Reproducibility of single-subject fMRI language mapping with AMPLE normalization.

PURPOSE: To evaluate the reproducibility of presurgical functional MRI (fMRI) language mapping based on test-retest scans, comparing traditional activation t-maps to relative activation maps normalized by activation mapping as percentage of local excitation (AMPLE).
MATERIALS AND METHODS: Language fMRI scans were performed by 12 healthy volunteer subjects undergoing a standard clinical presurgical mapping protocol in multiple independent scan sessions. Objective relative AMPLE activation maps were generated automatically by normalizing statistical t-value maps to the local peak activation amplitude within each functional brain region. The spatial distribution of activation was quantified and compared across mapping algorithms, subjects, scanners, and pulse sequences.
RESULTS: The spatial distribution of traditional blood oxygen level-dependent (BOLD) t-value statistical activation maps was highly variable in test-retest scans of single subjects, whereas AMPLE normalized maps were highly reproducible in terms of the location, hemispheric laterality, and spatial extent of relative activation. AMPLE map reproducibility was good regardless of scanner, field strength, or pulse sequence used, but reproducibility was best for scans acquired on the same scanner using the same pulse sequence.
CONCLUSION: Reproducibility of the spatial pattern of BOLD activation makes relative amplitude fMRI mapping a useful normalization tool for clinical imaging of language function, where reproducibility and quantitative measurements are critical concerns.