Quantitative comparison of functional magnetic resonance imaging with positron emission tomography using a force-related paradigm

The intention of our study was to compare functional magnetic resonance imaging (fMRI) with positron emission tomography (PET). We used the same force-related motor paradigm for both techniques, which allows for quantification of stimulus intensity. Regional cerebral blood flow (rCBF) was determined with PET in six male subjects (age 30 +/- 3) using the slow bolus injection technique and oxygen-15-labeled water. Scans were collected during six different conditions: at rest and during repetitive Morse key press at 1 Hz, with the right index finger at a range of different forces. In a second series of experiments fMRI data were acquired under similar conditions in six volunteers in a single slice parallel to and 51 +/- 3 mm dorsal to the anterior and posterior commissure (AC-PC). A conventional 1.5-T clinical magnetic resonance (MR) system and the FLASH technique were used. The data obtained in both series of experiments were subjected to the same statistical analyses. Statistical parametric maps (SPM) were generated by two different approaches: a correlation between peak force and rCBF or fMRI signal and using a categorical comparison of force exerted with rest, SPMs were coregistered with anatomical MR images. PET and fMRI measurements demonstrated activation in the primary motor cortex (M1) and posterior supplementary motor cortex in all subjects. Correlation analysis demonstrated foci in the M1 in four subjects with PET and in only one subject with fMRI. Locations of activation peaks differed by 2 to 8 mm between imaging methods. The relationship between fMRI signal or rCBF and peak force was logarithmic. The maximum increase in fMRI signal was 5.0% +/- 0.9 at 60% of the maximum voluntary contraction while the corresponding increase in rCBF was 13.7% +/- 1.2. The ratio of percentage rCBF change to percentage fMRI signal change was very similar across all force levels. The high degree of correspondence between PET and fMRI data provides good cross-validation for the two techniques. (C) 1996 Academic Press, Inc.