Updates in Ischemic Stroke Imaging: A Review

Stroke remains one of the top 5 causes of death and disability in the United States and the Americas.¹ Ischemic stroke is the most common type, with an annual incidence of 7 to 8 million cases globally.² Early diagnosis and clinical management of acute ischemic stroke (AIS) are imperative for optimizing outcomes.^3,4 Over the past few decades, the application of medical imaging has not only improved diagnostic accuracy for AIS (close to 97% sensitivity within minutes of onset),⁵ but also clarified its pathophysiologic characterization and disease course.⁶ In combination with rapid advances in computational technology, multiple medical imaging modalities, including CT, MRI, ultrasound, and fluoroscopy, have played a major role in standardizing clinical decision-making, management, and treatment of AIS, with concomitant improvement in clinical outcomes.^7-9

Multiple peer-reviewed articles reviewing decades of changes in guidelines and clinical uses of medical imaging in AIS are available, and continued discussion of the applicability of novel imaging modalities from different perspectives remains an essential component of scientific growth in this field.⁸ This review provides an up-to-date perspective on the role of medical imaging in the diagnosis, management, treatment, and prognostication of AIS.

The Penumbra and Early Treatment of AIS

Stroke penumbra—salvageable brain tissue at risk of ischemic infarction—is one of the earliest targets of clinical imaging in stroke. The definition of the stroke penumbra has changed over the years but primarily centers on electrophysiologic changes in the brain at the neuronal level after a reduction in the cellular fuel supply of adenosine triphosphate, a pathologic state that may be reversible as long as the cell membrane remains intact.¹⁰ Given that there is a limited time window before the penumbra progresses to a necrotic ischemic core, the primary goals of neuroimaging are early identification and measurement of the extent of salvageable penumbra. Positron emission tomography (PET) was the first technology to successfully calculate regional cerebral blood flow (CBF), metabolic rate, and oxygen extraction fraction (OEF); tissue with reduced CBF but preserved metabolic rate through increased OEF compensation was considered to represent the penumbra.¹¹ However, PET is unavailable in most emergency departments.

The first successful medical treatment of AIS was established in trials sponsored by the National Institute of Neurological Disorders and Stroke (NINDS), which showed that intravenous thrombolysis with tissue plasminogen activator (tPA) improved 3-month functional outcomes.¹² These positive results led to interest in quantifying early ischemic changes on CT¹³ and in estimating infarct core using MRI to identify potentially futile thrombolysis.¹⁴ Concerns regarding the risk of symptomatic intracranial hemorrhage (ICH) as well as questions about the benefits of tPA promoted several reanalyses and examinations of its effectiveness in clinical practice, using noncontrast CT (NCCT) as the only imaging modality.^15-18

Plain CT

Since the early days of thrombolysis, NCCT has been the mainstay for ruling out ICH. Even among the most difficult-to-detect types of bleeding, including subarachnoid hemorrhage, NCCT has high sensitivity (up to 95%) within the first 24 hours.¹⁹ The rapid acquisition time of NCCT (~1–2 minutes with modern scanners) and its relatively low index of harm to patients make it an invaluable imaging tool in AIS.²⁰ Highly specific NCCT results can also prove valuable for avoiding false-positive signs of hemorrhage, such as in calcified masses or cavernomas.²¹ To this end, high-throughput and machine-learning techniques continue to improve the specificity of NCCT for identifying all hemorrhages.²²

NCCT also has the potential to reveal signs of acutely infarcted core while showing chronic postinfarct changes, such as encephalomalacia and atrophy.²³ Brain parenchymal hypoattenuation, or hypodensity, on NCCT usually indicates early signs of tissue necrosis,²⁴ but distinguishing salvageable penumbra from irreversibly infarcted core remains challenging, even for experienced radiologists. Studies have shown little to no correlation between NCCT findings and time since stroke onset within 3 hours of symptom onset as well as failure of the Alberta Stroke Program Early CT Score (ASPECTS) derived from pretreatment NCCT to predict outcomes in the NINDS trials.^12,13,25

The feared complication of treating an already infarcted core with thrombolysis is hemorrhagic conversion due to a weakened capillary bed blood–brain barrier as well as worsening edema from reperfusion injury after recanalization.^26,27 Early European Cooperative Acute Stroke Study (ECASS) trials demonstrated diminishing benefit and increased hemorrhagic risk with delayed thrombolysis, particularly beyond 3 hours from onset, after which time the risks of thrombolysis administration may outweigh its benefits.^28,29 These findings were confirmed by the Alteplase Thrombolysis for Acute Noninterventional Therapy in Ischemic Stroke (ATLANTIS) trial, which found no benefit and increased symptomatic ICH risk with tPA administered between 3 and 5 hours after stroke onset.³⁰ The ECASS-III trial, using NCCT, demonstrated a safe and effective treatment window for systemic thrombolysis to 3 to 4.5 hours from symptom onset.³¹

ASPECTS, initially developed for evaluation of anterior circulation strokes, is an imaging metric designed to predict the extent of infarcted core by assessing hypodensity found on NCCT.³² Although ASPECTS quickly gained interest in the stroke community to guide eligibility for thrombolysis and patient selection in thrombectomy trials, its crudeness in evaluating diverse pathophysiologic changes across heterogeneous stroke mechanisms limits its reliability.³³ ASPECTS does not account for interindividual variability in cerebrovascular collaterals and cannot be applied to posterior fossa brain tissue, where bone artifacts affect resolution.³⁴ With the ability of NCCT to detect ICH,³⁵ the standard of care considers administration of thrombolysis in all disabling strokes as beneficial if no contraindications are noted on NCCT,³⁶ maintaining NCCT as the fastest front-line imaging test for evaluation of suspected stroke.

CT Angiography

Contrast CT angiography (CTA) availability expanded after NINDS trials showed its ability to demonstrate vascular patency with high sensitivity.^37,38 On CTA, lack of contrast distal to an occlusion has high specificity for poor tissue perfusion; however, it remains difficult to fully assess the extent of collateral flow and perfusion in routine clinical settings, even with high-resolution CTA.^37,39 Rapid development of cloud-based image transfer and artificial intelligence analysis technologies aims at automated detection of arterial occlusions, with several products commercially available.^40-42

With the emergence of intra-arterial therapeutics, large numbers of participants in the Prolyse in Acute Cerebral Thromboembolism (PROACT)–I and II trials were screened with digital subtraction angiography to confirm occlusions.^43,44 Lack of universal use of CTA at that time prompted investigators to use the National Institutes of Health Stroke Scale score as a substitute for occlusion presence in the Interventional Management of Stroke (IMS) III trial (NCT00359424), which had negative results.⁴⁵

The first successful trial deploying stent retrievers within 6 hours of onset—the Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands (MR CLEAN; ISRCTN10888758)—reignited interest in endovascular therapies.⁴⁶ Participant selection became more standardized and proved vital in the Endovascular Treatment for Small Core and Proximal Occlusion Ischemic Stroke (ESCAPE; NCT01778335) trial, which showed superiority of endovascular thrombectomy (EVT) over best medical management up to 12 hours from stroke onset.⁴⁷ ESCAPE trialists used multiphase CTA, which demonstrates good and intermediate collaterals, to exclude participants with poor collaterals, presumably improving selection of patients with small core infarcts.⁴⁷ EVT was shown to be beneficial, although severe complications of EVT were estimated at ~15% in 2018,⁴⁸ highlighting the need for improvement in reperfusion technologies and refined patient selection.

The success of early-window (≤6-hour) thrombectomy trials (MR CLEAN, ESCAPE, and Solitaire with the Intention for Thrombectomy as Primary Endovascular Treatment for Acute Ischemic Stroke [SWIFT PRIME]; NCT01657461) collectively provided the basis to explore the extended time window in trials such as DWI or CTP Assessment with Clinical Mismatch in the Triage of Wake-Up and Late Presenting Strokes Undergoing Neurointervention with Trevo (DAWN; NCT02142283) and Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke (DEFUSE 3; NCT02586415). Whereas early-window trials relied primarily on time from onset and NCCT or CTA results for participant inclusion, the extended-window trials introduced clinical core and perfusion mismatch paradigms to safely expand the treatment window up to 24 hours. Across all these studies, EVT consistently improved functional outcomes without resulting in increased rates of symptomatic ICH or mortality (Table 1).

CT Perfusion Imaging

With the acceptance of highly effective reperfusion therapy (ie, EVT) as the standard of care for individuals with large vessel occlusion (LVO), questions remain regarding how to extend the treatment window beyond the first 6 hours and which patients are unlikely to benefit from EVT. CT perfusion (CTP) imaging was added to multimodality CT for patient selection for EVT.⁴⁹ The Extending the Time for Thrombolysis in Emergency Neurological Deficits: Intra-Arterial (EXTEND-IA; NCT01492725) trial clearly demonstrated the value of CTP in patient “super-selection” (ie, identifying patients most likely to benefit from EVT). However, questions remain regarding which CTP cutoffs should be used to predict benefit or futility, as these EVT metrics were first established in the early 2000s.^14,50-53 Unlike the penumbra metrics derived from PET scans, CTP uses changes in flow rate and transit time distribution of administered iodine over time. CBF is measured from the average distribution of contrast within the vessels (biased by collaterals and multiphasing), cerebral blood volume is calculated by late-phase stagnation of contrast within the tissue (biased by leaky vessels), and maximum time is defined as the time it takes contrast to reach its peak volume from the middle cerebral artery.⁵⁰

A maximum time >6 seconds is thought to have high sensitivity in capturing tissue at risk (ie, the penumbra).⁵⁴ Although less sensitive and specific for estimating infarct core compared with presumed penumbra (96% accuracy), CTP, using an absolute cutoff of <30% CBF, still had relatively high accuracy in predicting infarct core (92% accuracy) in a nonrandomized criteria-derivation study.⁵¹ When tested in the setting of rapid and complete reperfusion, however, these CTP prognostic maps had only 50% to 60% probability of identifying salvageable vs irreversibly lost brain functions.⁵⁵

In theory, treatment of a large infarct core can be detrimental. Thus, pivotal trials that extended the EVT time window to 16 to 24 hours in participants with LVO focused on “slow growers” with smaller cores and larger mismatch ratios.^14,53 Slow growers were defined as participants presenting in an extended time window with smaller estimated cores and higher mismatch ratios, with larger presumed penumbral areas.

CBF estimates based on CTP and ASPECTS on NCCT can overestimate the core, defined as a “ghost core” (Figure 1).⁵⁶ The opposite can also be true, with CBF estimates from CTP having false-negative results (revealing no infarct core), particularly in smaller or deep brain penumbra with high collateral density (Figure 2).⁵⁷ Despite the global success of the EXTEND-IA, SWIFT PRIME, and Endovascular Revascularization With Solitaire Device Versus Best Medical Therapy in Anterior Circulation Stroke Within 8 Hours (REVASCAT; NCT01692379) trials in significantly reducing functional disability with EVT, critical appraisal of the data discouraged overuse of CTP to withhold EVT, particularly in the early time window.^51,58-60

Large Core Studies

The major randomized trials evaluating EVT in patients with large ischemic core infarcts, conducted between 2018 and 2023, include Recovery by Endovascular Salvage for Cerebral Ultra-Acute Embolism–Japan Large Ischemic Core Trial (RESCUE-Japan LIMIT; NCT02419794), A Randomized Controlled Trial to Optimize Patient’s Selection for Endovascular Treatment in Acute Ischemic Stroke (SELECT2; NCT03876457), Endovascular Therapy in Acute Anterior Circulation Large Vessel Occlusive Patients with a Large Infarct Core (ANGEL-ASPECT; NCT04551664), The Efficacy and Safety of Thrombectomy in Stroke With Extended Lesion and Extended Time Window (TENSION; NCT03094715), Large Stroke Therapy Evaluation (LASTE; NCT03811769), and Thrombectomy for Emergent Salvage of Large Anterior Circulation Ischemic Stroke (TESLA; NCT03805308).^61-66 Participants were generally adults with good premorbid functional status (baseline modified Rankin Scale score [mRS] of 0–1 or 0–2) and confirmed anterior circulation LVO. Imaging-based selection defined large cores using low ASPECTS (typically 3–5) or large infarct core volumes on CTP, with treatment windows extending up to 24 hours in most studies. Despite heterogeneity in imaging thresholds and time windows, all trials successfully challenged the historical exclusion of patients with large cores from EVT (Table 2).

Overall, these trials demonstrated that EVT provides a functional outcome benefit compared with medical management alone, although with higher absolute rates of disability and mortality in large-core populations than in early-window, small-core cohorts. ICH rates remained acceptable and generally comparable between EVT and medical therapy across studies. Mortality rates were similarly high in both groups in several trials. Collectively, these data support the safety and potential benefit of EVT in select patients with large cores, and underpin the recent paradigm shift toward tissue-based, rather than core-size–based, decision-making in AIS (Figure 3).

Medium Vessel Studies

Despite the success of EVT in the treatment of individuals with LVO, advancement of endovascular techniques, and their improved safety profiles, recent medium vessel occlusion trials, such as Endovascular Treatment to Improve Outcomes for Medium Vessel Occlusions (ESCAPE-MeVO; NCT05151172), Endovascular Therapy Plus Best Medical Treatment versus BMT Alone for Medium Vessel Occlusion Stroke (DISTAL; NCT05029414), and Evaluation of Mechanical Thrombectomy in Acute Ischemic Stroke Related to a Distal Arterial Occlusion (DISCOUNT; NCT05030142), failed to show a significant benefit of EVT in functional outcomes measured by the mRS.^67-69 However, appropriate patient selection for EVT using CTP imaging continues to allow stroke treatment in patients with a small core infarct and large salvageable penumbra (Figure 4). Medical imaging plays a key role in predicting which patients with medium vessel occlusion are likely to benefit from EVT.⁷⁰

Magnetic Resonance Imaging

MRI has undergone many technological modifications since the early 1990s, and has become a fundamental tool for the radiographic diagnosis of stroke. Its utility was initially considered limited, with little promise of capturing a stroke within a 7-day window.⁷¹ This view of MRI began to shift⁷² with the introduction of T2-weighted imaging, which enabled detection of subtle changes (eg, edema) in brain parenchyma,⁷³ and fluid-attenuation inversion recovery (FLAIR), which enabled highly sensitive detection of proteinaceous changes seen in necrosis, scarring, and abscesses.⁷⁴ Similar to NCCT, the lack of detectable tissue changes within the 6-hour treatable window initially limited the usefulness of MRI in hyperacute stroke workup.⁷² Moseley and colleagues⁷⁵ introduced diffusion-weighted imaging (DWI) in 1990; after multiple iterations, DWI became capable of capturing restricted water motion within human tissue, a highly sensitive finding even within the first minutes of AIS.⁷⁶ A head-to-head comparison of DWI MRI and NCCT demonstrated clear superiority of DWI in detecting early cerebral ischemia.⁷⁷

DWI was initially thought to show the core of infarction, whereas contrast perfusion MRI could demonstrate hypoperfused tissue at risk (presumed penumbra), thus enabling testing of the mismatch concept of tissue at risk.^78,79 Two thrombolysis trials in AIS used DWI/perfusion mismatch for patient selection and DWI changes before and after reperfusion as an outcome metric: the Echoplanar Imaging Thrombolysis Evaluation Trial (EPITHET; NCT00238537) and the original DEFUSE trials both showed reduction in infarct volume.^80,81 Although DWI provided an objective measure of acute stroke burden,⁸² infarct volume reduction did not become an acceptable primary end point in stroke trials, unlike clinical outcomes (ie, functional recovery at 90 days) measured by the mRS.⁸³ Early use of DWI/perfusion mismatch to guide EVT patient selection did not result in improvement in outcomes in the MR RESCUE trial, likely owing to ineffective endovascular devices.⁸⁴

Furthermore, the notion that a DWI lesion represents completed cerebral infarction may be incorrect in the early hours of ischemia. The Cambridge group compared DWI findings with CBF and OEF on PET.⁸⁵ Halted O2 extraction indicates loss of metabolism and completion of infarction within a DWI lesion. However, in the hyperacute DWI-positive studies, PET showed a heterogenous picture of reduced CBF and increased OEF as well as increased CBF and reduced OEF. The authors noted heterogeneity within and between DWI lesions, suggesting that hyperacute DWI may reflect metabolically dysfunctional tissue but not necessarily a complete infarction. This small but important study cautioned against excluding patients from treatment solely on the basis of DWI positivity, as withholding treatment risks a self-fulfilling prophecy in which untreated cerebral ischemia progresses to infarct growth, thereby making DWI appear predictive of the final infarct volume. Further evidence that absolute imaging cutoffs determining the ceiling effect of reperfusion therapies are relative can be found in the discussion of large core trials.

A wealth of diagnostic information can be derived by comparison of multimodal MRI sequences, and MRI offers safety advantages due to lack of radiation exposure and the low complication profile of gadolinium contrast (compared with CT iodine contrast), particularly in at-risk populations (eg, pregnant patients).^86,87 However, time delays related to safety screening (to avoid metallic interactions with the magnet), length of the imaging process (>20–30 minutes for multiple sequences), and higher rates of noncompliance (eg, claustrophobia, inability to lie flat or still) make MRI a relatively limited tool for frontline imaging in AIS.^88,89 Although some studies have evaluated MRI as an initial imaging modality, the risk of treatment delay often overweighs the risk of intravenous thrombolysis in patients with clinically suspected but imaging-negative AIS.^90,91

MRI was recently used for successful detection and treatment of patients with stroke with an unknown last known well time. The Efficacy and Safety of MRI-Based Thrombolysis in Wake-up Stroke (WAKE-UP; NCT01525290) trial revealed that patients who woke up with stroke symptoms or had an unknown last well time may benefit from thrombolysis if appropriately selected using MRI criteria⁹² (ie, DWI restriction without similar extension on T2-weighted FLAIR imaging [DWI/FLAIR mismatch]).⁹³

Conclusion

Medical imaging has fundamentally reshaped the diagnosis and treatment of AIS, enabling faster detection, better characterization of tissue injury, and more precise patient selection for reperfusion therapies. Advances in CT- and MRI-based techniques have shifted clinical decision-making from time-based thresholds toward tissue-based paradigms, safely expanding treatment windows and improving functional outcomes across a broader range of patients. As imaging technologies and analytic methods continue to evolve, their integration into clinical practice will remain central to optimizing stroke care and outcomes.