Master the acute stroke MRI protocol: DWI-ADC mismatch, minimized TE, susceptibility fixes, and the golden-hour scanning workflow radiographers need.

Acute Stroke MRI Protocol: 10 Critical Steps for the Golden Hour

Introduction

The acute stroke MRI protocol is, without meaningful competition, the single most time-critical sequence set in the entire cross-sectional imaging department, and arguably in the whole of diagnostic radiology. Unlike almost every other examination in this 30-day series, the acute stroke MRI protocol is not judged by image beauty — it is judged by minutes.

A patient arriving inside the thrombolysis window has a diagnostic and therapeutic clock running from the moment of last known well, and every additional minute of table time is measurable in neurons. The often-cited estimate from stroke neurology literature holds that approximately 1.9 million neurons are lost for every minute that a large-vessel occlusion remains untreated^[1]. That single statistic should frame every operational decision a radiographer makes when the “code stroke” page arrives.

Modern comprehensive stroke centres have largely converged on a hybrid model: non-contrast CT and CT angiography remain the first-line triage tool in most emergency departments because of speed and universal availability, while MRI is reserved for wake-up stroke, uncertain onset time, posterior circulation symptoms, or as the definitive problem-solving study when CT is equivocal.

Diffusion-weighted imaging (DWI) is substantially more sensitive than non-contrast CT for hyperacute ischaemia, detecting restricted diffusion within minutes of vessel occlusion, compared with the 31–60% sensitivity ceiling of non-contrast CT in the first six hours^[2]. This is precisely why MRI, despite its comparatively longer acquisition time relative to non-contrast CT, remains the internationally recognised reference standard imaging modality for confirming or definitively excluding acute cerebral infarction in equivocal or unclear presentations.

This article works through the complete acute stroke MRI protocol from a technologist’s-eye view and a radiologist’s-eye view simultaneously: the anatomy that must be understood before the first sequence is prescribed, the tissue relaxation values that explain why DWI-ADC mismatch is diagnostic, the ten-step scanning technique itself, the DSC perfusion contrast dynamics, SAR management at 3T, the top ten pathologies that mimic or complicate acute stroke, and — critically — the pitfall framework that separates a scanning error from an interpretation error from a downstream clinical error. Radiographers, radiologists, and non-radiology physicians each fail this protocol in structurally different ways, and understanding all three failure modes is what makes a stroke imaging pathway resilient rather than merely fast.

Globally, stroke remains one of the leading causes of long-term adult disability, and ischaemic stroke accounts for the large majority of cases, with haemorrhagic stroke and other causes making up the remainder^[10]. The acute stroke MRI protocol therefore sits at the intersection of two clinical imperatives that can pull in opposite directions: the imperative to move fast enough to preserve thrombolysis and thrombectomy eligibility, and the imperative to be thorough enough to correctly identify the substantial minority of “code stroke” presentations that turn out to be stroke mimics.

These mimics include seizure with post-ictal (Todd’s) paralysis, complicated migraine, functional neurological disorder, hypoglycaemia, or focal encephalitis^[11]. A well-designed acute stroke MRI protocol resolves this tension by front-loading the highest-yield, fastest sequences first, so that a confident answer — positive or negative — exists as early as possible in the examination.

The workflow implications of this extend well beyond the scanner room. Door-to-needle and door-to-puncture time targets set by organisations such as the American Heart Association / American Stroke Association (AHA/ASA) and the European Stroke Organisation (ESO) explicitly include imaging acquisition and interpretation time as a component of the total treatment delay^[12]. In practice, this means the protocol is not simply a radiology deliverable; it is a shared institutional metric tracked by emergency medicine, neurology, radiology, and hospital administration alike, and technologist-level protocol discipline has a directly measurable effect on system-wide quality benchmarks.

Anatomy

A safe and diagnostically complete acute stroke MRI protocol depends on the technologist’s working knowledge of cerebral vascular territories, because slice prescription, coil positioning, and even the decision to extend the protocol with perfusion imaging are all territory-dependent decisions made in real time at the console.

Gross vascular anatomy of the brain

The brain is supplied by two paired arterial systems that anastomose at the Circle of Willis: the anterior circulation (internal carotid arteries feeding the anterior cerebral artery [ACA] and middle cerebral artery [MCA]) and the posterior circulation (vertebral arteries joining to form the basilar artery, which supplies the brainstem, cerebellum, and posterior cerebral arteries [PCA]). Roughly 70–80% of acute ischaemic strokes occur in the anterior circulation, with the MCA territory being by far the most frequently affected single vascular bed^[3]. This statistical weighting is why MCA-territory changes — insular ribbon effacement, lentiform obscuration, loss of the grey-white differentiation — dominate stroke pattern recognition training, even though the protocol itself must remain territory-agnostic in its coverage.

Middle cerebral artery territory

The MCA arises from the internal carotid artery bifurcation and is conventionally divided into four segments: M1 (horizontal, sphenoidal segment, the classic large-vessel occlusion site targeted by mechanical thrombectomy), M2 (insular segment), M3 (opercular segment), and M4 (cortical branches). The lenticulostriate perforators arising from the M1 segment supply the basal ganglia and internal capsule — a territory with minimal collateral reserve, which is why lentiform nucleus infarction can appear on DWI within extremely short occlusion times.

Anterior cerebral artery territory

The ACA supplies the medial frontal and parietal lobes, including the primary motor and sensory cortex representing the lower limb. Isolated ACA infarcts are less common than MCA infarcts but carry a distinctive clinical signature — contralateral leg weakness with relative arm sparing — that should prompt the technologist to ensure full coverage of the interhemispheric fissure on the FLAIR and DWI slice stack.

Posterior circulation and vertebrobasilar system

The vertebrobasilar system supplies the brainstem, cerebellum, thalami, and occipital lobes via the PCA. Posterior circulation strokes are disproportionately under-detected by non-contrast CT because of beam-hardening artefact from the dense petrous and occipital bone, which is precisely why any patient with vertigo, diplopia, dysarthria, ataxia, or crossed sensorimotor signs should be escalated directly to the acute stroke MRI protocol rather than relying on CT alone^[4]. The basilar artery itself is a critical structure to interrogate on the TOF MRA source images — basilar occlusion carries a mortality approaching 80–90% if unrecognised and untreated.

Watershed and border-zone territories

The border zones between the ACA-MCA and MCA-PCA territories are watershed regions supplied by the most distal, lowest-pressure collateral vessels. These areas are exquisitely vulnerable to global hypoperfusion (cardiac arrest, severe hypotension, high-grade carotid stenosis) rather than focal embolic occlusion, producing a characteristic “string of pearls” pattern of cortical and subcortical DWI restriction that the radiographer should recognise as a distinct pattern from the wedge-shaped territorial infarct.

The Circle of Willis and collateral physiology

The Circle of Willis, formed by the anterior communicating artery and posterior communicating arteries bridging the anterior and posterior circulations, is the anatomical substrate for collateral flow. A complete, well-formed circle can substantially delay infarct core growth after large-vessel occlusion by recruiting leptomeningeal and communicating-artery collateral flow, which is the physiological basis for the extended thrombectomy windows (up to 24 hours) used in perfusion-selected patients under the DAWN and DEFUSE-3 trial criteria^[5,6]. The 3D TOF MRA sequence in this protocol directly images this anatomy and is often the deciding factor in whether a patient is escalated for interventional consultation.

Cerebellar arterial supply

The cerebellum receives blood from three paired vessels: the superior cerebellar artery (SCA), anterior inferior cerebellar artery (AICA), and posterior inferior cerebellar artery (PICA), all branches of the vertebrobasilar system. PICA territory infarction is the single most common cerebellar stroke and can present with isolated vertigo and gait ataxia that is easily misattributed to peripheral vestibular disease in the emergency department — a well-documented and clinically important stroke mimic in reverse, where the true stroke is mistaken for something benign rather than the other way around. Cerebellar infarcts carry a specific risk of mass effect on the fourth ventricle and obstructive hydrocephalus, so the radiographer should flag any cerebellar DWI restriction for expedited radiologist review regardless of infarct size.

Basal ganglia and thalamic perforator territory

The lenticulostriate arteries (from M1), the anterior choroidal artery (from the internal carotid artery), and the thalamoperforating and thalamogeniculate arteries (from the P1 posterior cerebral artery segment) supply the deep grey matter structures — caudate, putamen, globus pallidus, internal capsule, and thalamus. These are end-arteries with essentially no collateral supply, which explains why deep perforator infarcts evolve rapidly and are frequently the earliest visible DWI change in a hyperacute large-vessel occlusion, often preceding visible cortical change by many minutes.

Venous anatomy and the dural sinuses

Although the acute stroke MRI protocol is arterially focused, the technologist should retain working knowledge of the superior sagittal sinus, straight sinus, transverse sinuses, and sigmoid sinuses, since cerebral venous sinus thrombosis is an important stroke mimic covered later in this article. The SWI/GRE sequence acquired as part of this protocol is directly sensitive to venous thrombus via its paramagnetic susceptibility effect, meaning this standard stroke imaging protocol already carries meaningful diagnostic yield for venous pathology without any additional sequence time.

MR tissue relaxation values

Understanding the T1 and T2 relaxation behaviour of normal brain parenchyma versus ischaemic and haemorrhagic tissue is what allows a technologist to recognise a technically successful acute stroke MRI protocol at the console before the radiologist ever opens the study. The table below summarises approximate relaxation times at 1.5T and 3T for the tissues most relevant to stroke imaging, alongside their characteristic DWI and ADC behaviour.

The physiological basis for this signature is cytotoxic oedema. Within seconds of vessel occlusion, ATP-dependent sodium-potassium pump failure allows sodium and water to shift from the relatively unrestricted extracellular space into the intracellular compartment, where the diffusion of water molecules is markedly more restricted by cell membranes, organelles, and cytoskeletal structures. This intracellular water shift lowers the apparent diffusion coefficient — the quantity the ADC map directly encodes — well before any T1 or T2 relaxation change becomes visible on conventional sequences. This is precisely why DWI can detect infarction within minutes, while T2/FLAIR changes typically require 6–8 hours to become reliably visible, and non-contrast CT changes often take even longer.

As the infarct evolves over the following days, vasogenic oedema is superimposed on the initial cytotoxic process, extracellular water content rises, and both T1 and T2 relaxation times progressively lengthen — producing the classic transition from an isointense-to-CT, DWI-bright, ADC-dark hyperacute infarct into a T1-hypointense, T2/FLAIR-hyperintense subacute infarct over one to three weeks. Recognising where a given infarct sits on this relaxation timeline is often the deciding factor in whether a lesion represents the acute event responsible for the patient’s current presentation or an incidental finding from a prior, clinically unrelated episode.

Scanning technique

The acute stroke MRI protocol is built around speed without sacrificing the diagnostic core. The following ten-step sequence reflects the workflow used at high-volume comprehensive stroke centres, where net table time for the non-contrast core protocol is routinely held under six minutes.

1. Patient screening and coil setup. Confirm MRI safety screening is complete or, in a true code-stroke pathway, use the abbreviated emergency metal-screening checklist. Position the patient supine in a dedicated head coil with foam padding for immobilisation — motion is the single greatest threat to diagnostic yield in an often confused or hemiparetic patient.
2. Localiser / scout. Acquire a rapid three-plane localiser. Confirm midline symmetry and check for gross asymmetry or midline shift that may reprioritise slice angulation.
3. Axial DWI (b=0 and b=1000 s/mm²). This is the priority sequence and should be acquired first, before FLAIR, so that a diagnostic answer exists even if the patient cannot tolerate the remainder of the examination. Use minimized TE and echo-planar readout with fat suppression.
4. Auto-generated ADC map. Confirm the scanner’s inline ADC map reconstructs correctly and correlates spatially with any DWI hyperintensity before the patient leaves the bore.
5. Axial FLAIR. Acquired second, FLAIR provides the DWI-FLAIR mismatch assessment central to wake-up stroke triage — a lesion visible on DWI but not yet visible on FLAIR suggests onset within approximately 4.5 hours, supporting thrombolysis eligibility even with unknown onset time^[7].
6. Axial SWI or T2* GRE. Detects haemorrhagic transformation, microbleeds, and the susceptibility vessel sign — a hypointense clot signature that can localise the occluding thrombus directly.
7. 3D TOF MRA of the Circle of Willis. Non-contrast angiography confirms or excludes large-vessel occlusion and directly informs thrombectomy candidacy; extend coverage to the aortic arch origin of the great vessels where local protocol allows.
8. Optional single-shot DSC perfusion. Reserved for cases with uncertain onset time or extended-window thrombectomy evaluation; requires the gadolinium bolus described in the contrast section below.
9. Image review at console. Confirm DWI-ADC concordance, FLAIR correlation, absence of susceptibility artefact obscuring the territory of clinical concern, and TOF MRA vessel patency before releasing the patient.
10. Immediate transmission and verbal notification. Push images to PACS and verbally notify the stroke team without delay — in a comprehensive stroke centre, “door-to-needle” clocks continue to run during image review, and this step is timed as rigorously as the scan itself.

The specific sequencing of DWI before FLAIR, rather than the more conventional “localiser then anatomical then functional” order used in most other MRI protocols in this series, is a deliberate departure justified purely by clinical urgency. If a hyperacute stroke patient becomes too agitated, too hypoxic, or too haemodynamically unstable to complete the full examination, having already secured a diagnostic DWI-ADC pair means the referring team still receives an actionable answer. This principle — always acquire the single most decision-changing sequence first — is a genuinely useful heuristic that generalises well beyond stroke imaging alone to almost any time-critical or motion-limited MRI protocol encountered in daily practice.

Total table time is only part of the throughput equation; patient turnover time also depends heavily on how efficiently the preceding patient is cleared from the bore and how quickly the next code-stroke patient can be transferred, screened, and positioned. High-performing stroke imaging suites typically maintain a permanently pre-configured stroke coil setup and a visibly posted, laminated rapid-screening checklist at the scanner door specifically for this pathway, separate from the routine elective screening process used for scheduled outpatient MRI studies.

Patient preparation and communication in the hyperacute setting

Executing the acute stroke MRI protocol well is as much a communication exercise as a technical one. Many patients arriving under a code-stroke activation are hemiparetic, dysphasic, confused, or frankly uncooperative, and the technologist typically has only seconds to establish rapport, explain the examination in the simplest possible terms, and position the patient safely — all while a stroke team is often physically present at the scanner door awaiting results. Clear, calm, repeated verbal instruction (“stay very still, we are almost finished”) measurably reduces motion artefact even in patients with significant expressive aphasia, because comprehension of tone and simple cues frequently outlasts formal language function.

Family members or accompanying paramedics can be a valuable but under-used resource at this stage: they often know the exact last-known-well time, baseline cognitive status, and relevant implanted device history that the patient cannot currently communicate. A structured, rapid verbal handover — “any metal, any pacemaker, any recent surgery, what time did symptoms start” — delivered while the patient is being positioned rather than as a separate preceding step is one of the most effective time-saving habits in a high-volume stroke pathway, and it should be rehearsed as deliberately as any pulse sequence parameter.

Coordinating the code-stroke pathway across departments

The acute stroke MRI protocol functions inside a multidisciplinary pathway that typically includes emergency medicine triage, a stroke-alert paging system, the MRI technologist, the on-call or in-house radiologist, and an interventional or stroke neurology team. Each stage has its own internal time target, and the imaging stage sits in the middle of the chain — meaning delays introduced at the scanner propagate directly into every downstream decision. Comprehensive stroke centres that consistently meet national door-to-needle benchmarks typically share a common feature: a pre-agreed, protocolised imaging pathway with locked default sequence parameters, rather than an ad hoc, individually optimised protocol built fresh for every patient.

Contrast media protocol

In the acute stroke MRI protocol, gadolinium-based contrast is not required for the core diagnostic question of infarct detection — DWI, ADC, FLAIR, SWI, and TOF MRA are all non-contrast sequences. Contrast is reserved specifically for dynamic susceptibility contrast (DSC) perfusion imaging, used to characterise the ischaemic penumbra (salvageable tissue) versus the infarct core when extended-window thrombectomy is being considered or when onset time is unknown.

The high flow rate specified here — 5.0 mL/s, notably faster than the 1.5–3.0 mL/s used in most other neuro-MRI protocols in this series — reflects a fundamentally different physiological goal. Static post-contrast T1 imaging tolerates a slower, more diffuse bolus because it only needs contrast to reach steady-state distribution. DSC perfusion instead depends on a sharp, compact bolus producing a well-defined signal-drop curve as the contrast bolus first passes through the cerebral capillary bed — a shallow or prolonged bolus flattens the time-signal curve and degrades the accuracy of derived maps such as time-to-maximum (Tmax), cerebral blood flow (CBF), and cerebral blood volume (CBV).

Clinically, the perfusion maps derived from this DSC bolus are what allow a radiologist to distinguish the irreversibly infarcted core from the potentially salvageable penumbra — tissue that is functionally impaired but not yet structurally dead. In practice, the DWI lesion volume is typically used as a surrogate for the infarct core, while the volume of tissue showing delayed Tmax on perfusion imaging but preserved diffusion represents the penumbra. A large core-penumbra mismatch identifies patients who may still benefit from late-window thrombectomy, which is the entire clinical rationale for extending an otherwise non-contrast acute stroke MRI protocol with a rapid gadolinium bolus.

Because the DSC technique depends on first-pass susceptibility effects rather than steady-state enhancement, timing between injection and acquisition start is unforgiving. Most departments programme a fixed injector-to-scanner trigger delay validated against their specific power injector’s compliance and flow characteristics, rather than relying solely on operator judgement — this is one of the clearest examples in the entire 30-protocol series of where injector hardware precision directly determines diagnostic map quality rather than merely image cosmetics.

Specific absorption rate

Specific absorption rate (SAR) management in this protocol is a secondary but real consideration, particularly at 3T where multiple fast spin-echo and steady-state sequences (FLAIR, SWI acquisition trains) can approach regulatory SAR limits, especially in patients with a smaller body habitus or in paediatric stroke presentations.

Five practical strategies reduce SAR while preserving the diagnostic integrity of the acute stroke MRI protocol:

1. Increase TR on FLAIR sequences where table time permits, since SAR scales with the square of flip angle but is inversely related to TR — a modest TR increase yields a disproportionate SAR reduction.
2. Use parallel imaging acceleration, which reduces the number of RF refocusing pulses required per unit time, directly lowering deposited energy.
3. Select hyperecho or variable flip-angle refocusing trains on turbo spin-echo FLAIR, reducing peak RF power while maintaining T2/FLAIR contrast.
4. Confirm accurate patient weight entry at the console — SAR calculations are weight-normalised, and an inaccurate weight (common in an unresponsive or unmeasured stroke patient) can trigger unnecessary sequence throttling or, worse, underestimate true deposited energy.
5. Reserve first-level (normal) operating mode for the routine protocol; do not enter controlled operating mode for a non-research emergency stroke examination.

In practice, SAR rarely becomes a hard limiting factor for the core stroke imaging protocol at 1.5T, since DWI, SWI, and TOF MRA are all comparatively low-SAR sequences. The FLAIR sequence, with its long refocusing pulse train, is the main SAR driver, and at 3T in a smaller patient it can occasionally approach the permitted ceiling. When this occurs, the correct operational response is to accept a modest TR increase or echo-train adjustment rather than skip FLAIR altogether, since DWI-FLAIR mismatch remains one of the most clinically valuable determinations available for wake-up and unknown-onset stroke triage.

Paediatric and small-body-habitus considerations

Paediatric ischaemic stroke, while rare relative to adult presentations, carries a disproportionately high risk of both delayed diagnosis and SAR-related concern, since SAR calculations are weight-normalised and children present with a smaller absolute mass over which RF energy is deposited. Departments running a paediatric stroke pathway should maintain a separate, pre-validated low-SAR protocol variant with adjusted TR, reduced echo-train length, and confirmed weight-based dosing for any contrast-enhanced perfusion component, rather than manually re-deriving safe parameters under time pressure. The same susceptibility-mitigation principles — minimized TE, adequate bandwidth, appropriate parallel imaging — apply equally to the paediatric protocol and should not be relaxed simply because the patient is smaller.

Pregnancy and lactation considerations

MRI without gadolinium contrast is generally considered safe throughout pregnancy when clinically indicated, making the non-contrast core of the acute stroke MRI protocol — DWI, ADC, FLAIR, SWI, and TOF MRA — well suited to a pregnant stroke patient. Gadolinium-based DSC perfusion imaging should be reserved for cases where the additional diagnostic information is judged to materially change management, following institutional obstetric-radiology consultation pathways, since gadolinium crosses the placenta and its long-term fetal safety profile remains incompletely characterised.

Top 10 pathologies

The following ten conditions represent the differential diagnosis landscape a radiographer and radiologist must hold in mind while executing and interpreting this protocol — some are the target pathology itself, others are stroke mimics or complications that alter management. Reviewing this list before every shift is a genuinely useful clinical and operational discipline, because the protocol is deliberately built and engineered to remain sensitive across this entire broad differential diagnosis, not solely to the single most statistically common diagnosis of MCA territorial infarction.

Several patterns recur across this differential and are worth committing to memory as pattern-recognition shortcuts. A single wedge-shaped territorial lesion favours embolic or large-vessel occlusive disease. Bilateral, symmetric findings favour a global process — PRES, hypoxic-ischaemic injury, or toxic-metabolic derangement — rather than focal vascular occlusion. Lesions crossing arterial territory boundaries favour venous infarction. Multiple lesions of differing ages favour an embolic source requiring cardiac work-up, or a vasculitic/vasculopathic process requiring vessel-wall imaging. None of these heuristics replace formal radiologist interpretation, but they meaningfully speed pattern recognition at first pass, which matters when every minute of interpretation time is tracked against a national quality benchmark.

Imaging criteria supporting thrombectomy eligibility

Mechanical thrombectomy has transformed outcomes for large-vessel occlusion stroke, and the acute stroke MRI protocol described in this article generates every imaging component needed to support an eligibility decision. Three imaging elements are typically required: confirmation of a proximal large-vessel occlusion on TOF MRA (or CT angiography where MRI is unavailable), an infarct-core volume estimate derived from DWI that remains below the threshold generally associated with favourable outcome after reperfusion, and — for patients presenting beyond roughly six hours from last known well — a favourable core-penumbra mismatch on DSC perfusion imaging supporting the presence of salvageable tissue.

The DAWN and DEFUSE-3 trials established the evidence base for extending thrombectomy eligibility out to 24 hours in carefully selected, perfusion-imaged patients, fundamentally changing the clinical calculus for wake-up stroke and unwitnessed-onset presentations^[5,6]. This is precisely why this stroke imaging protocol retains an optional DSC perfusion extension rather than being a purely binary “infarct present or absent” examination — for a meaningful subset of patients, the perfusion data is what determines whether an intervention proceeds at all.

Pitfalls — radiographers

The primary scanning pitfall of the acute stroke MRI protocol is magnetic susceptibility distortion, particularly at the skull base, paranasal sinuses, and petrous temporal bones, where air-bone-tissue interfaces produce local field inhomogeneity that distorts the echo-planar DWI readout. Left uncorrected, this artefact can obscure or falsely simulate restricted diffusion in the inferior frontal and temporal lobes — precisely the territories most relevant to MCA and PCA stroke.

Of these, magnetic susceptibility distortion deserves particular emphasis because it is both the most common and the most preventable scanning-level failure in the acute stroke MRI protocol. The three-part mitigation — increased receiver bandwidth, parallel imaging, and minimized TE — should be built into the vendor protocol card as locked or default parameters rather than left to individual technologist discretion at 3 a.m. during a code-stroke activation, when cognitive bandwidth for manual parameter optimisation is at its lowest. Departments that audit their stroke DWI protocols periodically against these three parameters consistently find fewer non-diagnostic or equivocal studies requiring same-day repeat scanning.

Pitfalls — radiologists

The primary interpretation pitfall in acute stroke MRI reporting is misreading T2 shine-through as true restricted diffusion, or conversely, dismissing genuine restricted diffusion in the presence of susceptibility-related signal dropout as artefact rather than pathology.

The T2 shine-through pitfall is worth dwelling on because it is, in the authors’ clinical experience and in the published radiology education literature, the single most frequently cited diagnostic trap in acute stroke MRI reporting^[13]. The reflex habit of visually scanning only the DWI trace image — the brightest, most eye-catching sequence in the protocol — without deliberately cross-referencing the ADC map on every positive finding is a workflow shortcut that becomes dangerous precisely because it works correctly most of the time, making the exceptions easy to miss under time pressure.

Pitfalls — non-radiology physicians

Emergency physicians, neurologists, and hospitalists interpreting or acting on preliminary stroke MRI findings without full radiological context are exposed to a distinct set of clinical pitfalls.

Non-radiology physicians occupy a uniquely difficult position in the stroke pathway: they are frequently the first clinician to see the patient and the last clinician to see the final radiology report, often having to make an initial thrombolysis decision from a preliminary read or a verbal communication before the formal written interpretation is finalised. Structured, closed-loop verbal communication — where the receiving physician repeats back the key finding and its stated confidence level — meaningfully reduces the risk of the misunderstandings tabulated above, particularly under the time pressure and cognitive load of an active hyperacute stroke activation.

Pitfall comparison summary

This three-tier framework — scanning, interpretation, and clinical action — is deliberately structured so that each professional group can immediately locate the failure modes relevant to their own role rather than wading through an undifferentiated list. In departmental quality-improvement reviews, mapping a near-miss or adverse event onto this framework quickly clarifies whether the corrective action belongs in a protocol card update, a reporting-template change, or a clinical communication policy — three very different interventions that are easy to conflate if the pitfall taxonomy is not made explicit from the outset.

AI and automation

Automated large-vessel-occlusion detection and perfusion-mismatch software have become embedded infrastructure in comprehensive stroke centres rather than experimental add-ons. FDA-cleared and CE-marked platforms such as RapidAI (iSchemaView), Viz.ai, and Brainomix e-Stroke automatically process DWI, TOF MRA, and DSC perfusion data to flag suspected large-vessel occlusions and generate core/penumbra volume estimates within minutes of image acquisition, pushing direct mobile alerts to on-call stroke neurologists and interventionalists^[8].

Independent validation studies report that these platforms can reduce the time from imaging completion to treatment-team notification by a clinically meaningful margin, though real-world performance is protocol-dependent and hardware-dependent — algorithms trained predominantly on optimally acquired DWI data perform less reliably on studies degraded by the same susceptibility distortion and motion artefacts discussed above^[9]. This creates a direct feedback loop between scanning technique quality and AI triage reliability: a technically compromised stroke imaging study degrades not only radiologist interpretation but also every downstream automated tool built on top of it.

Beyond large-vessel-occlusion flagging, automated ASPECTS (Alberta Stroke Program Early CT Score) scoring adapted for DWI, automated infarct-volume segmentation, and mismatch-ratio calculators are increasingly embedded directly into PACS viewers rather than requiring a separate standalone workstation. This integration trend reduces the friction that historically limited AI adoption in time-critical pathways, since the stroke team no longer needs to open a second application or wait for a separate server round-trip to see automated measurements alongside the primary diagnostic images.

The 2026 European Congress of Radiology placed particular emphasis on the shift from AI as an isolated pilot project toward AI as embedded, everyday infrastructure across stroke, oncology, and cardiac imaging pathways alike^[14]. For the acute stroke MRI protocol specifically, this means the technical quality benchmarks established earlier in this article — minimized TE, adequate bandwidth, appropriate parallel imaging factor — are no longer simply “good radiographic practice.” They are now, in effect, the input-quality specification that every downstream AI triage tool implicitly depends on to perform at its validated accuracy.

Telestroke and interfacility transfer imaging

Many patients undergo their acute stroke MRI protocol at a smaller spoke hospital before transfer to a comprehensive stroke centre for thrombectomy, making rapid, reliable image transmission an essential extension of the scanning protocol itself. Cloud-based PACS and AI triage platforms increasingly automate this hand-off, pushing both the raw DICOM dataset and the automated large-vessel-occlusion flag directly to the receiving hospital’s interventional team simultaneously with the sending radiologist’s preliminary read, compressing what was historically a sequential, phone-call-dependent transfer workflow into a largely parallel one.

This telestroke integration places additional weight on protocol consistency across a hospital network: if the spoke site’s stroke imaging protocol uses materially different sequence parameters, slice coverage, or contrast timing from the hub site’s validated reference protocol, the receiving team may lose confidence in directly comparing pre- and post-transfer imaging, or the automated AI triage tool may perform outside its validated accuracy envelope on an unfamiliar acquisition. Standardising this protocol across every site in a referral network — down to the specific TE, bandwidth, and parallel imaging factor discussed throughout this article — is therefore not merely a single-site quality initiative but a network-level patient safety requirement.

Further reading

The five resources below extend directly on themes covered in this article — CT-based stroke triage as the frequent first-line comparator to MRI, contrast media safety underpinning the DSC perfusion bolus, general MRI sequence optimisation principles, and the AI infrastructure trends now embedded in modern stroke pathways.

1. Critical Non-Contrast Brain CT Parameters Every Radiographer Must Master — the CT-side companion to this protocol, including ASPECTS scoring and stroke-window triage.
2. 7 Proven Strategies for Optimizing MRI Sequences in 2026 — how injection precision and sequence parameter choices interact across advanced MRI protocols.
3. 2026 Worldwide Guidelines for Safe Contrast Media Administration — eGFR thresholds and safe gadolinium administration relevant to DSC perfusion bolus planning.
4. ECR 2026 Review: Major Updates, Keynote Lectures Summary & AI Highlights — includes the ECR 2026 stroke-AI keynote coverage referenced in this article.
5. Scaling Radiology AI 2026: Moving from Pilot Projects to Core Infrastructure — the broader infrastructure context behind automated large-vessel-occlusion detection tools.

Reducing artefacts with patients and parameters

Beyond the stroke-specific susceptibility mitigation covered above, four universal parameter categories govern image quality on every sequence in this protocol. Understanding their trade-offs allows a technologist to make defensible, real-time adjustments under time pressure.

Spatial resolution

Spatial resolution defines the ability to distinguish small details in an image. Matrix size: increasing the matrix size (frequency × phase) increases spatial resolution, but decreases SNR because the voxel size becomes smaller. Field of view (FOV): reducing the FOV increases spatial resolution; however, a smaller FOV results in smaller voxels and reduces SNR. Slice thickness: thinner slices provide higher spatial resolution and reduce partial volume averaging, but significantly decrease SNR — a critical trade-off on thin-slice DWI used to detect small lacunar infarcts.

Signal-to-noise ratio (SNR)

SNR represents the strength of the diagnostic signal relative to inherent background noise. A high SNR produces crisp, clear images, whereas a low SNR looks grainy. Number of averages (NEX/NSA): increasing averages acquires data multiple times, improving SNR, though doubling the averages roughly doubles scan time — rarely acceptable in the hyperacute stroke window. Receiver bandwidth: decreasing bandwidth limits recorded noise and boosts SNR, but lowers bandwidth increases scan time and chemical shift artefact — the opposite trade-off from what the susceptibility-mitigation strategy above requires, which is why bandwidth selection on DWI is always a deliberate compromise. Coil selection: using a dedicated, localised head coil rather than a body coil captures a much stronger signal and heavily improves SNR.

Image contrast

Contrast determines how different tissues are distinguished from one another. Repetition time (TR): TR is the time between consecutive RF pulses; a short TR maximises T1 tissue contrast, while a long TR minimises it — FLAIR relies on a long TR/TI combination to null CSF signal. Echo time (TE): TE is the time between the RF pulse and the peak of the echo signal; a short TE minimises T2 effects, while a long TE maximises T2 weighting, making fluid-filled areas appear very bright — and, as established above, a short TE is also the primary defence against susceptibility distortion on DWI. Flip angle: controls proton excitation and directly shapes tissue contrast, particularly critical on gradient-echo SWI sequences.

Artefact control

Phase-encoding direction: swapping the phase and frequency axes can shift motion-induced artefacts, such as CSF pulsation from posterior fossa vessels, away from the primary region of interest. Flow compensation / gating: uses physiological triggers to minimise blurring and ghosting caused by pulsatile motion, though cardiac gating is rarely used on brain stroke protocols given time constraints. Parallel imaging: utilises multiple coil elements simultaneously to reduce phase-encoding steps, significantly cutting scan time and reducing both motion artefact and — specifically relevant to this protocol — magnetic susceptibility distortion on echo-planar DWI.

Parallel imaging protocols and parameters

Parallel imaging acceleration (SENSE, GRAPPA, ASSET, or manufacturer-equivalent) is the single most effective lever a technologist can pull to simultaneously reduce scan time, motion sensitivity, and susceptibility distortion in the acute stroke MRI protocol. The turbo/echo-train factor and acceleration factor must be balanced against SNR loss, which scales with the square root of the acceleration factor.

Standardised reporting and documentation

A structured reporting template reduces the risk of the interpretation pitfalls described earlier in this article by forcing explicit documentation of every element the referring stroke team needs, rather than relying on free-text narrative that may unintentionally omit a critical negative finding. A well-designed stroke MRI report should explicitly state: the presence or absence of restricted diffusion with anatomical territory and estimated volume; DWI-ADC concordance status; presence or absence of susceptibility blooming suggesting haemorrhage; large-vessel patency on TOF MRA; and, where acquired, a qualitative or quantitative perfusion-mismatch statement.

Explicitly documenting a negative finding — “no restricted diffusion identified, TOF MRA patent, no susceptibility blooming” — is just as clinically important as documenting a positive one, since the absence of a clear negative statement can be misread by a time-pressured clinician as an incomplete or pending report rather than a genuinely reassuring result. Many comprehensive stroke centres now embed a dedicated “code stroke MRI” macro or structured template directly into the reporting software specifically to standardise this language across all radiologists in the group, reducing inter-reader variability in how equivocal or borderline findings are communicated to the treating team.

Quality assurance and continuous protocol auditing

A single well-designed acute stroke MRI protocol is only as good as the ongoing discipline that keeps it consistently executed across every shift, every technologist, and every scanner in a multi-site network. Departments with the strongest stroke imaging quality metrics typically run a formal, recurring audit cycle rather than treating protocol design as a one-time setup task completed at scanner installation.

Technologist competency validation

Because the acute stroke MRI protocol is executed disproportionately during night and weekend shifts by whichever technologist is on call, formal competency validation should not be assumed simply because a staff member has completed general MRI credentialing. A focused, protocol-specific competency checklist — covering coil setup speed, correct sequence order, verification of minimized TE and adequate bandwidth at the console, and recognition of a non-diagnostic DWI requiring immediate repeat — is a low-cost, high-value addition to annual staff competency review, and directly targets the specific susceptibility-artefact failure mode identified as the primary scanning pitfall throughout this article.

Metrics worth tracking

Useful, actionable quality metrics for this protocol include: the proportion of studies requiring same-day repeat DWI due to susceptibility artefact or motion; mean image-acquisition-to-PACS-transmission time; mean time from image availability to formal radiologist report; the rate of discordance between preliminary AI large-vessel-occlusion flags and final radiologist interpretation; and door-to-needle or door-to-puncture time stratified by whether MRI or CT served as the triage modality. Tracking these metrics longitudinally, rather than only reviewing individual cases after an adverse event, surfaces systemic protocol weaknesses — a particular scanner consistently producing more non-diagnostic DWI, for example — well before they contribute to a serious clinical miss.

Closing the loop with case review

Regular multidisciplinary case review, bringing together radiographers, radiologists, and the treating stroke or emergency medicine team, is one of the most effective mechanisms for reinforcing the three-tier pitfall framework described earlier in this article. Reviewing a small number of representative cases each month — including at least one technically challenging susceptibility-artefact case and one interpretation near-miss — keeps the specific failure modes of this protocol current in institutional memory, rather than allowing lessons learned from a single adverse event to fade after the initial incident review is closed.

Working with scanner vendors and application specialists

Scanner manufacturers’ clinical application specialists are an under-used resource when refining this protocol, particularly for susceptibility-mitigation parameter tuning that is highly specific to a given gradient system, coil design, and software platform version. A periodic joint review — bringing the department’s lead stroke radiologist, senior MRI technologist, and the vendor’s application specialist together to test the protocol card against phantom and retrospective clinical data — can identify incremental gains in bandwidth headroom or parallel imaging capability that become available with each software upgrade, gains that are easy to miss if the protocol card is simply carried forward unchanged from one scanner generation to the next.

Follow-up and surveillance imaging

The acute stroke MRI protocol described in this article is designed for the hyperacute diagnostic window, but many patients return for follow-up imaging at 24–48 hours to confirm final infarct territory and exclude interval haemorrhagic transformation, particularly after thrombolysis or thrombectomy. This follow-up study can generally use a lighter-weight version of the same protocol — DWI, FLAIR, and SWI are typically sufficient, since TOF MRA and DSC perfusion add little incremental value once the acute intervention decision has already been made and vessel recanalisation status is usually already known from the angiographic procedure itself.

Longer-term follow-up, at 90 days or beyond, shifts the imaging question from acute triage toward secondary-prevention work-up: identifying an embolic source, characterising carotid or vertebral stenosis, and assessing for silent infarcts suggesting an underlying vasculopathy or an as-yet-undiagnosed cardioembolic condition. These studies fall outside the scope of this hyperacute protocol proper and are typically built as distinct, non-emergency protocols elsewhere in a department’s neuro-imaging repertoire, but recognising the boundary between the hyperacute protocol and these downstream studies helps avoid unnecessary sequence duplication and keeps scanner utilisation efficient across a busy stroke service.

Conclusion

The acute stroke MRI protocol succeeds or fails on details that are individually small but collectively decisive. A properly minimised TE, an appropriately increased receiver bandwidth, and a sensibly chosen parallel imaging factor are the difference between a diagnostic DWI-ADC pair and a study degraded by magnetic susceptibility distortion at exactly the moment a clinician most needs a confident answer. Anatomy knowledge — vascular territories, perforator supply, watershed zones, the Circle of Willis — is what allows a technologist to recognise, in real time, whether coverage and slice prescription are adequate for the clinical question being asked, rather than mechanically running a fixed protocol regardless of context.

The relaxation-value framework presented earlier in this article explains, at a fundamental physiological and biophysical level, why DWI-ADC concordance is the single most important interpretive checkpoint in stroke neuroimaging: true restricted diffusion is DWI-bright and ADC-dark, while T2 shine-through is DWI-bright and ADC-normal or bright. This one distinction underlies the majority of both false-positive and false-negative errors documented across the radiographer, radiologist, and non-radiology physician pitfall tables in this article, and reinforcing it at every level of the stroke pathway — from console to reporting workstation to bedside — is arguably the single highest-yield educational intervention available to a stroke imaging programme.

The top ten pathology differential presented here is a deliberate reminder that the acute stroke MRI protocol is not solely a tool for confirming large-vessel MCA occlusion. It must also reliably identify venous infarction, PRES, hypoxic-ischaemic injury, haemorrhagic transformation, arterial dissection, and vasculitis — conditions whose management diverges sharply from standard ischaemic stroke pathways, and whose misclassification carries real clinical consequences. The three-tier pitfall framework — scanning, interpretation, and clinical action — exists precisely because these errors cluster at different points in the pathway and demand different corrective strategies: protocol-card engineering for radiographers, structured ADC cross-referencing for radiologists, and closed-loop communication discipline for treating physicians.

Finally, the growing role of FDA-cleared and CE-marked AI triage software does not remove the burden of technical excellence from the imaging team — it raises it. Automated large-vessel-occlusion detection, ASPECTS-DWI scoring, and perfusion-mismatch calculation all inherit the technical quality of the underlying acquisition, which means the susceptibility-mitigation and bolus-timing discipline covered throughout this article is now, in effect, an input specification for every downstream AI tool as well as for the human radiologist. A department that masters the fundamentals of the acute stroke MRI protocol described here is simultaneously building the technical foundation for reliable AI-assisted stroke triage, and that alignment between operational excellence and technological readiness is precisely what defines a modern comprehensive stroke centre.

Ultimately, every technical and clinical element covered in this article — vascular anatomy, relaxation physics, sequence order, contrast bolus dynamics, SAR management, pathology recognition, and the three-tier pitfall framework — exists in service of a single, unifying outcome: getting an accurate, actionable answer to the stroke team as fast as safely possible. The acute stroke MRI protocol will continue to evolve as faster gradients, higher-channel-count coils, and more capable AI triage tools become standard, but the underlying physiological logic connecting cytotoxic oedema, restricted diffusion, and the DWI-ADC signature will remain the diagnostic foundation this entire protocol is built upon for the foreseeable future.

References