MR Perfusion Imaging

During the last 10 years, a variety of MR techniques have been developed that can provide images of cerebral perfusion (1). These approaches include those that require the injection of paramagnetic contrast agents (bolus-tracking approaches) as well as those that magnetically tag water in arterial blood as it moves into the brain. The effects of “tagged” arterial water on brain MR images can be used to calculate quantitative CBF images that can be expressed in classical physiologic units (ie, cc/l00 g/min). The major drawback to these tagging techniques is that, with the current technology, they are rather insensitive and require relatively long imaging times (≈10 minutes). Given this restriction, it is unlikely that MR arterial spin-tagging approaches will be applied to the clinical evaluation of acute stroke in the near future. Nevertheless, they could play an important role in the clinical evaluation of cerebrovascular diseases that provide a longer diagnostic “window,” especially for those that require absolute quantitation.

Following the quantitative CBF response to cerebrovascular challenges is one scenario in which MR spin-tagging flow approaches could be very useful. Samuels et al (2) employed the MR spin-tagging response to acetazolamide challenge to study middle cerebral artery stenosis, and used the results to characterize specific patterns of impaired perfusion. In this issue of the AJNR, Kastrup et al (page 1233) suggest the use of MR arterial spin-tagging approaches with another variant of the cerebrovascular challenge—breath-holding. Kastrup and colleagues demonstrate that breath-holding can provide reproducible changes in CBF in control subjects that can be followed accurately, both regionally and globally, using MR spin-tagging techniques. The advantage of the breath-hold approach is that it obviates the need for acetazolamide injection or CO₂ inhalation; the disadvantage is that it cannot be used for patients with impaired respiratory function. Both Samuels et al and Kastrup et al underscore the importance of obtaining ancillary data (eg, T₁ relaxation time images) to enable MR spin-tagging data to be interpreted in terms of absolute CBF values. This ability to quantify CBF absolutely is potentially of great clinical importance.

Functional MR (fMR) imaging approaches using blood oxygen level dependent (BOLD) effects also have been used to follow the response to cerebrovascular challenges. BOLD approaches are more sensitive than MR spin-tagging approaches. Kastrup et al emphasize, however, that BOLD results are harder to interpret because fMR imaging responds to changes in various physiologic parameters (eg, CBF, cerebral blood volume, and cerebral oxygen consumption), wheras MR spin-tagging responds primarily to changes in CBF. Nevertheless, MR arterial spin-tagging approaches also present problems in quantitation of CBF. For example, calculated CBF values will be artifactually low when arterial transit times are abnormally long, which might occur in compromised brain regions that have extensive collateral circulation. This issue could be examined using MR bolus-tracking approaches (1), which can give information on arterial transit times in compromised brain areas. Further validation of the quantitative ability of the arterial spin-tagging technique is needed before the results can be applied to individual patients.

The results of Kastrup et al and Samuels et al demonstrate the usefulness of MR arterial spin-tagging approaches for studies of cerebrovascular reserve. These approaches have a number of advantages over other techniques (eg, PET, SPECT, CT, etc); they are noninvasive, easily repeatable, and have relatively good spatial resolution. In the near future, a number of technical advances, such as phased-array head coils and higher magnetic field strengths, undoubtedly will increase the sensitivity of MR arterial spin-tagging approaches, and could make them viable for routine clinical studies of cerebrovascular disease.

An interesting sideline to these studies of physiologic perturbations of CBF is the subtlety and possible pervasiveness of the effects. Although a gross respiratory change such as a 30-second breath-hold would be unlikely to occur during a conventional fMR imaging experiment, more subtle respiratory changes could accompany some activation paradigms, particularly those with “surprise” components. This could result in MR signal changes secondary to the unanticipated respiratory (or cardiac) responses. Statistical analysis might classify erroneously respiratory responses as noise, false localization of a cognitive task, or a true but secondary phenomenon. The latter possibility reemphasizes the complexity, as well as the richness, of these new functional imaging techniques.