AJNR-Presurgical Mapping with fMRI and DTI: Soon the Standard of Care?



American Journal of Neuroradiology 

Presurgical Mapping with fMRI and DTI: Soon the Standard of Care? 


The technique of fMRI has been around for over 30 years, and DTI for about 15 years. The first application of fMRI was by Ogawa et al, in 1990. In a rat model, this team was able to manipulate the blood oxygen level–dependent (BOLD) signal by inducing changes in deoxyhemoglobin concentrations with insulin-induced hypoglycemia and anesthetic gases. About a year later, Kwong and Belliveau published the first images of cerebral areas that responded to visual stimulation and vision-related tasks.

DTI was first described by Basser et al, who were experimenting on a voxel-by-voxel characterization of 3D diffusion profiles, which took into account anisotropic effects (instead of eliminating them, as in standard DWI). Tractography (or fiber tracking) was developed by applying statistical models to DTI data to obtain anatomic fiber bundle information.

Although both fMRI and DTI are now currently available in most scanners, well beyond the framework of academic institutions and research protocols, these techniques are not quite considered “standard of care.” Indeed, the processes that govern the translation of new technology into clinical practice are complex. Even more complex are the processes that lead to establishing clinical practice as standard of care, particularly at a time when established patterns of care delivery are being increasingly challenged and economic difficulties affect all aspects of society, certainly including health care.

However, some challenges, especially with fMRI, go back to basic cerebrovascular physiology. The cerebrovascular response to neuronal activation, also referred to as “functional hyperemia,” was first recognized in 1890 by Roy and Sherrington, who initially proposed a metabolic hypothesis to the phenomenon, ie, mediation via release from neurons of vasoactive agents in the extracellular space. The major role of astrocytes as key intermediaries in the neurovascular response — being interposed between blood vessels and neuronal synapses via their foot processes as modeled in the “tripartite synapse model” of the neurovascular unit — has since been recognized. Although complex, astrocyte response to changes in synaptic activity is primarily mediated by glutamate receptors through changes in intracellular Ca2+ concentration.

In fMRI, contrast is based on the BOLD effect, which reflects local shifts of deoxygenated-to-oxygenated hemoglobin ratios due to local increases in blood flow in excess of oxygen utilization following brain activity. As a result, the foundation of the fMRI BOLD signal is based on local changes in cerebral blood flow that are not linearly related to the metabolic changes inducing the flow change.

Therefore, BOLD fMRI rests on 3 major approximations: 1) the technique does not directly reflect neural activity, ie, generation and propagation of action potentials, synaptic transmission, or neurotransmitter release/uptake; 2) the changes in BOLD signal originate from that portion of the vasculature experiencing the greatest change in oxygen concentration, which occurs in the venules in the immediate vicinity of the active neurons; and 3) more importantly, fMRI signal relies on intact “neurovascular coupling,” the phenomenon that links neural activity to metabolic demand and blood flow changes.

The main reason fMRI is clinically useful most of the time is that under most circumstances neurovascular coupling remains fully intact, unaltered by confounding disorders that can interfere with this relationship. However, it has long been known that neuronal activation results in local blood flow increases that exceed local oxygen consumption, so that the oxygen utilized may constitute a small fraction of the amount delivered. Under normal conditions, the oxygen concentration in draining venules increases during neuronal activation. The original researchers who discovered this phenomenon named it “neurovascular uncoupling” or “neurovascular decoupling.” From a medical perspective, “uncoupling” or “decoupling” implies a pathologic condition, suggesting something abnormal about tissue that demonstrates this phenomenon. More recently, researchers have preferred the term “functional hyperemia” to describe the phenomenon. In fact, when there is interference with the mechanism producing functional hyperemia, the term "neurovascular uncoupling" has been re-applied, albeit with a completely opposite meaning from that originally used. Impairment in the flow response leads to neurovascular uncoupling and a reduced BOLD signal in response to neural activity, which can lead to false-negative errors in fMRI maps.

John Ulmer, reporting on a series of 50 patients, found that although accurate cortical activation could be demonstrated most of the time, various cerebral lesions could cause false negatives in fMRI results when compared with other methods of functional localization, suggesting contralateral or homotopic reorganization of function. He further suggested that pathologic mechanisms such as direct tumor infiltration, neovascularity, cerebrovascular inflammation, and hemodynamic effects from high-flow vascular lesions (ie, arteriovenous malformations and fistulas) could trigger “neurovascular uncoupling” in those patients. Neurovascular uncoupling, and other pitfalls of fMRI, are briefly discussed.

David Mikulis discusses “neurovascular uncoupling syndrome,” where lack of functional hyperemia during neuronal activation can have long-term consequences on the integrity of the tissue in the absence of acute ischemia.

Jay Pillai discusses the successful clinical application of a technique to improve the consistency of BOLD fMRI by using a breath-holding technique.

Aaron Field discusses the technique, clinical use, and some limitations of DTI and tractography, and describes patterns of alteration of white matter fiber tracts by neoplasms and other lesions.

Lastly, Wade Mueller shows that a neurosurgeon may obtain significant improvements in clinical outcomes and a drastic reduction in complication rates when working with a team that provides presurgical mapping of cerebral lesions by using fMRI and DTI (wisely, fully acknowledging their limitations) and when various team members clearly communicate using a common language.

Functional MRI and DTI are extremely useful techniques that have become increasingly available to neuroradiologists in recent years. As with any technique, these work best as parts of a whole. A good understanding of physiologic mechanisms is necessary to make us good “functional” specialists, and a good understanding of the limitations of any technique is necessary to make us better physicians.


  1. Ulmer JL, Hacein-Bey L, Mathews VP, et al. Lesion-induced pseudo-dominance at functional magnetic resonance imaging: implications for preoperative assessmentsNeurosurgery2004;55:569–79; discussion 580–1
  2. Mikulis DJ. Chronic neurovascular uncoupling syndromeStroke2013;44:S55–57, 10.1161/STROKEAHA.113.001081
  3. Pillai JJ, Mikulis DJ. Cerebrovascular reactivity mapping: an evolving standard for clinical functional imagingAJNR Am J Neuroradiol, published online before print April 30, 2014, 10.3174/ajnr.A3941
  4. Jellison BJ, Field AS, Medow J, et al. Diffusion tensor imaging of cerebral white matter: a pictorial review of physics, fiber tract anatomy, and tumor imaging patternsAJNR Am J Neuroradiol 2004;25:356–69
  5. Field AS, Filippi C, Kalnin A, et al. American Society of Functional Neuroradiology Guidelines for Clinical Application of Diffusion Tensor Imaging. Recommendations from the DTI Standards and Practice Subcommittee of the ASFNR Clinical Practice Committee. March 2012.


Image modified from: Jellison BJ, Field AS, Medow J, et al. Diffusion tensor imaging of cerebral white matter: a pictorial review of physics, fiber tract anatomy, and tumor imaging patterns.