Arterial spin labeling (ASL) is a magnetic resonance (MR) imaging technique used to assess cerebral blood flow noninvasively by magnetically labeling inflowing blood. In this article, the main labeling techniques, notably pulsed and pseudocontinuous ASL, as well as emerging clinical applications will be reviewed. In dementia, the pattern of hypoperfusion on ASL images closely matches the established patterns of hypometabolism on fluorine 18 fluorodeoxyglucose (FDG) positron emission tomography (PET) images due to the close coupling of perfusion and metabolism in the brain. This suggests that ASL might be considered as an alternative for FDG, reserving PET to be used for the molecular disease-specific amyloid and tau tracers. In stroke, ASL can be used to assess perfusion alterations both in the acute and the chronic phase. In arteriovenous malformations and dural arteriovenous fistulas, ASL is very sensitive to detect even small degrees of shunting. In epilepsy, ASL can be used to assess the epileptogenic focus, both in peri- and interictal period. In neoplasms, ASL is of particular interest in cases in which gadolinium-based perfusion is contraindicated (eg, allergy, renal impairment) and holds promise in differentiating tumor progression from benign causes of enhancement. Finally, various neurologic and psychiatric diseases including mild traumatic brain injury or posttraumatic stress disorder display alterations on ASL images in the absence of visualized structural changes. In the final part, current limitations and future developments of ASL techniques to improve clinical applicability, such as multiple inversion time ASL sequences to assess alterations of transit time, reproducibility and quantification of cerebral blood flow, and to measure cerebrovascular reserve, will be reviewed.
After reading the article and taking the test, the reader will be able to:
■ Describe the technical principles of arterial spin labeling
■ Discuss the limitations and pitfalls of ASL
■ Discuss the role of ASL in neurodegenerative and cerebrovascular disease, arteriovenous malformation, epilepsy, neoplasms, and neurologic disorders
■ Discuss the use of ASL to measure cerebrovascular reserve using a vascular challenge
Conclusion
Although ASL has been around for more than 2 decades, it only recently began to make the transition from a research tool to clinical use due to increasing awareness of radiologists and clinicians of its capability and technical improvements that have made this approach more reliable and available as product sequences on MR imaging platforms. For many diseases, including dementia, vascular diseases, neoplasms, and various psychiatric diseases, ASL provides additional and complementary information to that available from structural MR imaging. Current methodological developments aim to increase the robustness and decrease interimager variability of CBF estimation.
Essentials
■ Due to the close link between brain metabolism and perfusion, patterns of fluorine 18 fluorodeoxyglucose PET closely resemble those of arterial spin labeling (ASL).
■ ASL complements structural information of standard MR imaging in neurocognitive decline and may allow early diagnosis of dementia.
■ ASL may depict mismatch perfusion in acute stroke and help identify tissue at risk in chronic cerebrovascular disease.
■ ASL may depict arteriovenous shunting in arteriovenous malformation and fistulas.
■ ASL may be used to localize the epileptogenic focus in seizure disorders.
■ In neoplasms, ASL may replace standard gadolinium-enhanced dynamic susceptibility contrast-enhanced imaging, notably in patients with allergies or renal insufficiency or in children.
■ ASL may help detect subtle functional changes in psychiatric disorders such as posttraumatic stress disorder or mild traumatic brain injury, where standard structural MR imaging typically provides no detectable anomaly.
■ Emerging new applications of ASL include superselective ASL to map vascular territories and cerebrovascular reserve imaging.
ASLarterial spin labeling labeling schemes. A, In PASLpulsed ASL, an inversion slab is placed proximal to the imaging volume to label blood in the arterial feeding vessels supplying the brain. The pulse is short (∼10 msec) and all the blood is inverted simultaneously. B, In PCASLpseudocontinuous ASL, the inflowing arterial blood is continuously inverted as it flows through the labeling plane by means of a process known as flow-induced adiabatic inversion. The PCASLpseudocontinuous ASL labeling pulse train is typically applied for a period of approximately 1–2 seconds.
Figure 2:
Differences in labeling degree of ASLarterial spin labeling bolus for PASLpulsed ASL (left column) and PCASLpseudocontinuous ASL (right column). The top row shows the temporal profile of the bolus (1 = fully inverted; 0 = fully relaxed). Since the PASLpulsed ASL inversion slab is inverted at a single point in time (t = 0 on this graph), all the inflowing arterial blood undergoes the same amount of T1 recovery at all time points after this. In PCASLpseudocontinuous ASL, blood is labeled as it flows through the inversion plane and recovers en route to the imaging volume. A–D show the degree of labeling remaining at several time points after the start of labeling (t = 0): A, t = 0; B, t = arterial arrival time (ATT); C, t = bolus duration (τ); D, t = ATT + τ. Color scale represents the range from fully inverted (red) to fully relaxed (blue). It can be seen that the PCASLpseudocontinuous ASL labeling process produces a bolus with a higher overall degree of inversion than does the PASLpulsed ASL, resulting in a higher intrinsic SNRsignal-to-noise ratio for PCASLpseudocontinuous ASL.
Figure 3:
Example of an underestimation of relative CBFcerebral blood flow (relCBF) in ASLarterial spin labeling due to a proximal vessel stenosis. A, The estimated relative CBFcerebral blood flow based on DSCdynamic susceptibility contrast perfusion is within normal limits. B, In contrast, the relative CBFcerebral blood flow estimated by using a standard single–inflow-time ASLarterial spin labeling sequence demonstrates marked reduction in the left anterior and middle cerebral artery territories. The origin of this discrepancy is the increased perfusion delay, illustrated on, C, DSCdynamic susceptibility contrast-derived delay map, which exactly matches the altered perfusion in ASLarterial spin labeling due to the presence of a high-grade stenosis of the left internal carotid artery on, D, time-of-flight (TOF) image. The underlying principle is explained in E. The normal perfusion time series of DSCdynamic susceptibility contrast imaging (solid black line) is shifted to the right due to the presence of a proximal vessel stenosis and slower collateral flow (dotted line). DSCdynamic susceptibility contrast imaging acquires an entire time series, this shift in the bolus arrival simply causes a shift of the estimated curve, and the relative CBFcerebral blood flow can be accurately estimated in DSCdynamic susceptibility contrast. In contrast, a single–inflow-time ASLarterial spin labeling sequence with a standard inflow time (indicated by the vertical line) will underestimate the true perfusion simply because it is too early with respect to the peak of the perfusion curve.
Figure 4:
Examples of dementia. Transverse FDGfluorine 18 fluorodeoxyglucose and ASLarterial spin labeling images of, A, a healthy individual (male; age, 57 years; Mini-Mental State Examination [MMSE] score, 30), B, patient with ADAlzheimer disease (male; age, 52 years; MMSE score, 19), and, C, patient with frontotemporal lobar dementia (female; age, 53 years; MMSE score, 26). Functional images show predominant prefrontal abnormalities in FTD and parietal abnormalities in ADAlzheimer disease. Red color reflects normal metabolism and perfusion.
Figure 5:
Images in 50-year-old woman presenting with stroke, with 14 hours of right hemiparesis and aphasia, National Institutes of Health Stroke Scale score of 9 at the time of imaging. Patient was later found to have extracranial left internal carotid artery dissection. A, Diffusion-weighted images demonstrate irreversibly damaged tissue within the left caudate and putamen. By using a multidelay ASLarterial spin labeling sequence capable of acquiring both, B, CBFcerebral blood flow and, C, arterial transit time images, a larger region of perfusion abnormality is identified. D, Conventional DSCdynamic susceptibility contrast images show the region of perfusion abnormality (time to the maximum of the residue function) is concordant with the findings on ASLarterial spin labeling images.
Figure 6:
Example of multidelay ASLarterial spin labeling imaging. A fixed labeling duration of 2000 msec is used, but on subsequent images, different postlabel delays (PLD) ranging from 700 to 3000 msec are used, yielding ASLarterial spin labeling difference (control label) images as shown in A. From these data, and the use of general kinetic modeling, one can simultaneously measure, B, an arrival-time–corrected CBFcerebral blood flow and, C, the arterial arrival time itself. In this patient, there is near symmetric CBFcerebral blood flow but clear arterial arrival delay in the right hemisphere, as shown by the higher values on the arrival time map.
Figure 7:
Images in 51-year-old man with exertional headaches and an arteriovenous fistula imaged at 1.5 T. A, T2-weighted images demonstrate very subtle flow voids in the inferior frontal lobe (arrow). B, ASLarterial spin labeling images demonstrate linear high signal intensity (arrows) in the region of the right inferior frontal lobe, which extends to the right cavernous sinus, indicative of an arteriovenous shunt lesion. C, Collapsed and, D, source MR angiographic images confirm the presence of an ethmoid dural fistula (arrow), which was recognized only after the observation of the abnormal ASLarterial spin labeling signal within the venous structures draining the fistula. This case also demonstrates that while 3 T is preferable to 1.5 T, relevant clinical information can be obtained with ASLarterial spin labeling at 1.5 T.
Figure 8:
Comparison of ASLarterial spin labeling MR imaging and FDGfluorine 18 fluorodeoxyglucose PET in epilepsy. Each column represents imaging studies of a single patient. A, Preoperative axial T1-weighted MR images. B, Raw FDGfluorine 18 fluorodeoxyglucose PET images. C, Raw PASLpulsed ASL CBFcerebral blood flow maps. D, Image fusion of T1-weighted MR imaging and FDGfluorine 18 fluorodeoxyglucose PET. E, Image fusion of T1-weighted MR imaging and PASLpulsed ASL. Arrows indicate regions of focal cortical dysplasia that are epileptogenic. Numbers 1–5 refer to different patients. (Reprinted, with permission, from reference 80.)
Figure 9:
Images in 78-year-old man with unresectable anaplastic astrocytoma (arrow) (World Health Organization Grade 3), seen on, A, postcontrast T1-weighted and, B, FLAIR images. C, ASLarterial spin labeling image demonstrates increased CBFcerebral blood flow (arrow) in the region of the tumor. D, Increased cerebral blood volume (arrow) is also visible on bolus DSCdynamic susceptibility contrast image, though it is more difficult to appreciate due to the extensive number of arterial and venous vessels that surround the tumor. The findings of high CBFcerebral blood flow are characteristic of high-grade (grades III and IV) glial neoplasms.
Figure 10:
A, T2-weighted, B, postcontrast T1-weighted, and C, ASLarterial spin labeling images in a 40-year-old man with von Hippel Lindau disease and multiple hemangioblastomas after multiple prior surgeries. Arrows point to the extremely high CBFcerebral blood flow measured with ASLarterial spin labeling even in very small lesions.
Figure 11:
Images in 57-year-old man with lung cancer brain metastasis 15 months after resection, radiation, and chemotherapy. A, Axial T1-weighted gadolinium-enhanced image shows a new enhancing lesion, with, B, axial fluid-attenuated inversion recovery image demonstrating extensive vasogenic edema in the region of the resection cavity. C, ASLarterial spin labeling image demonstrates no increase in CBFcerebral blood flow in this region, similar to findings on, D, FDGfluorine 18 fluorodeoxyglucose PET image. These findings were considered to represent radiation necrosis and the patient is being followed serially rather than with re-resection.
Figure 12:
Images in 29-year-old woman with bilateral Moyamoya disease. Both, A, fluid-attenuated inversion recovery and, B, diffusion-weighted images show evidence of prior infarcts in the deep white matter on the left side. C, Multidelay ASLarterial spin labeling image acquired at baseline shows normal CBFcerebral blood flow in the bilateral anterior circulation. D, However, image obtained 10 minutes after intravenous administration of 1 g of acetazolamide demonstrates the expected CBFcerebral blood flow increase in the posterior circulation but marked reduction in CBFcerebral blood flow in both anterior circulations, compatible with cerebrovascular steal. This finding has been associated with a high risk of subsequent cerebrovascular events and prompted bilateral direct superficial temporal artery-middle cerebral artery bypass.
Figure 13:
Functional connectivity analysis of ASLarterial spin labeling raw data in 154 healthy elderly control subjects and 66 cases with mild cognitive impairment. Note that for illustrative purposes, the raw ASLarterial spin labeling data were directly analyzed by using tensorial independent component in FSL (www.fmrib.ox.ac.uk/fsl/), an established technique optimized for functional MR imaging analysis. A, Spatial representation of one of the independent components. B, The on-off labeling pattern of the raw ASLarterial spin labeling data is clearly visible on the corresponding temporal representation. Currently ongoing methodological adaptation and optimization to the specific properties of ASL data, such as implementing the known prior knowledge of the on-off time course, will improve the quality of ASL-derived functional connectivity networks.
