Recommendations for the use of arterial spin labeling in clinical neuroimaging

Summary

Arterial spin labeling (ASL) is a non-invasive MRI method used to image cerebral perfusion. Given increasing concerns regarding the use of gadolinium-based contrast agents and significant technical advancements in ASL implementation, the method is gaining attention in various diagnostic applications. This review article aims to familiarize readers with the fundamentals of ASL sequence implementation in neuroradiology, discuss optimal scanning parameters for achieving the highest quality and accuracy in data interpretation, and provide an overview of its diagnostic applications in the areas of cerebrovascular diseases, neuro-oncology, epilepsy, and neurodegeneration. Furthermore, we present illustrative radiological cases and explore the potential future developments of non-invasive ASL techniques.

Keywords:

Neuroimaging – magnetic resonance imaging – perfusion imaging

This is an unauthorised machine translation into English made using the DeepL Translate Pro translator. The editors do not guarantee that the content of the article corresponds fully to the original language version.

 

Introduction

The arterial spin labeling (ASL or arterial spin labeling) method is an advanced non-contrast MR imaging method that maps perfusion (blood flow) through a selected organ, most commonly the brain [1]. The method was originally developed experimentally in the early 1990s and was inspired in particular by the "steady state" perfusion imaging approach in ¹⁵O-H₂O-PET [2], which remains the gold standard. The principle of ¹⁵O-H₂O-PET is to measure regional blood flow using radiolabelled water (¹⁵O-H_(2)O) and detect the emitted radiation using PET. In the following years, ASL has been adapted for clinical use [3]. Since its inception, the method has undergone significant development, which has been supported by technical advances in MR, such as the increased availability of 3T scanners and multichannel head coils. ASL is currently available from all leading commercial MR instrument vendors [4] and is gradually gaining importance in the clinical evaluation of brain pathologies, specifically brain tumors, neurovascular and neurodegenerative diseases, and epilepsy [5]. The intention of this article is to introduce the basic principles of the ASL method, explain in detail the settings for its optimal use, and then demonstrate practical clinical applications and indications.

 

Perfusion MRI

In clinical practice, the two most commonly used MR perfusion imaging methods are Dynamic Susceptibility Contrast (DSC) and ASL. The main difference between DSC and ASL is the necessity of using an intravenously administered gadolinium-based contrast agent, which, due to its paramagnetic properties, causes an abrupt change in the T2 (*)-weighted signal [6]. Monitoring the dynamic signal change as the substance passes through the tissue allows perfusion to be measured with a relatively high signal-to-noise ratio [7] compared to perfusion-based ASL measurement. DSC has the advantage over ASL in the ability to obtain quantitative blood flow data in real time with higher spatial and temporal resolution, whereas ASL is limited by lower sensitivity and longer scan times (~5 min for ASL versus ~2 min for DSC). However, analysis of cerebral perfusion using contrast agent (DSC) has several disadvantages:

1.
Health risks: intravenous use of gadolinium-based contrast agents may be associated with a risk of allergic reactions and potential nephrotoxicity, especially in patients with renal impairment [8]. Long-term deposition of gadolinium has been described with linear carriers, especially in the globus pallidus and nucleus dentatus [9]. Therefore, since 2007, the European Medicines Agency has recommended the diagnostic use of only macrocyclic gadolinium-based contrast agents, which are considered safer due to their higher stability and lower risk of gadolinium release into the body. However, there is still a lack of published results of randomized prospective studies on the safety of macrocyclic contrast agents, which are currently ongoing and awaiting evaluation by the Food and Drug Administration (FDA) [10]. Thus, increased caution should be exercised when intravenous contrast agents are administered to children, pregnant women, and patients with renal impairment [11].

2.
Patient discomfort associated with intravenous contrast agent administration [12] also plays a role in terms of data quality, as it can cause motion artefacts, especially in paediatric patients, and thus complicate quantitative analysis [13].

3.
T2* artefacts around the tissue interface, which significantly affect the quantification of perfusion parameters of DSC in areas of necrosis, at the air-tissue and bone-tissue interface (e.g. at the skull base) [14].

4.
The higher costs are due to the price of the contrast agent as well as higher personnel costs, which are due to the requirements for the composition of the medical team defined in Decree No. 55/2011 Coll.

5.
Negative ecological impacts: anthropogenic gadolinium has been detected in the world's water sources, including groundwater, lakes and urban tap water [15].

6.
More complex implementation in practice: requires the use of contrast agent administration with a synchronized infusion pump and, above all, the coordinated cooperation and experience of MR radiology assistants.

 

In addition, the measurement without exogenous contrast agent using ASL allows the measurement of perfusion also in healthy volunteers, even repeatedly (longitudinally). Longitudinal measurement allows more precise and especially easier monitoring of the temporal evolution of pathologies and also comparison with control groups. Also for these reasons, there is a growing interest in validating and extending the ASL method into routine clinical practice.

 

ASL implementation

The basic principle of the ASL method is the endogenous labelling of hydrogen spins in the blood supplying the organ of interest (most often the brain) and the analysis of signal changes in the tissue as the labelled hydrogen flows through the vascular system of the organ. A perfusion-weighted image is created by subtracting the spin-labeled image from a control image, which is scanned in the same way but without the spin labels. Since cerebral perfusion normally changes approximately 1% of the water molecules contained in the brain in 1 s, the usual signal change in the labelled image is approximately 2% compared to the control image. Therefore, repeated measurements of pairs of labeled and control images are recommended to improve the quality of the resulting image. To quantify the typical cerebral blood flow (CBF) perfusion parameter, it is also necessary to measure the M0 (equilibrium magnetization) image to obtain an equilibrium value of blood magnetization to normalize the perfusion-weighted image.

 

ASL variants and parameters

Arterial spin labeling has different technical implementations of the sequence -⁠ PASL (pulsed ASL) and PCASL (pseudo-continuous ASL) (Figure 1) [16]. Spin labeling is performed using radiofrequency inversion pulses, and in the PCASL method, the area of their application is usually selected around the C2/C3 vertebrae, ideally in a plane perpendicular to the carotid artery (Figure 2) [5]. For the PASL method, a 15-20 cm long marking area is chosen, which is located 2-4 cm below the scanning area. The marking length is recommended to be 1,800-2,000 ms for adult PCASL methods, whereas it is almost instantaneous for PASL methods. "The 'waiting' time after spin labeling and before scanning, referred to as the post-labeling delay (PLD), is recommended to be in the range of 1 800-2 000 ms for PCASL and 1 600-2 000 ms for PASL. The control image is acquired using the same setup but without the inversion pulse spin labeling.

It is recommended to calculate the CBF according to the article by Alsop et al. [4]. Standard sequences of scanner manufacturers usually allow the calculation of CBF maps, but the algorithms for this calculation are not freely available, which makes it impossible to verify their compliance with published recommendations. In our experience, the results in some cases differ from those obtained by the recommended tools. Therefore, when quantitatively assessing CBF, e.g. in longitudinal studies or when comparing with normative values, we recommend using external tools and reconstruction algorithms. It is also advisable to verify the compatibility of the archival format with the scanner and advanced postprocessing before implementing ASL in clinical practice. For relative perfusion comparisons between regions, e.g. for visual identification of hypoperfused areas, the manufacturer's basic reconstruction algorithm can be used if the PLD is set up correctly. Several tools exist for processing ASL data. Among the most used are ExploreASL [17], FSL BASIL [18], nordicIce, Quantifiphyse and others. More information about the available tools and additional tutorials and terminology overview can be found in the ISMRM OSIPI (Open Science Initiative for Perfusion Imaging) resources [19].

One of the main assumptions of the perfusion model of ASL is that the PLD must be longer than the transit time of the labeled spin through the artery to the tissue of interest, otherwise the method loses its reliability. The labeled spin begins to relax immediately with the T1 time of the blood (approximately 1,650 ms in the 3T field), which significantly affects the quality of the resulting images. Because of the relatively short T1 time of blood, it is important to perform ASL before the possible administration of contrast agent, which due to the paramagnetic properties of blood shortens the T1 time even more. To increase the reliability of the measured perfusion, scanning with several different PLDs is also possible, which allows better modelling of the temporal signal profile and avoids artefacts during slowed blood flow through the artery, e.g. during stenoses or occlusions. Another approach to improve image quality is to suppress the signal from static tissue (background suppression) using a series of radiofrequency pulses between marking and image reading.

 

ASL artefacts

For correct interpretation of perfusion scans, knowledge of possible ASL-specific artefacts that can significantly affect the diagnostic evaluation is essential. We distinguish between artifacts that occur during spin labeling, spin passage through the artery, and during scanning [20].

Artifact markings:

                Failure of spin labeling in tortuous vessels due to a poorly planned labeling plane, which manifests as a lower or completely absent perfusion signal in the corresponding vascular territory.

Labeling of spins in the cerebrospinal fluid that moves into the scan volume, thereby creating a false perfusion signal in the perimedullary space.

 

Spin Passage Artifacts:

                The arterial transit time artifact occurs when the image was scanned before perfusion of the marked spins into the tissue and the signal remains prominent mainly in the feeder vessels, usually wider than 0.1 mm. This artefact indicates that the selected PLD is shorter than the artery transit time of the spin, and is often seen in elderly patients with reduced cardiac output, arterial stenosis or aneurysm.

Conversely, direct or rapid transit of blood from arteries to venous structures leads to increased signal in the veins and occurs in patients with arteriovenous shunts. In such a disease, the marked spins do not undergo tissue exchange.

Perfusion signal readout artifacts:

Since ASL is a subtraction method, motion has a significant negative effect on the quality of the resulting CBF images, which are blurred and/or contain significantly increased signal around the skull.

                Increased signal in occipital areas is usually a manifestation of physiological noise caused by activation of the visual cortex.

 

In the following sections, we discuss examples of the use of the method for advanced clinical diagnosis in a number of neuropathologies, including cerebrovascular diseases, brain tumors, epilepsy, and neurodegenerative processes (Figure 3). We conclude with new directions in ASL development and their potential contribution to clinical practice.

 

Cerebrovascular diseases

By definition, cerebrovascular diseases refer to a group of disorders in which cerebral blood vessels are damaged and cerebral perfusion is altered leading to ischemia or hemorrhage. It is for these reasons, i.e., significant alteration of perfusion, that ASL is the most logical clinical application. In steno-occlusive diseases, ASL has the potential to identify the ischemic area, which is characterized by reduced perfusion, and to localize the collateral pathway, which in turn is characterized by increased perfusion. An important factor in imaging perfusion in cerebrovascular disease with ASL is the prolonged arterial blood transit time, which may lead to the aforementioned arterial transit time artifact. Therefore, the choice of longer PLD and/or multi-PLD ASL is recommended here for higher quantification accuracy.

In patients with acute ischemic stroke, ASL can be used similarly to DSC to assess PWI-DWI (perfusion -⁠ and diffusion-weighted images) mismatch to determine penumbra, i.e. whether reperfusion therapy can benefit the patient [21,22]. ASL can also help to visualize collateral flow or arterial occlusions that cause delayed flow and thus a significantly increased CBF signal in the vessels [23,24]. ASL can also be used to assess perfusion changes after revascularization procedures, e.g., in patients with moyamoya disease [25], where post-vascularization hyperperfusion (luxury perfusion) and the corresponding stealth phenomenon are often seen (Figure 4). ASL is also useful for the identification of cerebral arteriovenous malformations and fistulas [26] due to the labeled spins that bypass the microvascular network and enter directly into the venous circulation or drainage veins, allowing their easy identification. These findings are characterized by a high perfusion signal outside the arterial tree in the region of the nidus or venous system. When evaluating cerebrovascular pathologies with ASL, increased caution should be exercised in clinical conditions where alterations in microvascular flow are expected, such as brain tumors and sickle cell anemia, which may indicate capillary shunting.

 

Brain tumours

Blood perfusion in tumor areas and peritumoral edema significantly correlates with angiogenesis, which is an important marker for grading and classification [27]. Because of the need for contrast administration in MR imaging of tumors, methods of imaging perfusion and blood-brain barrier permeability using contrast agent have traditionally been more widely used in practice compared with ASL. However, the increasing use of AI methods aimed at MR imaging with less contrast agent administration [28] or non-contrast MR [29], the validation of advanced non-contrast MR methods incl. ASL, T2* DSC artifacts around the tissue interface and leakage artifacts in cases of broken blood-brain barrier motivate the use of ASL in clinical neuro-oncology.

Arterial spin labeling allows perfusion imaging of primary and secondary intra -⁠ and extra-axial tumors while allowing repeat imaging for initial diagnosis, therapy monitoring, or monitoring the progress of resection [30]. ASL-derived CBF has the potential to differentiate between tumor stages [31], to distinguish pseudoprogression from true progression [32], and to noninvasively classify tumors according to genetic markers such as IDH and pTERT [33]. Higher intratumoral CBF values are indicative of increased vascular density and thus higher tumor aggressiveness, whereas elevated peritumoral CBF values are a hallmark of gliomas as opposed to hypervascular metastases [34].

For basic assessment of tumor perfusion and edema for clinical diagnosis, standard ASL imaging with a single PLD is appropriate. However, heterogeneous tumor vascularity, vessel compression by peritumoral edema [35], blood-brain barrier disruption, and vascular shunting [36] may bias CBF measurements. Therefore, either longer PLD times (>2000 ms) or scanning with multiple PLDs are recommended for more sensitive and accurate assessment of perfusion. It is also recommended to normalize CBF in the tumor or edema using CBF values in the contralateral, normally structured gray matter, obtained either by region of interest (ROI) location [37] or by automated segmentation of the entire hemisphere. An example of processed ASL and DSC scans in a patient with a brain tumor is shown in Fig. 5.

 

Epilepsy

Epileptiform activity can significantly alter metabolism and perfusion patterns in the epileptogenic zone (EZ) and its elocentric regions [38]. Identification of the EZ remains a significant challenge, especially in the case of non-lesional epilepsy (MR negative), which represents a very complex subset of patients often undergoing epileptogenic surgery, characterized by complicated design and long-term failure rates exceeding 50% [39]. Widespread advanced diagnostic imaging modalities for the detection of EZ include fluorodeoxyglucose PET (FDG-PET) and subtraction ictal SPECT, which is sensitive to ictal perfusion changes. ASL may play a significant role in such indications. A meta-analysis focusing on the diagnostic accuracy of ASL in detecting the epileptogenic zone shows a high pooled sensitivity of 0.74 [40], making it a suitable adjunct to clinical multimodal imaging protocols. Although a number of previous studies comparing ASL and FDG-PET have concluded good agreement between the methods, discrepancies remain, which explains why ASL serves as a complementary rather than alternative tool to FDG-PET [41-43]. One reason for the difference is the different parameters measured, perfusion and metabolism, which correlate with each other in the healthy brain by neurovascular coupling. However, in the case of EZ, the assumption of neurovascular coupling may not be valid [44,45], which may lead to a decorrelation of perfusion and metabolism in the ictal, postictal and/or pericardial phases. Previous studies have shown that the EZ exhibits hyperperfusion in the ictal phase and, in frequent seizures, in the interictal phase (Figure 6) and, conversely, hypoperfusion in the interictal and postictal phases [46,47]. This inconsistency is probably due to the different time interval between the last seizure and the acquisition of MR scans. This interaction has not yet been systematically addressed in the literature. A recent study also reveals an association between seizure duration and postictal perfusion changes; longer seizures correlate with more pronounced hypoperfusion, whereas shorter seizures lead to hyperperfusion [48].

 

Neurodegenerative diseases

Changes in cerebral perfusion are characteristic of a number of neurodegenerative diseases, incl. Alzheimer's disease, frontotemporal dementia and Parkinson's disease. These diseases may manifest early functional deficits before the onset of more pronounced structural changes [49,50]. Since the first signs of neurodegenerative changes, such as memory impairment, have limited indications for contrast agent administration in the early stages, ASL is a suitable method for initial monitoring of neurodegeneration. ASL has the potential to differentiate Alzheimer's disease from frontotemporal dementia and provides useful markers that are associated with disease progression and amyloid burden [51,52].

Due to age-related physiological changes as well as pathological changes associated with neurodegeneration, longer PLDs (2000-2 500 ms) are recommended for optimal acquisition of ASL images. In addition, early stages of neurodegeneration may cause only subtle changes in regional perfusion; therefore, general physiological factors affecting perfusion should be considered and perfusion parameters should be interpreted with caution.

 

Other directions of ASL development

Recently, new variants and improvements of ASL acquisition and ancillary measurements have been developed to improve image quality, provide more accurate quantification of cerebral blood flow, or measure other physiological parameters [53]. One emerging method is the measurement of cerebrovascular reserve (CVR) using a cerebral stress test with acetazolamide administration or hypercapnia induction [54]. CVR measurement has potential for the diagnosis of cerebrovascular disease [55] and brain tumors [56].

The increased prevalence of stronger magnets and technological advances have enabled the development of ASL-based techniques that measure not only perfusion but also blood-brain barrier permeability, such as multi-TE ASL (Figure 7) [57] and diffusion-weighted ASL [58]. These newly developed methods are expected to be more sensitive to less pronounced blood-brain barrier disruptions due to the smaller size of water molecules compared to gadolinium-based contrast agents.

One of the important innovations in ASL is velocity-selective ASL (VSASL), which eliminates the problems of vessel transit time artifact and can significantly improve the signal-to-noise ratio [59]. VSASL differs from other ASL acquisition methods by its labeling strategy, which is not limited by spatial localization and uses rate coding. Another direction in the development of ASL application is non-contrast angiography, which allows imaging of both large vessels and microvasculature.

 

Conclusion

Arterial spin labeling is a non-invasive method of perfusion MR that is available in MR scanners from all major manufacturers and is relevant for many clinical applications. Greater use of ASL in clinical practice has so far been hampered mainly by the lack of validation in large clinical trials that demonstrate specific patient benefits. At the same time, there is a general lack of experience with data acquisition and reading and interpreting ASL images in clinical practice. However, after a long phase of technical development, attention is now turning primarily to these practical issues, and greater use of ASL in the diagnosis of cerebrovascular diseases, neuro-oncological pathologies, epilepsy and neurodegeneration can be expected in the near future. ASL is already a suitable alternative to conventional contrast perfusion (DSC) and is particularly useful in clinical risk groups such as paediatric patients and patients with renal impairment.

Acknowledgements

We would like to express our sincere gratitude to Jan Brhel and Veronica Borovcova, radiology assistants, who were invaluable in helping to install and test the MR research protocols and in obtaining data for our studies. We would also like to thank our colleagues from the Department of Neurology (doc. Aleš Tomek, prof. Petr Marusič, MUDr. Adam Kalina), the Department of Child Neurology (prof. Pavel Kršek, ing. Radek Janča, MUDr. Anežka Bělohlávková, MUDr. Alena Jahodová, MUDr. Martin Kudr and MUDr. Matyáš Ebel), the Department of Paediatric Haematology and Oncology (doc. Michal Zápotocký, MUDr. David Sumerauer, MUDr. Kateřina Trková) and the Department of Neurosurgery for Children and Adults (doc. Vladimír Beneš, MUDr. Jana Táborská).

 

Financial support

Supported by project No. LX22NPO 5107 (Ministry of Education and Science): funded by EU -⁠ Next Generation EU. Supported by grants from the Agency for Medical Research of the Ministry of Health of the Czech Republic with reg. no. NU21-08-00228 and NU23-08-00460.

 

Conflict of interest

The authors declare that they have no conflict of interest in relation to the subject of the paper.