How do we do fMRI?

• We ask the patient to lie in the MRI scanner while we collect a series of MR scans.

• First we collect anatomical scans to visualize the particular sulci and gyri (cracks and bumps) of the patient's brain.

• We next ask the patient to perform certain motor tasks, such as touching her fingers to her thumb successively, or simply moving her wrist.

• While the patient is doing this task, we collect a series of Echo-planar images, time-locked to each of these specific tasks.

• Next we ask the patient to perform a series of memory and language tasks which we use to determine the exact location in her brain where language reception and production takes place.

• Once again we collect a series of Echo-planar images at the precise time the patient is hearing sentences or thinking about specific definitions of words.

• Since fMRI images degrade with patient motion, and since the brain is "in motion" whenever someone swallows, or when someone speaks in an animated way, we are careful to ask patients to lie still. As nearly forty percent of the blood pumped by the heart goes directly to the brain, the brain pulses and beats in synchrony with the heart. Therefore, precise timing and co-registration of images, both spatially and temporally, is important for high accuracy.

• We use a series of mathematical algorithms to cancel the effects of this motion.

• We then treat the series of images in a statistical fashion, normally performing linear or parametric based mapping statistic.

• We construct brain maps (images) based on the results of the statistical analyses. Since there is a large number of images (in excess of 1000 images per session) and since many important regions of these images are compared, the number of comparisons can reach into the hundreds of thousands. Great care and great precision are the hallmark of this approach.