Non-Linearities of the BOLD Response in the Human Brain

In this study, event-related functional magnetic resonance imaging (fMRI) studies were performed on healthy volunteers at 7 Tesla (T), the highest field strength currently available for human studies. This study assessed the linearity of the hemodynamic response to short-duration stimuli of 0.25s, 0.5s, 1.0s, and 2.0s. This study expected clearer resolution at 7T. Results showed that the hemodynamic response at short duration stimulus exhibited non-linear characteristics. Analysis of response location showed a clustered dispersion to the observed nonlinearity. Noteworthy discrepancies were found between the non-linear aspects of the hemodynamic responses in this study when compared to a previous study done at 4T. Unique to this study was the initial creation of a linearity index map, showing the spatial clustering of the non-linearity. Introduction In previous neuroimaging studies of the human brain, it has been suggested that hemodynamic response to long-duration stimuli (visual or auditory) shows a linear relationship. The purpose of this study was to determine the linearity of the hemodynamic response to short duration stimuli less than two seconds. It is important to determine whether the relationship between stimulus duration and BOLD response is different for different durations so that the physiologic significance of the response can be accurately determined. The hypotheses of this study were: 1) There would be a non-linear relationship between stimulus duration of less than two seconds in the hemodynamic response using event-related BOLD functional magnetic resonance imaging (fMRI) at high magnetic field strength of 7 Tesla (T). 2) The higher field strength would give sufficiently clear resolution to show that the nonlinearity is spatially clustered in the human brain. Background The technique of fMRI is used to image brain activity at video rates (30 milliseconds/image) while the brain is functioning. Since active brain cells use oxygen, the activity of the cells can be detected and seen as ‘activated’ against a background of less active cells. Changing levels of deoxyhemoglobin are sufficient to cause measurable changes in the blood-oxygen level in the visual cortex (1). Mapping regions of the brain that control motor functions is possible because of these changes. Most fMRI studies assume a linear relationship to model stimulus duration versus corresponding BOLD response (2), but Vazquez and Noll in 1998 suggested that at short duration stimuli the relationship might be non-linear (3). Therefore knowing the true nature of the relationship between stimulus duration and BOLD response is crucial to response modeling. Vazquez and Noll reported that the hemodynamic response to stimuli of less than four seconds at 1.5T was non-linear; therefore short duration stimuli could not be used to predict responses to long-duration stimuli (3). In a 2000 study, Pfeuffer (4) showed that non-linear relationships existed for stimulus duration (SD) less than two seconds at 4T. Using short-duration stimuli, he showed that the BOLD response exhibited strong non-linear characteristics in both amplitude and width. Yacoub et. al. (5) showed that the sensitivity and spatial specificity are improved at 7T. This suggests that ultrahigh field magnetic resonance systems are advantageous for functional mapping in humans. The study reported in this paper dealt with the non-linear aspects of the hemodynamic response done at 7T in order to improve spatial specificity. Since nonlinearity in the BOLD response has been reported using 1.5 and 4T instruments, and 7T instruments provide more sensitive and accurate readings, this study was designed to use the most sensitive techniques presently available to establish whether the BOLD response to short duration stimuli in non-linear.