The first role of imaging is to diagnose ischemic stroke and exclude hemorrhagic stroke, which would contraindicate revascularization treatment. Computed tomography (CT) and magnetic resonance imaging (MRI) based on T2*-weighted sequences are equivalent in detecting acute hemorrhage and a target thrombus in an occluded artery (susceptibility vessel sign) [15-17]. MRI has better accuracy for the detection of small petechial hemorrhagic transformation in infarction and previous hemorrhage and/or microbleeds. The high incidence (15–27%) of cerebral microbleeds in patients receiving IVT has led to concern about the risk of symptomatic intracerebral hemorrhage in these patients after IVT [18]. Studies have consistently reported that the risk of symptomatic intracerebral hemorrhage in patients with >10 cerebral microbleeds (30–47%) is significantly greater than that in patients without cerebral microbleeds [18]. A meta-analysis showed that the presence of cerebral microbleeds is associated with worse outcomes after IVT compared with patients without microbleeds [19]. However, it is unclear whether these negative effects of cerebral microbleeds fully negate the benefit of IVT. Considering these studies, the 2019 guidelines newly recommend the following: IVT is reasonable in eligible patients who previously have had a small number (1–10) of cerebral microbleeds and may be reasonable in patients with >10 cerebral microbleeds if there is a potential for substantial benefit (Table 1).

Assessment of the core on baseline neuroimaging is critical to select eligible patients for IVT and EVT (Tables 1, 2). Noncontrast CT (NCCT) is the most widely used first-line imaging tool for this purpose because of its advantages of quick acquisition and easy accessibility. NCCT findings of the core within a few hours after symptom onset are loss of gray-white matter differentiation because of a decreased CT density of gray matter structures. This appears on NCCT as a loss of distinction between the basal ganglia and adjacent white matter and as a blending of the densities of the cortex and underlying white matter in the insula (insular ribbon sign) and over the convexities (Fig. 1A, B). Sulcal effacement, another early NCCT finding, is produced by the subsequent swelling of the gyri. More advanced ischemia induces an obviously hypodense core on NCCT, which is considered irreversible tissue damage. Except for eligible patients for IVT within a 3-hour treatment window, patients with a core involving more than one-third of the middle cerebral artery (MCA) territory were excluded from revascularization treatment because the patients had a lower chance of good outcomes after revascularization treatment (Tables 1, 2) [8]. However, identification of patients following the 1/3 rule was unreliable in clinical practice due to a considerable lack of interrater agreement in recognition and quantifying early ischemic NCCT findings [20]. Addressing this limitation, a semiquantitative way to report early ischemic NCCT findings, the Alberta Stroke Program Early CT Score (ASPECTS), was developed [21]. ASPECTS evaluates the presence of early ischemic NCCT findings (loss of gray-white matter differentiation and sulcal effacement) in 10 regions of the MCA territory: 4 for subcortical structures (caudate, lentiform, internal capsule, insula) and 6 for cortical structures that are labeled M1-M3 at the ganglionic level and M4-M6 at the supraganglionic level (Fig. 1C). The rater subtracts 1 point for every abnormal region, and the total score ranges from 10 (without early ischemic change) to 0 (core involving the whole MCA territory). A recent meta-analysis performed a central reading of all baseline NCCT from the major 5 EVT trials in 2015 [3-7] and found a clear benefit of EVT in patients with ASPECTS ≥6 [22], which has been adopted in current guidelines. However, for accuracy, the ASPECTS has the innate problem that NCCT itself has a significantly lower sensitivity than MR to depict cores in early time periods, with an overall sensitivity of 57–71% in the first 24 hours of symptom onset and only 12% in the first 3 hours. The 10 regions in the MCA territory defined in the ASPECTS are based on anatomical structures, and thus, the individual regions cover different amounts of brain tissue. Therefore, the correlation of the ASPECTS with the core volume of diffusion-weighted MR imaging (DWI) varies considerably depending on the lesion location. As core volume is a strong predictor of functional outcome, the regional unequal weighing of the ASPECTS could compromise treatment decisions. Contrary to the low sensitivity of NCCT to depict cores, DWI has 73–92% sensitivity to depict cores in the first 3 hours and up to 95–100% sensitivity in the first 6 hours after ischemic onset (Figs. 1A, 1B, 1J, 1K, 2A, 3A, 3B, 4A) [15,16]. Cytotoxic edema (condition of restricted diffusion) in the ischemic brain occurs when cerebral blood flow (CBF) decreases under 30/100 g/min increased DWI signal intensity within minutes after ischemic onset. The DWI signal change is more conspicuous than NCCT, so direct measurement of early core volume is possible (Fig. 1). Current guidelines do not present DWI criteria for the recommendation of IVT or EVT in patients within 6 hours after symptom onset. Studies and clinical trials have shown that a DWI-ASPECTS ≥5 and DWI lesion volume <70 mL could be optimal thresholds to predict the benefit of EVT [4,7]. Manual outlining of DWI lesions has been used to extract volume data, but this method is time-consuming and operator dependent. To overcome this limitation, automated estimation of the core using an apparent diffusion coefficient threshold of 620×10^-6 mm²/second is currently applied in practice [23].

Considering hemorrhagic complications and the considerable rate of futile treatment [3-7], the ideal patient for revascularization treatment should have a salvageable brain (penumbra), which is a brain at risk of infarction due to critical hypoperfusion that can survive with reperfusion. Perfusion imaging using CT or MRI techniques is the most used method for the assessment of penumbra. CT perfusion (CTP) and MR perfusion (MRP) obtained by continuous scanning over 45 to 90 seconds after administration of contrast material provide information on tissue enhancement according to time, which represents the amount of contrast material, that is, the amount of blood reaching the brain. Cerebral blood volume (CBV), CBF, time-to-maximum (Tmax), and mean transit time (MTT) are displayed by calculation based on tissue density (or signal intensity) time curves of perfusion imaging. Studies have tried to identify critically hypoperfused tissue and core by applying thresholds for CBV, CBF, MTT, or Tmax. The mismatch area between the critically hypoperfused tissue and core is considered the penumbra, which is called the ‘target mismatch’ because the mismatch area is a target for revascularization treatment. A Tmax threshold >6 seconds (Tmax >6 s) has been a reasonable estimate of critically hypoperfused tissue in the absence of reperfusion (Figs. 1H, 3D, 4C) [24]. The mismatch area can be estimated with the difference between the volume of MRP Tmax >6 s and the DWI lesion when we use MRI as a baseline imaging tool. However, estimating the penumbra when the baseline imaging tool is CT is not simple because we cannot measure the volume of the core on CT images due to the inconspicuousness of early NCCT findings. Studies comparing CTP maps with DWI lesions showed that a threshold of <30% of the contralateral mean CBF (CBF <30%) was a reasonable CTP threshold of the core [25]. Therefore, in the CT workup, the mismatch area can be estimated with a difference between the volume of Tmax > 6 s and CBF <30% (Figs. 1A, 1B, 1H, 1J, 1K, 2A, 2C–F, 3A, 3B, 3D, 3E, 4A, 4C, 4D). Clinical trials using a target mismatch (EPITHET [26], SWIFT-PRIME [5], EXTEND-IA [7], etc.) to select patients for revascularization treatments showed a greater treatment effect relative to clinical trials that did not use the target mismatch concept (NINDS [1], ESCAPE [3], REVASCAT [4], MR CLEAN [6], etc.). In 2018, DAWN [12] and DEFUSE 3 [13] trials showed the benefit of EVT compared with medical treatment in patients last known to be normal 6 to 24 hours (DAWN) and 6 to 16 hours (DEFUSE 3) earlier. The DAWN trial used CTP to estimate the core with a threshold of CBF <30% and enrolled patients with a small core infarction that was mismatched to the clinical severity (NIHSS). DEFUSE 3 enrolled patients with a target mismatch on CTP or MRP. The 2019 guidelines promptly accepted the inclusion criteria of DAWN and DEFUSE 3 to select eligible patients for EVT in the late time window (Table 2).

Based on the WAKE-UP trial in 2017 [27], the 2019 guideline included another imaging-based concept called DWI-fluid-attenuated inversion recovery (FLAIR) mismatch, which is a DWI-positive and FLAIR-negative lesion, for identification eligibility for IVT in patients who awoke with stroke or have an unclear onset time >4.5 hours from the last known well or at the baseline state (Tables 1, 3). FLAIR hyperintensity, which requires an increase in total tissue water (vasogenic edema), lags behind DWI hyperintensity due to diffusion restriction (cytotoxic edema) as the stroke evolves. DWI hyperintense lesions without corresponding FLAIR hyperintensity indicated a stroke within 4.5 hours of onset with an 83–87% positive predictive value [28], and the WAKE-UP trial could use the DWI-FLAIR mismatch concept to select eligible patients for IVT in the late time window. Recently, Broocks et al. [29] showed that quantitative net water uptake between ischemic core and mirrored contralateral brain, which was calculated based on CT density, was comparable to DWI-FLAIR mismatch for identification eligibility for IVT in this patient group.

Current guidelines adhere to immediate administration of intravenous rt-PA after core evaluation (ASPECTS or core volume) if the patient meets the inclusion criteria within 4.5 hours after symptom onset and has no contraindications (Table 1). Delay of IVT for additional imaging workup is not permitted, so the next imaging work-up proceeds during rt-PA administration. The next goal of imaging is to find the occluded vessel that is the cause of the infarction to select eligible patients for EVT. Contrast-enhanced CT angiography (CTA) and MR angiography (MRA) are widely used for this purpose and accomplish the task with less than 90 seconds of acquisition time. Contrast-enhanced CTA (Figs. 1D, 2C) and MRA (Figs. 3C, 4B) can directly show not only the causative vessel but also the overall information of cervicocerebral arteries, such as anatomy of the aortic arch, tortuosity of the proximal extracranial neck vessels, atherosclerosis of the carotid bifurcation, cervicocephalic dissection, or other cerebrovascular diseases, such as vasculitis, intracranial aneurysm, or arteriovenous malformation. These factors affect EVT decisions and are important to neurointerventionists when planning an endovascular procedure. Modern EVT practice uses multiple devices and techniques tailored to the needs of a specific situation. Previous information on the aortic arch, neck vessels, and intracranial circulation, including thrombus location, helps in choosing appropriate devices and techniques, enabling more rapid and safe recanalization. Timeof-flight (TOF) MRA is a vascular imaging method that does not use contrast material (Fig. 2B). The sensitivity and specificity for detecting arterial occlusion are lower than those of contrast-enhanced MRA, and the acquisition time is relatively lengthy (approximately 3–4 minutes). However, it can be a good imaging option to detect the causative vessels in patients with renal impairment. Contrast-enhanced CTA and MRA and TOF-MRA show only a single static vascular status restricted to the early arterial phase. Slow-flow dynamics due to distal steno-occlusive lesions can result in pseudoocclusion appearing upstream of the exact steno-occlusive lesion on these single-phase angiographies (Fig. 2B, C). Multiphase CTA and dynamic contrast-enhanced (DCE)-MRA provide dynamic flow information to overcome this limitation and identify the location of the steno-occlusive lesion, which is misled on arterial phase images, on delayed phase images more exactly (Fig. 2D, E) [30].

A thrombus lodged in the occluded artery is shown as a hyperdense artery on NCCT, which is called a hyperdense artery sign due to increase density of thrombus by clotted blood. The hyperdense artery sign is present in <50% of patients on standard 3–5 mm-slice NCCT but is highly specific for vessel occlusion with a specificity of 90–100% [31]. This finding is shown as a susceptibility vessel sign on SWI, which is defined as a hypointense signal that exceeds the diameter of the contralateral artery located at the site of the thrombus. CT hyperdensity and paramagnetic susceptibility of thrombi are due to the higher proportion of RBC-rich clots that have more concentrated deoxyhemoglobin than fibrin-rich clots. Although it is still debatable, this fact represents the association of the hyperdense artery sign (or susceptibility vessel sign) with the cardioembolic cause of acute ischemic stroke [31]. The role of identifying vessel occlusion of the hyperdense artery sign on NCCT is not as significant today as it was when CTA was not popular. However, direct visualization of the location and extent of the thrombus and assuming the cause of stroke are quite useful information for neurointerventionists to decide on the appropriate thrombectomy devices, position the devices, and plan the subsequent procedure.

Cerebral collaterals are alternative vascular channels for maintaining blood perfusion to the ischemic brain distal to arterial occlusion. In cerebral ischemia due to large-vessel occlusion, cell viability depends on the collateral circulation, which has high interindividual variability. Therefore, collateral estimation provides the most potentially powerful prognostic insights. Estimation of the core and penumbra by NCCT, DWI, and perfusion imaging is an indirect method of collateral estimation. Initially, contrast-enhanced CTA was studied for direct collateral estimation. The arterial scoring method based on this single-phase CTA had a critical limitation of underestimating the collaterals with a longer transit time due to early triggering of a static acquisition [32]. Patients excluded from revascularization treatment because of poor collaterals on single-phase CTA may have quite good collaterals that simply arrived after CTA acquisition. Multiphase CTA overcame this limitation by 3 consecutive acquisitions after administration of contrast material to capture the delayed flow (Fig. 1E–G) [33]. A secondary analysis from the MR CLEAN [6] trial showed that a significant benefit of EVT was found in patients with good and intermediate collaterals on single-phase CTA and none in patients with poor collaterals [34]. The ESCAPE trial used multiphase CTA in imaging selection of eligible patients for EVT and showed that it helped to determine the clinical outcome [3]. Based on these studies, the AHA/ASA recommends that the incorporation of collateral flow status into clinical decision-making in some candidates to determine eligibility for EVT may be reasonable in the 2018 guidelines, and they kept this recommendation in 2019 [8]. However, the benefit of collateral-based patient selection over the perfusion imaging-based method in selecting patients who are most likely to benefit from EVT is questionable. The treatment effect of EVT in the ESCAPE trial [3] was inferior to the SWIFT-PRIME [5] and EXTEND-IA [7] trials, which used perfusion imaging to select patients with a target mismatch. A post hoc analysis of the SWIFT-PRIME [5] trial found a benefit for patients undergoing EVT regardless of collateral status on single-phase CTA [35]. We suppose that the inferiority of the collateral-based method compared to the perfusion imaging-based method was due to the limitations of current collateral imaging methods. Neither single-phase CTA nor multiphase CTA provides parenchymal perfusion information but only pial arterial information at 3 fixed time points according to CT table speed, which does not reflect individual hemodynamics, and it is still unclear whether the flow reaches and permeates brain tissue (Fig. 1E–G). The venous outflow concept has emerged from this limitation of artery-oriented collateral estimation. Brain perfusion is not only determined by the arterial inflow but also by downstream resistance due to tissue pressure or capillary and venous characteristics. Faizy et al. [36] showed that favorable venous outflow profiles, which were assessed by counting veins on CTA, correlate with favorable tissue-level collaterals and clinical outcomes. However, venous outflow profiles based on CTA have the similar limitation of underestimating the collaterals with a longer transit time or slow hemodynamics like single-phase CTA. A novel multiphase collateral imaging method, called collateral map, using dynamic signals of dynamic susceptibility contrast-enhanced MRP (DSC-MRP) and DCE-MRA was developed recently [37]. The collateral map consists of images of arterial, capillary, early venous, late venous, and delay phases determined according to the hemodynamics of each patient and provides more detailed information on brain perfusion and vessels from arterial phase to venous phase (Figs. 1L–O, 3F–J, 4E–I). Kim et al. [38] showed the clinical prognostic value of the collateral map in acute ischemic stroke due to anterior circulation large vessel occlusion. We expect that sophisticated collateral perfusion imaging with temporal information can provide precise information on the penumbra (Fig. 3). If this approach is clinically validated, then it will be possible to estimate the causative vessel, penumbra, and collateral perfusion status of a patient based on 1 imaging acquisition (DCE-MRA). And the MRI protocol can be simplified with a <4 minutes acquisition time composed of DWI (acquisition time, 34 seconds), SWI (97 seconds), and DCE-MRA (60–90 seconds). Additionally, Lee et al. [39] showed the predictability of symptomatic hemorrhagic transformation (parenchymatous hematoma 2) with collateral map and permeability imaging derived from DCE-MRA in patients with poor collateral perfusion status.

The prominent vessel sign, which is demonstrated as prominent cortical and/or medullary veins in ischemic brains on SWI, has been introduced as an imaging marker for collateral assessment [31,40]. The prominent vessel sign has been positively correlated with the amount of deoxyhemoglobin in cerebral veins secondary to an increased oxygen extraction fraction of the ischemic brain, which shortens T2* relaxation and decreases the signal intensity of the veins in the ischemic region with respect to that in a normal brain [31]. Therefore, ischemic tissue with poor collaterals requires more oxygen and contains more deoxyhemoglobins in tissue vessels, demonstrating a more prominent vessel sign on SWI than that of tissue with good collaterals. The prominent vessel sign is another indirect finding of collateral perfusion status, such as various perfusion parameters. Lu et al. [41] reported that mismatch between SWI (quantitatively measured volume of a region with a prominent vessel sign) and DWI showed good consistency with target mismatch of perfusion imaging and provided a reliable method to reflect the penumbra. SWI is an essential sequence in the baseline MRI protocol used for evaluating patients with acute ischemic stroke to exclude patients with intracranial hemorrhage from revascularization treatment. Therefore, assessment of collaterals and penumbra by SWI is an efficient triage option in time-pressured environments of acute ischemic stroke care, particularly in patients for whom contrast material is not recommended.