All neuroimaging techniques are sensitive to artifacts associated with data collection that can appear as either false positives (signal change unrelated to brain activity) or false negatives (failure to detect real changes in brain activity). With PET, the most significant artifacts are associated with subject movement and variability in the timing or amount of radiotracer delivery. Many of these can be measured and adjusted for, but they may nevertheless impact the quality of PET results.
Functional MRI is more prone than PET to artifacts, with even small movements producing large changes in signal within individual voxels. Signal changes caused by periodic motion from breathing or cardiac pulsation can similarly hinder detection of changes in brain activation (Hu, Le, Parrish, & Erhard, 1995). Indeed, in many cases artifactual changes in MRI signals are substantially larger than the changes in BOLD signal associated with neural activation. Moreover, certain areas of the brain are extremely difficult to measure with fMRI because of signal dropout caused by boundaries between brain tissue and air (Farzaneh, Riederer, & Pelc, 1990). Because of this signal dropout, PET can detect changes in certain brain regions where many fMRI studies will produce false-negative results (particularly in ventromedial frontal and anteriormedial temporal regions). Many techniques exist to address these problems, but the quality of the data must be considered on a case-specific basis, especially when considering negative findings.
Was this article helpful?