Arterial Spin Labeling (ASL) — Generic Standard Protocol

Arterial Spin Labeling (ASL) — Generic Standard Protocol

MRIninja Knowledge Base | Master / General Protocol Page Cluster: Perfusion Imaging — Non-Contrast Techniques Version 1.0 — August 2026

1. Executive Summary

Arterial spin labeling (ASL) is a non-contrast MRI perfusion technique that uses magnetically labeled arterial blood water as an endogenous, freely diffusible tracer to quantify cerebral blood flow (CBF, expressed in mL/100 g/min). Unlike dynamic susceptibility contrast (DSC) perfusion MRI — which requires gadolinium bolus injection — ASL uses the patient's own arterial water protons as the tracer, producing quantitative perfusion maps without contrast administration, ionising radiation, or dependence on an intact blood-brain barrier.

This fundamental difference defines ASL's clinical niche: it is the method of choice for perfusion imaging when gadolinium must be avoided (renal impairment, paediatric patients, serial monitoring), when quantitative absolute CBF is required (longitudinal dementia follow-up, cerebrovascular reserve testing), and when the pattern of hypoperfusion itself is diagnostically informative without the need for dynamic bolus-tracking kinetics.

The 2015 ISMRM Perfusion Study Group / European ASL in Dementia consortium consensus paper [1] established the reference implementation: PCASL labeling, background suppression, 3D segmented readout, M0 calibration, and age-adjusted single post-label delay (PLD). A 2023 ISMRM update from the same group extended these recommendations to disease-specific clinical scenarios [2]. These two documents form the evidence base for this protocol.

ASL SNR is inherently low — the perfusion signal is approximately 0.5–1.5% of the static tissue signal — which explains why background suppression, signal averaging, 3T field strength, and 3D readouts are all components of the consensus standard protocol.

1.1 Core Strengths

No contrast requirement: the primary clinical advantage. ASL is the only perfusion MRI technique that provides quantitative perfusion maps without gadolinium, making it universally applicable in patients with eGFR < 30 mL/min/1.73 m², history of GBCA allergy, or patients requiring serial monitoring without cumulative gadolinium exposure (dementia follow-up, childhood epilepsy, cerebrovascular disease monitoring).

Quantitative absolute CBF: ASL produces CBF maps in absolute physiological units (mL/100 g/min) — not simply relative perfusion ratios. Normal grey matter CBF at rest: 50–80 mL/100 g/min; white matter: 20–30 mL/100 g/min. This quantitative output enables: normative database comparison; longitudinal within-subject comparison; and, in multi-centre studies, cross-site reproducibility when protocol is standardised.

Repeatability for longitudinal monitoring: because ASL requires no exogenous tracer, it can be repeated as frequently as clinically necessary within a single session or across serial examinations. This makes it uniquely suited to cerebrovascular reserve testing (pre/post acetazolamide or CO2 challenge), post-revascularisation monitoring in Moyamoya disease and carotid stenosis, and dementia progression monitoring.

Clinically useful perfusion pattern recognition: the characteristic ASL hypoperfusion territories of AD (posterior cingulate, precuneus, angular gyrus), FTD (frontal predominant), and DLB (occipital hypoperfusion with relative posterior cingulate sparing — the "cingulate island sign") can guide differential diagnosis of neurodegenerative syndromes without gadolinium or FDG-PET [1,2,7].

1.2 Intrinsic Limitations of the Generic Protocol

Low intrinsic SNR: the 0.5–1.5% perfusion signal requires averaging, background suppression, and 3T field strength to achieve diagnostic quality. At 1.5T, ASL is feasible but less reliable; for thin or focal perfusion deficits, 3T is required.

Arterial transit time (ATT) confound: single-PLD ASL assumes labeled blood arrives at the imaging voxel before the chosen PLD. When ATT is prolonged (steno-occlusive disease, severe carotid stenosis, collateral flow in Moyamoya) or highly variable (elderly patients), the single-PLD assumption fails — producing underestimation of CBF and characteristic arterial transit artefacts. Multi-PLD acquisition corrects for this but is more time-consuming and requires kinetic-model post-processing.

No BBB-independent enhancement information: unlike DSC or DCE perfusion, ASL does not provide enhancement kinetics or BBB permeability information. For tumour grading where rCBV and leakage correction are the primary metrics, DSC remains the reference standard. ASL provides tumour CBF information that is complementary, not equivalent.

When dedicated child protocols are required: ASL in brain tumour grading and treatment response; ASL in neurodegenerative dementia differential diagnosis (AD, FTD, DLB); ASL in steno-occlusive disease and Moyamoya (multi-PLD acquisition, cerebrovascular reserve testing); ASL in epilepsy (ictal and interictal); ASL in acute stroke; ASL in paediatric neuroimaging; multi-PLD and time-encoded ASL advanced protocols.

2. Main Clinical Indications

2.1 Standard Indications

Neurodegenerative dementia work-up is the leading clinical indication for ASL in most departments. The characteristic regional hypoperfusion patterns of AD (posterior cingulate, precuneus, temporoparietal), FTD (frontal and anterior cingulate), and DLB (occipital, with the cingulate island sign — relative preservation of posterior cingulate CBF despite global hypoperfusion) are reliably detected at 3T with the standard single-PLD PCASL protocol [2,3]. ASL provides functional information that structural MRI alone cannot. In centres without FDG-PET, ASL is the recommended perfusion surrogate for the dementia differential.

Brain tumour perfusion assessment is the second major indication. Tumour CBF from ASL correlates with DSC-derived rCBV for high-grade glioma detection, with pooled sensitivity 90% and specificity 91% across meta-analyses [4,5]. ASL is particularly useful when DSC is technically suboptimal (large haemorrhagic component, metalwork susceptibility artefact preventing DSC bolus tracking) and for serial follow-up where accumulated gadolinium is a concern.

Cerebrovascular disease and steno-occlusive disease — ASL detects haemodynamic impairment (territorial hypoperfusion, arterial transit artefact as collateral flow marker) in patients with carotid stenosis, large vessel occlusion, and Moyamoya disease. Cerebrovascular reserve testing (pre/post acetazolamide or CO2 challenge, with ASL at each time point) quantifies the haemodynamic reserve non-invasively. Note that single-PLD ASL in this indication requires the arterial transit artefact to be recognised and correctly interpreted as collateral flow, not true hyperperfusion [2,6].

Epilepsy focus lateralisation — the epileptogenic zone shows increased CBF ictally and reduced CBF interictally. ASL-detected ictal hyperperfusion has reported concordance with SPECT/PET, and interictal hypoperfusion is useful for focus lateralisation in drug-resistant epilepsy evaluation [2].

Paediatric perfusion imaging — ASL is particularly valuable in children because it avoids gadolinium administration and radiation. Indications include: arteriovenous malformation assessment, post-stroke perfusion, Moyamoya, and neurodegenerative conditions. Age-specific PLD adjustment is critical — children have faster circulation than adults and require shorter PLD (general guidance ≥ 1500 ms, but paediatric-specific protocols use further adjustment by age and condition) [2,7].

Renal impairment — gadolinium contraindicated: any perfusion-dependent clinical question (brain tumour, dementia, stroke) in a patient with eGFR < 30 mL/min/1.73 m² or history of severe GBCA reaction becomes an ASL indication by default [1].

2.2 Urgent Red Flags

ASL is not an emergency modality. The 3–5 minute ASL acquisition requires patient co-operation and is typically performed as part of a planned brain MRI protocol. However:

3. Preparation Reference

Universal MRI safety screening belongs to the general MRI preparation page and is not repeated here.

3.1 Anatomy-Specific Preparation Items

No contrast required: ASL requires no gadolinium administration and no IV access in the standard protocol. This is its primary operational advantage — no pre-examination renal function check is required for ASL-only protocols.

Patient co-operation and stillness: because the perfusion signal derives from subtraction of paired control/label acquisitions, inter-pair motion corrupts the subtraction and produces artefacts. Patient instructions about remaining still during the 3–5 minute ASL acquisition are particularly important. Sedation should be considered for patients with movement disorders or inability to co-operate with a 4–5 minute still period.

Caffeine and vasodilatory drugs: caffeine is a cerebral vasoconstrictor that reduces CBF. Caffeine consumption in the 4–6 hours before ASL imaging reduces global CBF by approximately 20–30% and may produce false-positive apparent hypoperfusion. Instruct patients to avoid caffeine for at least 4 hours before the examination. Document caffeine status in the protocol notes.

Cerebrovascular reserve testing (acetazolamide/CO₂ challenge): when ASL is used for cerebrovascular reserve testing, the pharmacological or physiological challenge must be administered with precise timing relative to the ASL acquisition. This is a child-page-specific preparation requirement.

Labeling plane position may be affected by carotid anatomy: in patients with tortuous carotid arteries, atherosclerotic plaques near the planned labeling level (C2/C3), or prior carotid surgery, the labeling efficiency may be reduced. A quick review of prior neck imaging or carotid ultrasound results before the examination is worthwhile when ASL is the primary clinical question.

3.2 Patient Positioning on the MRI System

Position: supine, head-first. Standard 16–32 channel head coil. Identical to standard brain MRI positioning.

Isocentre: at the level of the AC-PC line (anterior commissure – posterior commissure), ensuring the brain is at the B0 field centre. This is the standard brain MRI isocentre and is not modified for ASL.

Labeling plane planning (technologist responsibility): after positioning and before starting the ASL acquisition, the labeling plane must be placed perpendicular to the carotid and vertebral arteries at the level of C2/C3. This requires a sagittal localiser review. The labeling plane is the most critical planning step for ASL and is described in full in Section 4.6.

Head symmetry: standard brain MRI requirements. No rotation.

4. Standard Protocol Design

The standard ASL protocol follows the ISMRM 2015 consensus recommendations [1] as the baseline, updated by the 2023 ISMRM clinical guidance [2].

4.1 Mandatory Core Sequences

4.2 Conditional Sequences

4.3 Rationale Summary Per Sequence

PCASL Labeling — Why PCASL and Not PASL or CASL

The 2015 consensus paper [1] recommends PCASL as the optimal default labeling scheme, superseding both PASL and continuous ASL (CASL). The reasons:

vs PASL: PASL uses a single short adiabatic inversion pulse applied to a thick slab. It has high instantaneous labeling efficiency but produces an ill-defined bolus with variable duration — making absolute CBF calculation dependent on assumptions that are harder to validate. PCASL's train of discrete RF pulses with gradient blips approximates continuous flow-driven adiabatic inversion at a thin labeling plane, producing a well-defined labeling duration that directly enters the kinetic model and improves quantification accuracy.

vs CASL: CASL used a long continuous RF pulse (2–4 seconds) — theoretically ideal, but the continuous RF produces large magnetisation transfer (MT) effects on brain tissue, distorting the control/label difference signal. It also has hardware compatibility limitations on clinical MRI systems. PCASL achieves similar efficiency to CASL without the MT problem and with full hardware compatibility.

PCASL labeling plane placement: the labeling plane must be placed perpendicular to the major labeling vessels (carotid arteries and vertebral arteries) at approximately the C2/C3 level. The reason for this specific level: (a) the arteries are relatively straight and parallel at C2/C3, maximising flow-driven adiabatic inversion efficiency; (b) it is sufficiently proximal to ensure complete labeling of blood before it enters the cranial vault; (c) it avoids the carotid bifurcation (below) and the skull base foramina (above) where geometry complicates labeling.

3D Segmented Readout (GRASE or Stack-of-Spirals FSE) — Why Not 2D EPI

The 2015 consensus [1] recommends a 3D segmented readout over 2D single-shot EPI. The key reasons:

Background suppression compatibility: background suppression (BS) — multiple adiabatic inversion pulses applied to static tissue immediately before readout — is the most important strategy for reducing the static tissue signal and improving perfusion signal detectability. BS is most effective with 3D segmented readouts because the readout is spin-echo-based (less sensitive to T2* inhomogeneity); with 2D EPI, BS must be applied slice-by-slice, creating slice-to-slice inconsistency in the degree of suppression.

Reduced geometric distortion: 3D GRASE and stack-of-spirals use short readout segments that accumulate less B0-induced phase than the long single-shot EPI echo train. The resulting geometric distortion is substantially less severe — important for preserving anatomical accuracy of the CBF map relative to structural sequences.

Higher SNR efficiency: 3D acquisition provides √N_z SNR advantage (see the 2D vs 3D Acquisition parameter deep dive). For the inherently low SNR of ASL, this advantage is significant — it is one of the reasons the consensus recommendations shifted toward 3D readouts.

Slice-timing-dependent effective PLD: in 2D multi-slice EPI, each slice is acquired at a different time after labeling. The effective PLD varies by slice — producing slice-dependent CBF quantification errors. 3D acquisitions collect all slices within a short, well-defined readout window after a single PLD, eliminating this confound.

Background Suppression — The SNR Enabler

Background suppression (BS) applies adiabatic inversion pulses to static tissue immediately before the ASL readout. The goal: null the large static tissue signal (which contributes only noise to the control–label difference) while preserving the small perfusion-weighted signal (which has already exchanged into tissue and has a different T1 trajectory). This dramatically improves the signal-to-noise ratio of the control–label difference image.

Without BS: the static tissue signal dominates the image; motion between control and label acquisitions produces large subtraction artefacts. With BS: static tissue signal is suppressed to near-zero; inter-pair motion produces much smaller residual artefacts; the perfusion signal is relatively more conspicuous.

BS is incompatible with certain clinical uses of the control image (e.g., as an anatomical reference) but this is a minor trade-off for the substantial SNR gain in the difference image.

M0 Calibration Scan — Enabling Absolute CBF

CBF quantification requires dividing the control–label difference signal by a reference signal proportional to the equilibrium magnetisation of blood (M0). Without M0, only qualitative relative perfusion maps can be generated. The M0 calibration scan is a proton-density-weighted acquisition (very long TR ≥ 6 s, no BS) acquired as part of the same protocol sequence. The resulting M0 map, combined with known acquisition parameters (labeling efficiency α, blood T1, PLD, labeling duration), enables absolute CBF calculation in mL/100 g/min.

T1 Structural (MPRAGE) — Anatomical Reference and Partial Volume Correction

The T1-weighted 3D structural is not specifically an ASL sequence, but its acquisition in the same session provides: (a) anatomical reference for CBF map overlay and regional analysis; (b) input for grey/white matter segmentation required for partial volume correction algorithms; (c) voxel-to-voxel co-registration enabling accurate regional CBF extraction for ROI-based analysis (hippocampal CBF in dementia, tumour CBF extraction, etc.).

4.4 Sequence Matching and Cross-Sequence Consistency

The CBF map must be co-registered with the structural T1 for regional analysis. Most vendor ASL post-processing pipelines perform this registration automatically. Verify that: (a) the CBF map and T1 cover the same anatomical volume; (b) the registration is adequate (no residual displacement at the cortical surface or in the basal ganglia).

For longitudinal serial ASL monitoring (dementia follow-up, post-treatment tumour, Moyamoya post-revascularisation): the same labeling scheme, PLD, readout, field strength, and M0 protocol must be reproduced at each time point. Changes in any of these parameters produce systematic CBF differences that cannot be attributed to true clinical change.

4.5 Fat Suppression

ASL uses 3D readouts (GRASE/spiral FSE) that are naturally T2-weighted (spin-echo based). Fat suppression is not a standard component of the ASL acquisition. The perfusion-weighted image and the derived CBF map are not affected by fat signal in the same way as standard T1/T2 sequences because the short-T2 fat signal does not contribute meaningfully to the long-TE window of the 3D ASL readout.

4.6 Slice Positioning — Complete Technical Reference

The Labeling Plane — The Critical ASL-Specific Positioning Step

Unlike any other brain MRI sequence, ASL requires positioning a spatially distinct labeling zone (the labeling plane or labeling slab) separate from the imaging volume. This positioning step has no equivalent in standard brain MRI and is the most important technologist skill specific to ASL.

Labeling Plane Anatomy

Target vessels: internal carotid arteries (ICA) and vertebral arteries (VA), which supply approximately 85% and 15% of cerebral blood flow respectively. Both must be included within the labeling volume for complete brain perfusion labeling.

Optimal level: C2/C3 vertebral body level. At this level, the ICA and VA are: running in approximately vertical, straight trajectories (maximising flow-driven adiabatic inversion efficiency — the vessels must be perpendicular to the labeling plane); well below the skull base (no interference with the brain imaging volume); well above the carotid bifurcation (avoiding labeling of external carotid branches that do not contribute to brain perfusion); at a consistent, reproducible anatomical landmark.

Published guidance [1,2]: the labeling plane should be placed perpendicular to the carotid/vertebral arteries at the neck, with the labeling plane positioned approximately 20–25 mm inferior to the bottom of the imaging volume (to ensure labeled spins travel through the entire circulation distance before reaching the imaged brain).

Planning Sequence

• Sagittal localiser of the head and upper neck (standard brain sagittal includes the upper cervical vertebrae)
• Identify C2/C3 disc level on the sagittal localiser
• Place the labeling plane at C2/C3, perpendicular to the cervical spine (and therefore perpendicular to the longitudinal axes of the ICA and VA at this level)
• Verify the gap between the inferior border of the imaging volume (brain) and the superior border of the labeling plane: this gap should be ≥ 20 mm (usually achieved automatically when the labeling plane is at C2/C3 and the imaging volume covers the brain)
• Verify on the axial localiser that the labeling plane encompasses both ICA and VA

Imaging Volume Positioning

The imaging volume (the 3D brain slab from which CBF maps are derived) follows standard brain MRI positioning: parallel to the AC-PC line, covering the entire brain from vertex to cerebellum/brainstem. This is identical to any standard 3D brain acquisition.

Phase Encoding Direction

The 3D segmented readout (GRASE or stack-of-spirals) uses k-space ordering within each segment. Phase encoding direction: A-P for axial acquisitions. This is the standard for brain imaging and is not modified for ASL. See the Fold-over Direction parameter deep dive for the general principles governing this choice.

Common Labeling Plane Positioning Errors

Too close to the imaging volume (labeling plane < 10 mm from imaging volume): labeled spins may not have sufficient transit time to exchange into brain tissue before the PLD expires → underestimation of CBF; also, the labeling RF pulse may partially excite brain tissue at the skull base → image artefact.

Angled through the vessel obliquely (labeling plane not perpendicular to ICA/VA axis): flow-driven adiabatic inversion efficiency is maximised when blood flows perpendicularly through the labeling plane. Even 20–30° angulation reduces labeling efficiency by 15–30%, producing global CBF underestimation.

Labeling plane placed through the carotid bifurcation: at the bifurcation level, the ECA branches are also labeled. These labeled spins perfuse facial and scalp tissue — not brain — and do not contribute to brain CBF. Their inclusion dilutes the effective labeling efficiency for brain perfusion. Place the labeling plane at C2/C3, above the bifurcation.

Not checking for tortuous ICA: some patients have highly tortuous ICA at the C2/C3 level. If the vessel is running parallel to the labeling plane at that level (rather than perpendicular), labeling efficiency collapses. Visualise the ICA trajectory on the sagittal localiser before finalising the labeling plane position.

5. Optimisation Strategy

5.1 Artefact Reduction by Source

Arterial transit artefact (ATA) / transit-time bias — the most important ASL-specific artefact. When ATT exceeds the chosen PLD, labeled spins remain intravascular at the time of imaging rather than having exchanged into tissue. This produces serpentine/gyriform high-signal bands along superficial arteries and sulci — characteristic of collateral or delayed flow — rather than the smooth parenchymal CBF map expected from normal anterograde flow [6].

• Physical cause: PLD too short for the local ATT; most common in steno-occlusive disease, Moyamoya, severe carotid stenosis, and older patients with generally prolonged ATT
• Appearance: bright serpentine signal following the course of cortical arteries; does not follow expected territorial perfusion boundaries
• What it mimics: can be misread as true hyperperfusion in the affected territory — the opposite of the true haemodynamic state. In Moyamoya, the ATA reflects delayed collateral blood, not increased perfusion
• Reduction: increase PLD (from 1800 to 2000 ms or more); use multi-PLD acquisition to estimate ATT and correct for it; recognise the pattern for its collateral-flow significance
• When invalidating: when the CBF map shows widespread serpentine artefact and no meaningful territory can be assessed quantitatively — switch to multi-PLD acquisition

Motion artefact — the most common quality-degrading problem in routine clinical ASL.

• Physical cause: inter-pair motion between the control and label acquisitions produces subtraction artefacts (rim-like bright/dark signal at tissue boundaries)
• Appearance: ring-like bright signal at the brain surface; dark signal immediately interior to the bright rim; may simulate a hyperperfusion cortical pattern or mask true hypoperfusion
• Reduction: background suppression (reduces static tissue signal, reducing the amplitude of motion-induced subtraction artefact); volume-selective motion outlier rejection during post-processing; patient instruction and head immobilisation; sedation when necessary
• When invalidating: extensive rim artefact throughout the cortex making regional CBF assessment unreliable

Labeling plane malposition — produces systematically incorrect CBF values.

• Physical cause: labeling plane not perpendicular to ICA/VA; labeling plane too close to imaging volume; labeling plane through carotid bifurcation (includes ECA labeling)
• Appearance: globally reduced CBF (suggesting global hypoperfusion); or asymmetric reduction (asymmetric ICA labeling from tortuous vessel)
• Reduction: verify labeling plane placement on sagittal localiser before every acquisition; standard C2/C3 perpendicular positioning as described in Section 4.6
• When invalidating: global apparent hypoperfusion with no clinical correlate — always suspect labeling plane malposition as a first explanation

2D EPI-specific artefacts (when 2D EPI readout is used instead of 3D segmented):

• Susceptibility distortion at skull base and temporal poles (standard EPI distortion from air-tissue interfaces — see the Fold-over Direction deep dive for the underlying mechanism)
• Slice-dependent effective PLD producing z-dependent CBF bias
• Inferior background suppression performance (slice-by-slice BS inconsistency)
• Reduction: switch to 3D segmented readout (GRASE or stack-of-spirals) per consensus recommendation [1]

Insufficient signal averaging — produces noisy CBF maps.

• Physical cause: too few control-label pairs (NSA too low for the inherently low ASL SNR)
• Appearance: grainy, patchy CBF map with apparent focal abnormalities that are actually noise
• Reduction: standard protocol uses 30–60 control-label pairs (2–4 minutes at 2 pairs/8 s); at 1.5T, more pairs may be needed; at 0.55T, ASL is generally not clinically viable
• When invalidating: CBF map SNR insufficient for regional analysis — rare at 3T with standard pair count, more common at 1.5T

5.2 Protocol Efficiency and Throughput

The standard PCASL + 3D GRASE protocol at 3T requires approximately 3–5 minutes for the ASL acquisition itself. With the M0 calibration scan (30–60 seconds) and the anatomical T1 (typically already acquired as part of the standard brain protocol), total ASL-specific scan time is 4–6 minutes. This is entirely manageable within a standard 30–40 minute brain MRI protocol.

Abbreviated protocol (for scan-time-constrained settings): PCASL + 3D GRASE + M0 alone (without full structural T1) provides a clinically useful CBF map in approximately 4 minutes. The absence of the structural T1 precludes automated regional analysis but is adequate for qualitative and semi-quantitative CBF pattern assessment.

Multi-PLD protocol: adds 2–4 minutes for the additional PLDs. Time-encoded (Hadamard) ASL achieves multi-PLD sampling within a scan time comparable to single-PLD (3–5 minutes), at the cost of more complex post-processing.

5.3 Field Strength Considerations

3T is the preferred clinical standard for ASL for three independent reasons:

1. Higher blood T1 at 3T (T1_blood ≈ 1700 ms at 3T vs ≈ 1200 ms at 1.5T): longer blood T1 means less label decay during the post-label delay → higher perfusion signal at the imaging voxel → better SNR for a given PLD
2. Higher intrinsic SNR (approximately 1.7–2× compared to 1.5T): directly benefits the low-SNR ASL signal
3. Better background suppression performance at 3T (longer tissue T1 at 3T means inversion nulling is more complete at the time of readout)

1.5T ASL is feasible but requires: more control-label pairs (longer scan time) to achieve adequate SNR; slightly shorter PLD may be optimal due to shorter blood T1; reduced reliability for thin grey matter structures and subtle focal hypoperfusion. Single-PLD 1.5T ASL is adequate for detecting gross hemispheric CBF asymmetry and large territorial hypoperfusion (stroke, large AVM) but less reliable for the subtle regional patterns of neurodegeneration.

0.55T ASL: not clinically viable with current technology for the standard protocol. Blood T1 at 0.55T (≈ 750 ms) produces rapid label decay; the perfusion signal is so small as to be indistinguishable from noise at standard pair counts within any clinically acceptable scan time.

SAR at 3T: PCASL labeling trains add RF power to the sequence. At 3T, especially in combination with high-flip-angle structural sequences or other SAR-intensive sequences in the same session, PCASL SAR contribution must be monitored. Standard PCASL with typical labeling flip angle (15–25°) and labeling duration (1.5–2 s) does not typically exceed SAR limits when combined with a standard brain protocol, but SAR monitoring is mandatory.

6. Contrast Use Principles Specific to ASL

6.1 Non-Contrast — Standard Protocol

ASL is by definition a non-contrast technique. The standard clinical ASL protocol requires no contrast agent. This is its defining advantage over DSC and DCE perfusion.

6.2 Gadolinium in the Context of an ASL Examination

When ASL is performed within a standard brain MRI that also includes post-contrast T1 sequences, the ASL module must be acquired before gadolinium administration. If gadolinium is administered before ASL:

• Gadolinium shortens blood T1, which accelerates label decay during the post-label delay → CBF underestimation
• The M0 calibration scan (acquired at long TR) will reflect the gadolinium-shortened blood T1 rather than the native blood T1 → systematic quantification error

Rule: in any brain MRI session that includes both ASL and post-contrast T1 sequences, the ASL module (including M0 calibration) must be positioned before the gadolinium injection in the sequence order.

6.3 Post-Contrast Timing

Not applicable to the ASL sequence itself. When ASL is combined with DSC perfusion in a single session (as complementary sequences for tumour evaluation), the sequence order is: ASL first (before gadolinium) → DSC bolus injection and tracking → post-contrast T1 sequences. This order ensures ASL data is acquired in native blood T1 conditions.

7. Reporting Essentials

7.1 Interpretation Framework

ASL CBF maps are interpreted on two levels: qualitative pattern recognition (which territory is hypoperfused? which is hyperperfused?) and quantitative assessment (what are the absolute CBF values, and how do they compare to normative values or prior examinations?).

Normal CBF reference values (3T, single-PLD PCASL, adult): grey matter 40–80 mL/100 g/min (mean approximately 55–60 mL/100 g/min); white matter 15–30 mL/100 g/min (mean approximately 20–25 mL/100 g/min); global cortical CBF declines approximately 0.5–1% per year of normal ageing.

Diagnostic pattern axes:

Hypoperfusion (↓ CBF): territorial (vascular territory) → ischaemic stroke, carotid stenosis, occlusive disease; posterior parieto-occipital / temporoparietal → Alzheimer's disease, posterior cortical atrophy; frontal predominant (bilateral) → FTD; occipital with relative posterior cingulate sparing (cingulate island sign) → DLB; focal peritumoral → tumour mass effect, treatment effect after radiation/bevacizumab; global → consider labeling plane malposition first, then true global hypoperfusion (cardiac low output, severe bilateral ICA stenosis).

Hyperperfusion (↑ CBF): ictal/peri-ictal → epileptic focus (during or immediately after seizure); intratumoral (high rCBF relative to contralateral) → high-grade glioma, meningioma; serpentine/gyriform → arterial transit artefact (collateral or delayed flow), not true hyperperfusion; luxury perfusion (matched with DWI-positive stroke territory) → post-ischaemic reperfusion.

The cingulate island sign in DLB: relative preservation of posterior cingulate CBF (appearing as a bright "island" on the CBF map) surrounded by occipital hypoperfusion — a pattern specific to DLB that is the opposite of the posterior cingulate hypoperfusion in AD. This pattern is detectable on ASL CBF maps and provides a non-invasive alternative to FP-CIT SPECT for DLB characterisation.

7.2 Mandatory Reporting Checklist

Technique documentation: field strength (3T / 1.5T); labeling scheme (PCASL / PASL / other); readout (3D GRASE / 3D stack-of-spirals / 2D EPI — document if non-consensus readout); PLD in ms (document if non-standard for age/condition); background suppression (on/off); number of control-label pairs acquired; M0 calibration included or not (if absent: CBF map is qualitative only); labeling plane level (C2/C3 standard, or other — document).

CBF map quality: global CBF technically adequate (not globally suppressed from labeling plane error); motion artefact absent/mild/severe (region affected); arterial transit artefact absent/present (location and clinical implication); 2D EPI distortion not applicable (3D readout) or present (document).

Regional CBF assessment: global cortical CBF normal/globally reduced/globally increased; regional asymmetry present/absent (if present, which territory, quantify ratio); specific patterns (territorial / posterior predominant / frontal predominant / occipital / cingulate island sign); tumour CBF if applicable (absolute CBF mL/100 g/min, comparison with contralateral grey matter).

Comparison with prior studies: if serial examination, document CBF values at same ROIs and change over time.

7.3 Structured Reporting Template

Technique: 3T head coil; PCASL labeling (C2/C3, perpendicular to ICA/VA); PLD [ms]; 3D [GRASE/spiral] readout; background suppression ON; [N] control-label pairs; M0 calibration included; absolute CBF map generated (mL/100 g/min).

Findings: Global cortical CBF [normal/reduced/increased; quantify if possible]. [Regional findings: describe territories, laterality, pattern]. [Arterial transit artefact: if present, describe location and clinical implication]. [Correlation with structural sequences if relevant].

Impression: [Quantitative CBF finding; perfusion pattern; clinical correlation].

Limitations: [Labeling plane quality; motion; PLD limitation if single-PLD used in patient with suspected delayed transit; 1.5T if not 3T].

7.4 Incidental Findings — Clinical Decision Framework

Usually benign: mild global CBF reduction in the context of known ageing (normal); minor asymmetry < 10% in the absence of clinical symptoms.

May require further investigation: asymmetric territorial CBF reduction in the absence of a structural lesion (may reflect haemodynamic impairment without infarction; recommend MRA for vascular evaluation); subtle posterior cingulate CBF reduction in a patient evaluated for cognitive decline (may represent early AD; recommend clinical correlation and neurocognitive testing).

Clinical communication: perfusion-diffusion mismatch (DWI-positive region with surrounding ASL hypoperfusion but no ischaemic T2 change) in an acute or subacute context → discuss immediately with the referring neurologist for acute stroke pathway assessment.

8. MRI Technologist Pearls

8.1 Sequence Order Logic

The mandatory rule is: ASL before gadolinium. The sequence order in any brain MRI session that includes ASL: (1) structural sequences (T2, FLAIR, DWI, T1 pre-contrast); (2) ASL (PCASL + M0) ← acquired before any gadolinium; (3) gadolinium injection (if required for the clinical indication); (4) post-contrast T1 sequences.

8.2 Positioning Tricks

Always review the sagittal localiser for the ICA trajectory before placing the labeling plane: tortuous ICAs at the C2/C3 level will produce inefficient labeling if the standard C2/C3 perpendicular placement is used without adjustment. If the ICA is tortuous at C2/C3, find the level where the vessel runs most vertically (usually 1–2 cm superior or inferior to C2/C3) and place the labeling plane there instead.

Verify the labeling plane gap on the coronal localiser: confirm that the superior edge of the labeling slab is at least 20 mm inferior to the inferior edge of the imaging volume. If the gap is < 10 mm, the labeling RF pulse may partially saturate brain tissue at the inferior boundary of the imaging volume.

Check the CBF map before releasing the patient: a globally dark CBF map (suggesting global hypoperfusion) is almost always a technical error — most commonly a malpositioned labeling plane. If this pattern is seen, repeat the ASL acquisition with verified labeling plane placement before releasing the patient.

8.3 Fast Salvage Protocol

The minimum viable ASL dataset is the PCASL acquisition + M0. Without M0, the CBF map is qualitative only (relative CBF values; no mL/100 g/min quantification). This is still clinically useful for pattern recognition.

8.4 Common Avoidable Errors

9. Quality Control Checklist

• Sequence order verified: ASL before gadolinium injection
• Labeling plane placement confirmed on sagittal localiser: C2/C3 level, perpendicular to ICA/VA
• Gap between labeling plane and imaging volume: ≥ 20 mm
• Labeling plane angle: perpendicular (< 10° angulation from perpendicular) to ICA/VA on sagittal
• Background suppression: ON
• PLD value documented: ___ ms (consistent with age/condition: ≥ 1800 ms for < 70 y; ≥ 2000 ms for ≥ 70 y or suspected delayed transit)
• Number of control-label pairs completed: ___
• M0 calibration acquired and archived
• Global CBF map quality: technically adequate (not globally suppressed)
• Motion artefact assessed: absent / mild / severe (document if severe)
• Arterial transit artefact assessed: absent / present (document location and clinical context)
• CBF map (mL/100 g/min) sent to PACS and archived
• Co-registration with structural T1: adequate / suboptimal (document)
• Field strength documented in report: 3T / 1.5T
• Labeling plane level and PLD documented in report

10. Advanced Technical Parameters

10.1 PCASL Labeling — Physics and Optimisation

Tissue Contrast Logic

PCASL achieves flow-driven adiabatic inversion using a train of discrete RF pulses with gradient blips applied in the same direction as blood flow. The combination of RF pulses and gradients creates a rotating effective magnetic field that inverts flowing blood spins through adiabatic passage — maintaining inversion of spins moving through the labeling plane while leaving static tissue largely unaffected (because static tissue spins do not experience the adiabatic condition).

The labeling efficiency α is the fraction of blood water spins successfully inverted. For PCASL in optimal conditions: α ≈ 0.85–0.95. Factors that reduce α: RF/gradient timing not optimised for the local blood velocity; cardiac arrhythmia (variable blood velocity); tortuous arteries at labeling level (non-perpendicular flow); high field (slight B1+ variation).

The control condition of PCASL applies the same RF pulses but with a phase cycling scheme that prevents the adiabatic condition — so the same amount of RF energy is deposited (preventing MT asymmetry) but without inversion of blood spins.

Key Parameters

Quantification Model

Standard single-compartment simplified kinetic model (Buxton et al., 1998, as simplified in [1]):

CBF = [6000 × λ × ΔM] / [2α × M₀ × T1_blood × (e^(−PLD/T1_blood) − e^(−(PLD+τ)/T1_blood))]

where: ΔM = control–label difference signal; M₀ = equilibrium blood magnetisation (from M0 scan); λ = blood-tissue partition coefficient (assumed 0.9 mL/g); α = labeling efficiency (typically assumed 0.85 for PCASL); T1_blood = 1650 ms (3T) or 1200 ms (1.5T) — major field-strength difference in quantification; τ = labeling duration; PLD = post-label delay.

The factor 6000 converts units from mL/g/s to mL/100 g/min.

This formula shows why accurate T1_blood is essential: a 10% error in assumed T1_blood propagates directly to a 10% error in absolute CBF. The consensus paper provides recommended values by field strength [1].

10.2 3D GRASE Readout — Technical Specifics

GRASE (Gradient And Spin Echo) combines the efficiency of EPI-like gradient echo readout with spin echo refocusing pulses inserted into the echo train. The result: spin echoes reduce T2* sensitivity (less geometric distortion than EPI); gradient echoes between spin echoes maintain efficiency; 3D acquisition with short k-space segments provides whole-brain coverage in a single readout per label/control pair.

Typical 3D GRASE ASL parameters at 3T:

10.3 Vendor Implementation

Important note: the historical default readouts listed above reflect platform generations current in 2024. Siemens and Philips have both moved toward GRASE-based 3D readouts as their clinical default; some Philips platforms still ship with 2D EPI as default. Verify the actual readout type on the scanner console before protocol specification and note it in reports [2].

11. Evidence Gaps and Ongoing Debate

Single-PLD vs multi-PLD for routine clinical use: the 2015 consensus [1] recommended single-PLD as the clinical standard for simplicity, accepting transit-time bias as a known limitation. The 2023 update [2] recommends considering multi-PLD in specific scenarios (elderly, steno-occlusive disease, paediatric) but does not mandate it universally. Whether multi-PLD or time-encoded ASL should become the clinical routine — given their superior correction of ATT bias and their increasing commercial availability — is an area of active debate.

Normalisation strategy for dementia: direct comparison of absolute CBF values between patients requires matched acquisition protocols, field strengths, and M0 calibration. In practice, many centres use ratio-normalised CBF (ROI CBF / global grey matter CBF or cerebellar CBF) as a more reproducible metric than absolute CBF. No consensus exists on the optimal normalisation strategy for clinical dementia reporting.

AI-based ASL reconstruction and noise reduction: deep learning denoising applied to ASL difference images has shown promising SNR improvement in research settings, potentially enabling fewer control-label pairs (shorter scan time) or lower-field ASL with maintained CBF map quality. No clinically validated, regulatory-cleared AI ASL reconstruction tool exists at the time of writing.

7T ASL: the higher intrinsic SNR at 7T and the longer blood T1 (≈ 2200 ms at 7T) provide further advantages for ASL. However, B1+ inhomogeneity at 7T complicates the PCASL labeling efficiency (which depends on precise B1+ delivery at the labeling plane). Parallel transmit (pTX) is required for reliable 7T PCASL. Clinical 7T ASL protocols are not yet standardised.

Standardised CBF normative databases: absolute CBF quantification is clinically useful only when compared to validated normative values for age and sex. No single multicentre normative ASL database with standardised acquisition protocol has been widely adopted as a clinical reference. This limits the clinical use of absolute CBF thresholds for individual patient diagnosis.

12. Evidence-Based References

A. Guidelines / Consensus / Society Recommendations

B. Systematic Reviews / Meta-analyses

C. Important Prospective / Original Studies

D. Technical MRI Papers

E. Landmark Historical References

End of document — Arterial Spin Labeling (ASL) Generic Standard Protocol — MRIninja v1.0 — August 2026