Research - Neurofunctional Imaging
ATTENTION - EVOKED ACTIVATION IN THE VISUAL CORTEX
Yue Cao, Ph.D. Jie Huang, Ph.D.
Department of Radiology, Michigan State University,
East Lansing, MI USA
INTRODUCTION
A recent study demonstrated that visual attention, without presentation of a stimulus, evoked a response in the early visual cortex (V1, V2, and V3), and the attention-related activity strongly depended upon the difficulty of tasks [1]. In this study, we investigate contributions of visual attention and pattern stimulation to measured BOLD responses in the visual cortex during ultra-short visual stimulation.
METHODS AND MATERIALS
Subjects: Six normal subjects (3 female, 3 male,
ages 21 to 26 years) participated in the study.
Protocol: All participants had two event-related
(ER) fMRI scans. Each scan consisted of one dummy trial and 20 15-second long
experimental trials. Of the 20 trials, 10 started with a 200ms cue signal
(a 0.3 degree white square), immediately followed by brief pattern stimulation
(17ms or 50ms), and finished with a fixation point (a white cross). For the
other 10 trials, the stimulus was absent after the cue signal. The order of
the trials was randomly chosen.
 |
Top: trial with a stimulus
Bottom: trial without a stimulus
Stimulus: The stimulus, spanning a 13.4 degree visual
angle, consisted of a pattern of black/white stripes alternating 3 cycles/degree,
and a fixation point at the center of the pattern. The subject was instructed
to keep eyes on the fixation point and to attend to stimulation when the cue
signal appeared.
FMRI Parameters: During ER-fMRI, five axial-oblique sections
oriented parallel to the Calcarine Fissure (CF) with 6mm thickness per section
were acquired using a gradient-echo EPI pulse sequence on a 1.5 T NV/i scanner
(GEMS). Other MRI parameters were TR/TE=500ms/50ms, FOV=24 cm, matrix size=64x64,
and flip angle=75 degrees.
Pre-processing: Functional images were corrected for any
possible in-plane translation and rotation. Signal intensity time course was
corrected for possible slow baseline drifts by fitting 0, 1st, and 2nd order
polynomials to each voxel's time series, and normalized to allow signal averaging
over voxels and across subjects. Functional images in each series (scan) were
sorted, according to stimulus presentation (with or without).
Activation in the Visual Cortex: The signal intensity
time course obtained during 50ms stimulation was analyzed by temporal cross-correlation
(cc) with a gamma reference function to yield an activation map (estimated
p < 0.003) [2-4]. The activated voxels in the cortex adjacent to the CF
defined a region-of-interest (ROI) for analysis of the BOLD response to 0ms,
17ms, and 50ms visual stimulation (Figure
1). Different signal intensities between 50ms and 0ms, and between 17ms
and 0ms were calculated to yield the hemodynamic responses to the stimulus
only.
RESULTS AND DISCUSSION
Activated cortical regions delineated by the cc analysis were shown in Figure 1.
The ROIs defined by the activated voxels in the visual cortex in panels (a) and (c) of Figure 1 were applied to functional images obtained without stimulus present and with 50ms and 17ms stimulation to yield corresponding BOLD signal intensity changes (left panel, Figure 2). Without stimulus present, MRI signal intensities increased in the visual ROI, and peaked at 6 seconds which was approximately 1.5 seconds later than the peaks of the responses to stimulation (Figure 2). Then, the non-stimulus elevated signal changes were removed from the BOLD signals measured during stimulation to yield stimulus-evoked visual activity (right panel, Figure 2).
Areas under the stimulus-evoked BOLD response curves were 5.7 and 1.9 (% x second) for 50ms and 17ms duration of stimulation. Without stimulus present, the elevated BOLD signals produced 1.0 (% x second) of the area. When the non-stimulus-related activity was removed from the measured BOLD signals during stimulation, the stimulus-evoked BOLD signals (areas under the response curves) are linearly related to stimulus-duration within a 2% error (Figure 3). Linear fitting of stimulus duration vs. activation with a zero intercept yields a slope of 0.112 (% x second)/ms (solid line in Figure 3). If the non-stimulus-related activity was not accounted for and subsequently removed from the measured visual BOLD signals during stimulation, the observed BOLD signals (areas) could not be related to duration of stimulation linearly. Linear fitting of the observed BOLD response (area) during 17ms stimulation via the zero intercept yields a slope of 0.171 (% x second)/ms (dotted line in Figure 3). This fitting would predict 8.53 (% x second) of the BOLD response (area) to 50ms duration of stimulation (open triangle in Figure 3), which overestimated the measured value (closed triangle) by 27%.
During visual stimulation, an increase in attention can markedly elevate neural activity in the early visual cortex, and therefore offset the measured BOLD signals in response to the stimulus. The elevated MRI signal intensities without stimulus present is possibly due to attention-related activity which was a delayed, re-entrant feedback from higher visual areas or a sustained activity during attention [5]. Activation in the anterior cingulate, dorsal-lateral prefrontal, and parietal cortex that are part of attention networks supports that attention was increased with and without stimulus present. It is essential to correct the BOLD signals measured during ultra-short visual stimulation to delineate the stimulus-evoked activity. After removal of the non-stimulus-related activation, the stimulus-evoked activity is accurately related to the stimulus duration in a linear fashion.
REFERENCES
- D Ress, et al., Nature Neurosci., 3:940-945, 2000.
- PA Bandettini, et al., Magn. Reson. Med. 30:161-165, 1993.
- GM Boynton, et al., J Neurosci 16:4207-4221, 1996.
- AM Dale, et al., Human Brain Mapping 5:329-340, 1997.
- Martinez A, et al., Nat Neurosci, 2:364-9, 1999.
