Functional magnetic resonance (fMRI) imaging provides a powerful, non-invasive window on brain function. Blood oxygen-level depletion (BOLD) fMRI tracks changes in neural activity by measuring the decrease in deoxyhemoglobin, corresponding to gradually increased blood flow to active regions. The limited spatial resolution of this technique has prevented analysis of activity within the neural columns that provide the functional architecture of much of the cerebral cortex. Greater fMRI resolution can be obtained by measuring a highly localized, transient increase in deoxyhemoglobin corresponding to rapid oxygen uptake by active cells, but the correlation of this signal with neural activity has been in doubt. In the 14 February Science, Jeffrey K. Thompson and colleagues at the University of California, Berkeley, US, show that this signal does tightly correlate with neuronal firing activity, enabling sub-column analysis of brain function (Science 299:1070-1072, February 14, 2003).

Thompson et al. performed simultaneous recording of oxygen concentration and single-cell neural activity with dual microelectrodes in a double-barrel micropipette. Measurements of fine-scale (<60 μm) oxygen depletion during standard measurements of orientation selectivity and ocular dominance in the cat brain showed that the largest neural response — corresponding to optimal orientation of stimulation — led to maximal oxygen depletion. Similarly, higher activity from the dominant eye led to more complete oxygen depletion. Oxygen response was significantly correlated with neural firing rate from 3 to 14 seconds after stimulus, and the oxygen response could be used to predict the orientation selectivity of neurons to within 15 degrees.

These results indicate that fMRI detection of rapid, highly localized increases in deoxyhemoglobin can accurately track neural activity on a scale finer than current standard methods. This technique has the "potential to resolve functional cortical columns that are thought to be fundamental for a wide variety of brain processes," conclude the authors.