic response function with an experimental paradigm time course is assumed identical across subjects. Although some added-on properties to the model BOLD time course, such as temporal and dispersion derivatives, may be used to account for different BOLD response onsets, they can only account for the BOLD onset deviations to the extent of less than one repetition time （TR）. Methods In this study, we explicitly manipulated the onsets of model BOLD time course by shifting it with -2, -1, or 1 TR and used these temporally shifted BOLD model to analyze the functional magnetic resonance imaging （fMRI） data obtained from three acupuncture fMRI experiments with GLM analysis. One involved acupuncture stimulus on left ST42 acupoint and the other two on left GB40 and left BL64 acupoints. Results The model BOLD time course with temporal shifts, in addition to temporal and dispersion derivatives, could result in better statistical power of the data analysis in terms of the average correlation coefficients between the used BOLD models and extracted BOLD responses from individual subject data and the T-values of the activation clusters in the grouped random effects. Conclusions The GLM analysis with ordinary BOLD model failed to catch the large variability of the onsets of the BOLD responses associated with the acupuncture needling sensation. Shifts in time with more than a TR on model BOLD time course might be required to better extract the acupuncture stimulus-induced BOLD activities from individual fMRI data.
