Purchase this article with an account.
CW Tyler, L Kontsevich, H Baseler; Independent Spatiotemporal Components in the fMRI BOLD response . Invest. Ophthalmol. Vis. Sci. 2002;43(13):4745.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose:Typical blood-oxygenation-level-dependent (BOLD) responses have a peak time of around 5 sec, implying a slow integration of the underlying neural activity. Methods of extracting multiple temporal components of the response across voxels therefore permit a fine-grain spatial analysis of the fMRI time course. Methods:Observers were stimulated with On/Off random-element stimuli with a 45 sec period. The response time series were recorded in an fMRI experiment conducted at 3T with spiral acquisition throughout the occipital lobe. Significantly-activated voxels were identified on the basis of correlation ≷ 0.5-0.7 with the Fourier fundamental. Independent components were isolated from the response variation over voxels on the basis of deviation from the uniform Gaussian distribution. The time-series was then compared with the input signal convolved with a model BOLD impulse response of an n-pole filter, together with independent early and late negative lobes. The impulse response parameters were optimized separately for best fit to each time-series component. Results:Of the significantly activated voxels (∼1% of the total) at least two independent components could be identified. Correlations of up to 0.8 were obtained with the model BOLD signal with optimized impulse response parameters. The best-fitting impulse responses had peak times from 2-6 sec, different for different components. The impulse responses had pronounced negativities either before or after the peak. These contrasting time-course components predominated in different regions of occipital cortex. Conclusion:The data indicate that some cortical areas exhibit and early negativity before the main response peak, while other cortical areas exhibit a late negativity after the response peak. The early negativity had a peak time as short as 1 sec. Separation of these two negative components provides insight into the temporal processing reflected in the BOLD response. The identification of independent components allows the localization of rapid components in the BOLD signal comparable with the time course of neural signals.
This PDF is available to Subscribers Only