A GUIDE TO THE FUNCTIONAL IMAGING OF COGNITIVE-PROCESSES

We review methodologic issues involved in the study of changes in cere bral blood now and metabolism associated with the performance of speci fic tasks using positron emission tomography (PET), single photon emis sion computed tomography (SPECT), and functional magnetic resonance im aging (FMRI). Our particular focus is changes in association cortices associated with complex behavior. The basic principles underlying thes e imaging modalities-image registration, region of interest placement, resolution, image normalization, signal-to-noise ratio, statistical i ssues, task design, and relationship to the ablative paradigm-are cons idered. To date, PET and SPECT have been significantly limited by prob lems of statistical noise, image registration, region-of-interest plac ement, signal-to-noise ratio, and task design. Great methodologic adva nces have been made, primarily by PET investigators, but statistical n oise, the localization of metabolic or blood-flow changes to specific gyri, and the teasing apart of complex tasks to emphasize a single com ponent continue to be major challenges. PET will continue to be limite d by its expense and lack of availability. SPECT will continue to be l imited by the very small number of studies that can be done in any one subject. FMRI is free of many of the intrinsic limitations of PET and SPECT, but it has been sharply constrained by the problem of signal-t o-noise ratio. This limitation is rapidly yielding to ingenious techni cal solutions, and in the long run, the major drawbacks of FMRI will b e the intrusiveness of the imaging environment, susceptibility to move ment effects, and the same problems in task design that face PET and S PECT investigators. There is reason to believe that these various appr oaches to functional imaging will be complementary for the foreseeable future. The value of functional imaging currently rests far more on i ts promise than its accomplishments, but its promise is great.