Abstract - MNTP - University of Pittsburgh

Group J
An fMRI study of visual search: A learning experience
Wenzhu Bi, M.S.
Department of Biostatistics
Yanni Liu, Ph.D.
Department of Psychology
David Roalf, B.S. Dept. Behavioral Neuroscience
Xingchen Wu, MD/PhD DRCMR, MR Dept.
University of Pittsburgh
University of Michigan
Oregon Health Science Univ.
Copenhagen Univ. Hospital
Hvidovre Denmark
Facutly Instructor: Mark Wheeler, Ph.D.
TA: Jeff Phillips, M.S.
A visual search task was used to explore fMRI experimental design and data
analysis. Previous studies suggest that visual search tasks elicit robust activation
of the attention network, including occipital and parietal brain regions. Our visual
search task included easy (feature search) and difficult (conjunction search) trials.
It was hypothesized that the conjunction search condition would result in slower
reaction time, increased errors and increased activation in the attention network as
compared to the feature search condition. We compared results from block vs.
jittered fast event related designs for six volunteers. Behavioral results show that
the conjunction trials were responded to more slowly and less accurately than the
feature trials. Comparisons of fMRI data included: block vs. event-related analysis,
spatial smoothing vs. non-smoothing, and analysis of the time course of regions of
interest. Results of the functional data analysis show that we were able to replicate
previous studies of visual search. Both the block and event related design resulted
in activation of the attention network. The block design analysis resulted in more
robust activation than the event related design. The use of a spatial smoothing
parameter increased the functional SNR and resulted in more regions reaching
statistical significance than the non-smoothed data, but it reduced the spatial
precision. Finally, analysis of the ROI time course analysis for the block design
data allowed us to look at the temporal activation for each individual subject and
create a composite group time course for each region of interest.