THE SPATIAL, TEMPORAL, AND INTERPRETIVE LIMITS OF FUNCTIONAL MRI

Since the inception of functional magnetic resonance imag-ing (fMRI) in 1991, an explosive growth in the number of users has been accompanied by steady widening of its range of applications. A recent search of the National Library of Medicine database for articles with fMRI or BOLD (blood oxygenation-dependent) in the title revealed more than 1,000 citations. Improvements continue in pulse sequence design, data processing, data interpretation, and the tailor-ing of cognitive paradigms to the unique advantages and limits of the technique. This chapter describes the receding limits of fMRI. Specifically, the limits of spatial resolution, temporal resolution, interpretability, and implementation are discussed. The goal is to give the reader a perspective of the evolution of fMRI in the past 9 years and a sense of excitement regarding its ultimate potential. A user of fMRI primarily is interested in extracting at least one of three types of neuronal information: where neu-ronal activity is happening, when it is happening, and the degree to which it is happening. To extract this information optimally, an understanding of the basics of some of the more esoteric details is necessary, which are presented in this chapter. First, the basics of fMRI contrast are discussed. Second, the key of fMRI interpretation, the neuronal–hemo-dynamic transfer function, is described. Third, an overview of methods by which neuronal activation is played out in fMRI subjects and subsequently measured is provided. In this section, the popular technique of event-related fMRI (ER-fMRI) is described in detail, along with emerging methods of neuronal information extraction. Fourth, the issues of temporal and spatial resolution are discussed. Fifth, the limits of interpretation and the potential for further neuronal–hemodynamic information extraction are dis-cussed. Lastly, some implementation limits are finally given as a practical guideline