A BOLD New Look at Neuronal MRI

Functional MRI (fMRI) is a powerful technique for measuring subtle magnetic resonance changes in active neurons, but it does have limitations. One commonly used fMRI method called BOLD, or blood oxygen level-dependent MRI, measures decreases in deoxyhemoglobin content in the vicinity of active neurons. However, BOLD has limited resolution because deoxyhemoglobin in surrounding blood vessels can interfere with the measurement. Now, in today's Science, researchers in Ralph Freeman's lab at the University of California, Berkeley, demonstrate that the resolution of this technique may be dramatically improved by focusing on transient increases in deoxyhemoglobin, which occur immediately upon neural activation.

These increases, causing an initial dip in the BOLD signal, are followed by a delayed decrease in deoxyhemoglobin, which results as oxygenated hemoglobin rushes in from surrounding blood vessels to replenish the active area. This delayed decrease in deoxyhemoglobin elicits the stronger BOLD resonance that is typically used to measure neural activity. But what does the "initial dip" tell us?

Some researchers have proposed that the dip directly results from oxygen consumption by neurons (see Malonek and Grinvaid 1996), but this contention has remained controversial, and alternative hypotheses such as changes in blood volume also explain the dip. Now, first author Jeffrey Thompson and colleagues addressed this issue directly, by simultaneously measuring oxygen consumption and neural activity.

At the heart of their experiment was a combined oxygen electrode/platinum electrode sensor, which measures the electrical activity of a single neuron and the oxygen concentration in its immediate vicinity (30 μm radius). Thompson and colleagues trained this sensor on a single neuron in the visual cortex. Neurons in the visual cortex are exquisitely sensitive to optical stimuli from particular directions or specific eyes, and the authors used this selectivity to show that oxygen consumption correlated beautifully with electrical spikes from this one neuron. Stimuli that failed to elicit neural spikes also failed to register O2 changes, while the number of electrical spikes correlated with the magnitude of O2 loss. In addition, Thompson et al. iced the cake by recording delayed increases in O2 that explain the delayed BOLD signal.

The data confirms that the dip is indeed due to oxygen consumption, and would, in principle, seem to open the door for the BOLD dip to be used as a legitimate-and spatially more accurate-measure of neuronal activity. In practice, however, "there are still important gaps in our understanding of neuronal metabolism, with important consequences for interpreting BOLD fMRI signals," cautions John Mayhew, University of Sheffield, England, in an accompanying commentary. For example, he points out that there is considerable heterogeneity in the density of capillaries and the number of mitochondria in the brain. Other problems may also need to be ironed out, the authors write, as the initial dip is not always observed.—Tom Fagan

Comments

  1. Although little more that a decade old, fMRI already has an established pantheon of "giants in the field." Seiji Ogawa, one of these giants, was the first to observe that when rodents breathe, 100 percent oxygen-diminished contrast is observed in brain images acquired with T2*-weighted signal. Based on prior work, most notably by Thulborn, Ogawa was correct in assuming that the change in signal contrast was due to a decrease in deoxyhemoglobin levels coursing through the vasculature. It is this observation that sparked the wildfire spread of fMRI throughout the field of cognitive neuroscience. The elegant paper by Thompson et al. resolves both a stylistic imprecision and a conceptual error made by these early pioneers.

    Although Ogawa coined the term BOLD—Blood Oxygen Level Dependent—to account for the observed increase in blood vessel signal, MRI is essentially blind to blood oxygen levels, per se; deoxyhemoglobin, the true source of signal change, did not, apparently, lend itself to a snappy eponym. Since those early experiments 13 years ago, essentially all studies have focused on the relationship between neuronal metabolism and deoxyhemoglobin (or oxyhemoglobin), using various flavors of MRI, spectroscopy and optic imaging. Coupling an oxygen-sensitive microelectrode with signal unit recording in the cat visual cortex, the current study is noteworthy for establishing a relationship between a true blood oxygen level-dependent signal and neuronal metabolism.

    Although Ogawa was correct in attributing the BOLD response to a shift in deoxyhemoglobin, he initially erred in predicting how this response will map onto changes in metabolism. Justifiably, he predicted that an increase in neuronal metabolism will lead to an increased consumption of oxygen, an increase in deoxyhemoglobin, and, therefore, a decrease in T2*-weighted signal intensity (Ogawa et al., 1990). It was only after he and his colleagues systematically shifted the chronic metabolic state of neurons, using various drug infusions, that they reformulated their prediction: that an increase in metabolism will result in a decrease in deoxyhemoglobin, perhaps through an overflow of fresh blood, leading to an increase in signal intensity, while a decrease in metabolism will result in a reverse response (Ogawa et al., 1990). By and large, this has turned out to be true when one measures the BOLD signal a few seconds after a shift in metabolism has been initiated. Why the hemodynamic response is designed this way, and what are the biochemical mediators of this delayed response are still under investigation. In any case, since brain cells are thought to be aerobic energy-producers, it was fair to assume that at least for an instant in time the increased metabolic demands of neurons (and maybe glia) would be associated with a decrease in local oxygen levels. This assumption has been indirectly corroborated by MRI studies describing the early dip in T2*-weighted signal, and by optic imaging. Finally, thanks to the current study, a direct proof is this assumption is established.

    Does this definitive finding have direct implications for fMRI’s clinical use? Any advance that increases the precision of the technology is welcomed; however, the precision we gain from the current study does not fill the void that currently limits fMRI’s clinical utility. The "early dip" in oxygen and T2*-weighted signal is spatially more precise, possessing higher anatomical fidelity than the more conventional "delayed rise" in signal. This is so because the early dip is capillary-based, while the delayed rise is predominantly venule-based, and the early dip is, therefore, more focal and closer to cell bodies. Although this gain in spatial precision might aid in detecting brain dysfunction in relatively small brain areas like the hippocampal subregions or the nuclei of the basal ganglia, the gain will likely be most helpful in addressing basic questions regarding submillimeter circuits—such as the ocular dominance columns. The trade-off is that the amplitude of the early dip is much smaller than the amplitude of the delayed rise in signal, and, of course, signal-to-noise is critical for diagnostic purposes.

    This leads into what many consider the more burning fMRI imprecision that hampers its clinical utility. As typically practiced, a patient is asked to perform an activation task and the acute change in the BOLD response above some reference baseline is used as the metric of brain function. The problem is that we now know that the baseline state—sometimes called the resting state, but more appropriately termed the chronic state—is different across brain areas, and, more disturbingly, different between patients and controls. So, for example, we know from PET studies that the chronic metabolic state of the occipital cortex is higher than that of the frontal cortex (Raichle et al., 2001), and we know from MRI studies that the occipital cortex has a higher level of deoxyhemoglobin in the chronic state (Davis et al., 1998), and has shorter T2* relaxation times (Wansapura et al., 1999). There is now a rising chorus of studies showing that the chronic state will influence the amplitude of the acute BOLD response, and so it is difficult to make comparisons between groups regarding differences in the amplitude of the acute response. Thus, although many papers, including work from my own lab, have shown that AD patients have a diminished acute response in BOLD signal, we should be cautious in our interpretation of these findings. Certainly, these unresolved sources of noise diminish the potential of using fMRI for diagnostic purposes. Instead of increasing spatial fidelity of the MRI signal—the main gain of the early dip—what we really need are ways to increase signal validity, particularly in the context of a diseased brain. Issues of signal validity are currently being explored, either by attempts at calibrating the acute BOLD response or by attempting to understand and exploit the chronic metabolic state in order to "localize the lesion.

    References:

    Ogawa S, Lee TM, Nayak AS, Glynn P. Oxygenation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn Reson Med. 1990 Apr;14(1):68-78. PubMed.

    Ogawa S, Lee TM, Kay AR, Tank DW. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci U S A. 1990 Dec;87(24):9868-72. PubMed.

    Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci U S A. 2001 Jan 16;98(2):676-82. PubMed.

    Davis TL, Kwong KK, Weisskoff RM, Rosen BR. Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci U S A. 1998 Feb 17;95(4):1834-9. PubMed.

    Wansapura JP, Holland SK, Dunn RS, Ball WS. NMR relaxation times in the human brain at 3.0 tesla. J Magn Reson Imaging. 1999 Apr;9(4):531-8. PubMed.

    View all comments by Scott Small

Make a Comment

To make a comment you must login or register.

References

Paper Citations

  1. Malonek D, Grinvald A. Interactions between electrical activity and cortical microcirculation revealed by imaging spectroscopy: implications for functional brain mapping. Science. 1996 Apr 26;272(5261):551-4. PubMed.

Further Reading

Papers

  1. Arai H, Ishiguro K, Ohno H, Moriyama M, Itoh N, Okamura N, Matsui T, Morikawa Y, Horikawa E, Kohno H, Sasaki H, Imahori K. CSF phosphorylated tau protein and mild cognitive impairment: a prospective study. Exp Neurol. 2000 Nov;166(1):201-3. PubMed.
  2. Ishiguro K, Ohno H, Arai H, Yamaguchi H, Urakami K, Park JM, Sato K, Kohno H, Imahori K. Phosphorylated tau in human cerebrospinal fluid is a diagnostic marker for Alzheimer's disease. Neurosci Lett. 1999 Jul 30;270(2):91-4. PubMed.

Primary Papers

  1. Thompson JK, Peterson MR, Freeman RD. Single-neuron activity and tissue oxygenation in the cerebral cortex. Science. 2003 Feb 14;299(5609):1070-2. PubMed.
  2. Mayhew JE. Neuroscience. A measured look at neuronal oxygen consumption. Science. 2003 Feb 14;299(5609):1023-4. PubMed.

Annotate

To make an annotation you must Login or Register.