Statistica Sinica 18(2008), 1395-1419

THE ACQUISITION AND STATISTICAL ANALYSIS

OF RAPID 3D FMRI DATA

Martin A. Lindquist$^1$, Cun-Hui Zhang$^2$, Gary Glover$^3$ and Lawrence Shepp$^2$

$^1$Columbia University, $^2$Rutgers University and $^3$Stanford University

Abstract: In this work, we introduce a new approach toward the acquisition and statistical analysis of fMRI data. Our acquisition strategy is based on repeatedly measuring the low spatial frequencies present in the MR signal, allowing us to obtain a low spatial resolution snapshot of the brain with extremely high temporal resolution ($100 \; ms$ compared to the standard $2,000 \; ms$). The increased resolution allows us to study changes in oxygenation in the 3D brain immediately following activation. This in turn opens the possibility of shifting the statistical analysis of brain function closer toward the actual time frame of the underlying neuronal activation driving the process than is possible in standard fMRI experiments. However, this ability necessitates the introduction of new statistical techniques for analyzing the resulting data. We introduce one such approach in this paper. The feasibility and efficiency of the combined acquisition and analysis technique is confirmed using data from a visual-motor and an auditory-motor-visual task. The results of these experiments provide a proof of concept of our combined rapid imaging and analysis technique. It also indicates that our approach may provide important information regarding the initial negative BOLD signal, which can be used to obtain accurate temporal ordering of the various regions of the brain involved in a cognition experiment. Conversely, we show that the conventional approach of studying the positive BOLD signal will, at times, give inaccurate temporal ordering of the same regions. Thus, we believe that our approach will become an important tool for studying any cognition task which involves rapid mental processing in more than one region.

Key words and phrases: Echo-volumar imaging, fMRI, negative dip, latency, rapid imaging, temporal resolution, time series analysis.