Non-linearities in the blood-oxygenation-level dependent (BOLD) response measured by functional magnetic resonance imaging (fMRI)

A central question in the analysis of functional magnetic resonance imaging (IMRI) data is whether the measured fMRI signal summates in a linear fashion over repeated inputs. Most fMRI studies collect images sensitive to blood-oxygenation-level dependent (BOLD) contrast, which measures the local amount of deoxygenated hemoglobin (dHb). When neurons are active, more oxygenated hemoglobin is supplied than is needed for their metabolic demands, resulting in a decrease in dHb and an increase in MR signal. For analysis of fMRI data, researchers must therefore create experimental hypotheses of the measurable BOLD response based upon the predicted neuronal activity. An influential early model of the fMRI BOLD response assumes that BOLD activity is a linear transformation of neuronal input, representing the filtering effects of the vascular system. Recent studies have called this interpretation into question, due to observed differences in the pattern of linearity across brain regions that serve distinct functions.