Practical Aspects of Acute Stroke CT

Plain CT has a long track record of a safe and feasible exam in acute stroke patients. Only an occasional patient will need sedation or anesthesia for CT-based neuroimaging. Two or three injections of contrast are needed to obtain sufficient CT angiography (CTA) images of head and neck vessels (starting at the aortic arch) and perfusion CT (PCT) from supratentorial structures. Contrast nephrotoxicity or allergic contrast reactions are a rare occurrence especially if contrast is not administered to patients with a pre-existing history of renal failure. In our institution, contrast is injected for CTA and PCT before knowledge of the creatinine clearance in hyperacute patients who are potential candidates for revascularization treatment, with no significant increase in subsequent renal impairment [1]. Several precautions may be applied [2] to help limit nephrotoxicity. In addition to avoiding iodinated contrast in patient with a creatinine clearance <30 ml/min/m², uncontrolled thyroid disorders and known contrast allergy would be contraindications.

Non-Contrast CT (NCCT)

NCCT can be performed in less than a minute with a helical CT scanner, and is considered sufficient to select patients for intravenous thrombolysis with intravenous recombinant tissue plasminogen activator (RTPA) within 4.5 hours, and for endovascular treatment within 6–8 hours. It is a highly accurate method for identifying acute intracerebral hemorrhage (ICH) and subarachnoid hemorrhage, but quite insensitive for detecting acute ischemia. The approximate sensitivity of CT and PCT in different ischemic stroke subtypes is depicted in Figure 3A.1. The “fogging effect” on NCCT relates to the potential disappearance of hypoattenuation that can occur from approximately day 7 and up to 2 months after the acute stroke. It may result in false-negative NCCT in the subacute stage of ischemic stroke.

Figure 3A.1 Approximate likelihood of detecting ischemic stroke on non-contrast CT (NCCT) in territorial (continuous line) and lacunar (dashed line) infarcts. * indicates the fogging effect observed in the subacute phase on NCCT. The dotted line indicates approximate sensitivity of perfusion CT in non-lacunar supratentorial strokes. PCT data are from the ASTRAL registry. (P Michel, CHUV, Lausanne.)

Focal hypoattenuation (hypodensity) is very specific and predictive for irreversible ischemia, whereas early edema without hypoattenuation indicates low perfusion pressure with increased cerebral blood volume (CBV) and therefore represents potentially salvageable tissue [3]. In addition to hypoattenuation, loss of gray–white matter differentiation including in the insular cortex (“insular ribbon sign”) and isodense basal ganglia constitute the four early ischemic changes (EIC) that are mostly irreversible and are used to calculate the Alberta Stroke Program Early CT Score (ASPECTS) [4, 5]. Better detection and quantification of EIC has been shown in clinical practice when this score is used. Also, ASPECTS improves the prediction of clinical outcome independently of demographic, clinical, and treatment variables.

Regarding treatment response, ASPECTS based on NCCT does not add substantially to prediction of response to intravenous thrombolysis [6, 7]. Similarly, the relative benefit of endovascular recanalization treatment does not depend on the ASPECTS score, although there is a possibility that patients with very low scores benefit less [8]. EIC on NCCT predict post-thrombolysis symptomatic ICH independently of other factors [9, 10]. On the other hand, the extent of EIC on very early NCCT has not been shown to predict mass effect after acute ischemic stroke.

Perfusion CT (PCT)

PCT with iodinated contrast may be used in two ways:

• 
  

  as a slow-infusion/whole-brain technique

  
  
• 
  

  as dynamic PCT with first-pass bolus-tracking methodology.

The latter is preferable as it is quantitative and allows accurate identification of the ischemic penumbra [11]. In a patient with suspected acute ischemic stroke, a non-contrast baseline cerebral CT is immediately followed by PCT. Then, a CTA of the head and neck, and a contrast-enhanced CT of the brain are performed, requiring about 15 minutes for the entire examination. If the patient fulfills criteria for intravenous thrombolysis based on the NCCT, treatment may be started in the scanner while the patient is undergoing PCT and CTA. Similarly, image acquisition and processing usually overlap.

PCT examinations usually consist of two 40-second series separated by 5 minutes. For each series, CT scanning is initiated 7 seconds after injection of 50 ml of iso-osmolar iodinated contrast material into an antecubital vein using a power injector. Multidetector-array technology currently allows the acquisition of data from four adjacent 5–10 mm sections for each series. The lowest of these eight cerebral CT sections usually cuts through the midbrain and hippocampi; the other slices cover most of the supratentorial brain.

The PCT data are analyzed according to the central volume principle to create parametric maps of regional cerebral blood volume (rCBV), mean transit time (MTT), and regional cerebral blood flow (rCBF). The rCBV map is calculated from a quantitative estimation of the partial size averaging effect, which is completely absent in a reference pixel at the center of the large superior sagittal venous sinus. The MTT maps result from a deconvolution of the parenchymal time–concentration curves by a reference arterial curve. Finally, the rCBF values can be calculated from the rCBV and MTT values for each pixel using the following equation: rCBF rCBV / MTT. The maps can then be displayed graphically (Figure 3A.2).

Figure 3A.2 A 77-year-old patient, found on awakening with aphasia and right hemiparesis, NIHSS = 20. Perfusion CT maps depicting (A) regional cerebral blood flow, (B) regional cerebral blood volume, (C) mean transit time, and (D) core infarct maps according to a threshold model [12]. In (D), green: reversible ischemia (penumbra), and red: low likelihood of survival (infarct).

Raw maps of PCT images may be interpreted in a non-quantitative way by comparing the different parameters.

MTT is the most sensitive measure for decreased blood flow, but overestimates ischemia. rCBF is more specific in identifying salvageable tissue, and rCBV is the most specific parameter for irreversibly damaged tissue [13, 14], also in white matter [15]. Threshold maps separate reversible from irreversible ischemia [12, 16] and result in high interobserver agreement [16].

The 64-slice CT scanners allowing for eight or more brain slices have increased the detection rate for acute ischemic stroke [17], but diffusion-weighted MRI (DWI) remains more sensitive for small and infratentorial lesions. PCT has an overall sensitivity of about 75% for ischemic stroke, above 85% for non-lacunar supratentorial infarcts (Figure 3A.1), and a high specificity for ischemia [1, 13, 16].

Several PCT-based threshold models to differentiate ischemic, non-viable tissue (“core”), viable tissue (“penumbra”), and non-threatened tissue (“benign oligemia”) have been developed. According to calculations by Wintermark’s group [18], the ischemic area (penumbra and infarct) is defined by pixels with a greater than 145% prolongation of MTT compared with the corresponding region in the contralateral cerebral hemisphere [18]. Within this selected area, 2.0 ml/100 g represents the rCBV threshold: pixels belong to the infarct core if the rCBV value is inferior to the threshold, and to the penumbra if the rCBV value is superior to the threshold. Salvageable penumbra is displayed in green, and tissue with low likelihood of survival (infarct core) is displayed in red (Figure 3A.2). According to Parsons’ group, penumbra is present if the relative delay time is >2 seconds [19], and infarct if the mean rCBF is <31% of the contralateral side [20]; this group has also shown, however, that this threshold may be lower if rapid recanalization through thrombectomy is achieved [21]. Finally, a threshold value initially developed for MR perfusion is now also used for PCT in the RAPID® software: it uses time to maximum (Tmax) maps with a threshold of above 4–6 seconds to differentiate penumbra from benign oligemia, and relative rCBF or rCBV <30% of normal to differentiate penumbra from core [22].

The latter thresholds seemed superior to other models in a recent direct comparison [23]. Most importantly, however, seems the fact that PCT identifies core [24] and penumbra more accurately than NCCT. PCT also shows brain perfusion alterations in about 25% of patients with transient ischemic attacks, which are sometimes still present after the resolution of the patients’ symptoms [25]. Focal hyperperfusion in relationship with epileptic seizures has been described, and focal hypoperfusion is rare [26, 27]. During the migrainous aura, poorly delimited hypoperfusion contralateral to the aura symptoms is found occasionally [28] and may be mistaken as ischemic stroke.

Overall, in the absence of an abnormality on PCT in a patient with stroke symptoms, one might suspect a posterior fossa stroke, a lacunar stroke, small cortical stroke [17], or a stroke-imitating condition (migraine, Todd’s paralysis, venous thrombosis, encephalitis, conversion syndrome). Acute recanalization treatments might be inappropriate in some of these patients.

Baseline PCT volumes correlate with stroke severity in the acute stage, and do so better in left-sided infarctions [29]. Overall, using PCT and CTA seems to be associated with a better long-term prognosis after stroke [1], also in the posterior circulation [30]. Several PCT parameters have been associated with clinical outcome, but few of them were tested in combination with well-established clinical variables, such as age or initial stroke severity. Initial penumbra volume seems to be an independent outcome predictor depending on recanalization: if recanalization occurs, large initial penumbra is an indicator of favorable prognosis and vice versa [31, 32].

PCT predictors of treatment response are partially established; it has been shown that thrombolysis saves salvageable tissue as identified by PCT [33], and the presence of large penumbra volumes is associated with better clinical outcome if recanalization is achieved [32]. In patients undergoing endovascular treatment, smaller CBV volumes are associated with better outcomes [34], and a favorable mismatch profile predicted better treatment response [35]. Still, patients treated endovascularly within the 6 hours who have an acceptable NCCT ASPECTS score seem to have a benefit independently of the PCT profile [36].

Several recent randomized controlled trials (RCTs) with positive results used PCT selection to demonstrate superiority of certain recanalization methods over others (intravenous tenecteplase; EXTEND-IA [37]; EXTEND-IA TNK). The most striking demonstration of the benefit of perfusion imaging-based patient selection for recanalization comes from two RCT treatments beyond 6 hours of stroke onset (or of last proof of good health if onset unknown). Both the DAWN and the DEFUSE-3 trials used mainly PCT to identify patients with a favorable imaging profile for endovascular treatment, and both showed a major long-term benefit [38, 39].

PCT-based predictors of post-thrombolytic symptomatic ICH are only partially known. Severity and volume of hypoperfusion on PCT seems to play a role [40, 41]. Similarly, PCT has the potential to predict mass effect after middle cerebral artery (MCA) stroke. No consensus has been found of the most promising marker, however; volumes of decreases in CBF or CBV [42] and blood–brain barrier permeability [43] have been proposed.

CT Angiography

Cerebral and cervical CTA is performed using intravenous administration of 50 ml of iodinated contrast material at a rate of 3 ml per second, and an acquisition delay of about 15 seconds. Data acquisition is performed from the origin of the aortic arch branch vessels to the circle of Willis and reconstructed as maximum-intensity projections (MIP) and three-dimensional reconstructions (Figure 3A.3).

Figure 3A.3 Same patient as Figure 3A.2. Upper row: imaging at 12 hours after going to bed: (A) plain CT, (B) CT angiography with occlusion of the middle cerebral artery (white arrow), and (C) perfusion CT with threshold maps. The patient was then given intravenous thrombolysis with rtPA at 13 hours after going to bed and 2.5 hours after awaking (approved study protocol with informed consent from family). Lower row: (D) plain CT at 24 hours with a small left basal ganglion bleed (dotted arrow), (E) CT angiography with repermeabilization, and (F) diffusion-weighted MRI at 5 days, showing a small, partially hemorrhagic lesion.

CTA has been shown to identify the site of arterial occlusion in acute ischemic stroke patients [44] with similar accuracy as digital subtractive angiography (DSA) and probably better than MR angiography (MRA) [45, 46]. Clot length can be assessed by thin-sliced NCCT [47] and CTA, although the latter may overestimate the length unless late images with collateral filling are considered [48]. Similar to clot length, the clot burden score can be calculated and correlates with stroke severity [49]. Clot presence, localization, length, and burden seem to predict clinical outcome [49–52]. Clot length and site also seem to predict recanalization after intravenous thrombolysis [47, 53, 54].

Good collateral circulation is associated with smaller early infarct volume and lower National Institutes of Health Stroke Scale (NIHSS) scores [53, 55, 56]. Correlation with better clinical outcome has been shown [53, 55, 57, 58]. A more dynamic method of collateral imaging is multiphase CTA: by adding two more axial acquisitions after the usual arterial phase, vessel filling in the peak venous and the late venous phases is assessed. Patients with a favorable multiphase CTA showed a very good response to endovascular recanalization in the ESCAPE trial up to 12 hours after stroke onset [59].

As an alternative to PCT and collateral imaging, CTA source images (CTA-SI) have been used to estimate infarct core and penumbra in anterior [60, 61] and posterior circulation [62].

CT and Intracranial Hemorrhage

Hyperintensity in acute intracranial hemorrhage (ICH) is present on NCCT from its onset in virtually all patients. Intraparenchymal calcifications or melanin-containing metastases may sometimes give false-positive results. Adding CTA is debated, but is probably useful in patients with higher risk of vascular malformations underlying the ICH, such as patients with superficial (lobar) ICH, without hypertension, and of younger age. One main advantage of adding iodinated contrast in ICH is that contrast extravasation (“leakage”) is an independent predictor of hematoma growth and poorer clinical outcome [63]. It is now a target for immediate hemostatic therapy in RCTs. Various radiological methods, including PCT [64], indicate that there is no significant ischemia around the hematoma.

Practical Aspects of Acute Stroke MRI

Magnetic resonance imaging (MRI) can be used as the first and sole modality for the emergency imaging of patients with suspected acute stroke. An acute multiparametric stroke imaging should combine diffusion-weighted imaging (DWI), fluid-attenuated inversion recovery (FLAIR), and T2*-weighted imaging. Total acquisition time of these initial sequences with 1.5T or 3T systems is about 10 minutes and enables examinations of acute stroke patients with moderate cooperation. Once contraindications are excluded, medication such as thrombolysis can be administered on the table and additional sequences can be acquired. This should include MR angiography (MRA: time-of-flight MRA and/or magnetic contrast agent based) of head and neck arteries, perfusion-weighted imaging (PWI; usually using contrast agent), and T1 imaging following injection of contrast. Fast image reconstruction makes the results of MRA or PWI available within a few minutes. A typical sequence of multimodal imaging including DWI, FLAIR, PWI, and susceptibility-weighted imaging (SWI) in an acute stroke patient is shown in Figure 3B.1.

Figure 3B.1 Multimodal MR-based imaging in a patient with a right MCA stroke. SWI (a) shows a mild cortical hemorrhagic transformation which is a contraindication for intravenous thrombolysis. A typical mismatch pattern can be seen with scattered DWI lesions on DWI (b) and ADC (c) and perfusion deficit on MTT map (e). MRA (f) shows MCA occlusion on the right. Thrombolysis was withheld with regard to the hemorrhagic transformation of one cortical infarction (a) and the different ADC values (c) and FLAIR (d) signal intensities indicating that stroke had likely occurred over several time points, some of which more than 4.5 hours before imaging.

Such a comprehensive set of acute examinations not only identifies ischemia in most patients, but provides valuable information on potential stroke imitators, high bleeding risk with thrombolysis, prognostic information, and pathogenetic mechanisms of the current stroke.

Contraindications to acute stroke MRI are most implantable electronic devices (such as cardiac pacemakers) and other metallic elements in the head region (such as first-generation aneurysm clips or foreign bodies in the eye). Agitated and claustrophobic patients will usually not tolerate this exam well enough to obtain high-quality images. In pregnant women, the potential effects on the fetus are poorly known, particularly in the first trimester, and magnetic contrast is in general contraindicated. Magnetic contrast agent should not be injected in patients with a creatinine clearance <30 ml/min because of the risk of systemic nephrogenic fibrosis.

Diffusion Images, FLAIR

Water molecules show random motion in brain tissue which is limited by intracellular organelles, increased along white matter tracts, and reduced perpendicular to tracts and bundles. This anisotropy of diffusion caused by white matter tracts is technically compensated by diffusion measurements in three orthogonal directions and calculation of mean diffusion (trace) images (Figure 3B.2).

Figure 3B.2 Technical aspects of DWI. Two diffusion sensitizing magnetic field gradients are added to a T2-weighted echo-planar imaging sequence (a). On DWI source images the signal is reduced if diffusion is measured in the direction of a tract (corpus callosum; b) and increased if measured perpendicular (c). On post-processed isotropic DWI (d), healthy tissue such as the right hemisphere shows low contrast between gray and white matter. Acute infarction is hyperintense on DWI (d; left caudate and lentiform nucleus and the insular cortex). The apparent diffusion coefficient (ADC) is reduced in the infarcted tissue (e). On DWI there is a much stronger contrast between affected and healthy tissue compared to conventional FLAIR images (f).

Sensitivity of DWI at standard slice thickness of 5–6 mm is 80–90% [65, 66], which is about twice the sensitivity of acute non-contrast CT examinations (30–60%) [66, 67]. DWI lesions represent infarction in the majority of stroke patients, but a few other conditions are associated with them, as well. Transient global amnesia is associated with hippocampal DWI positive spots [68] and DWI hyperintensities have been reported after seizures [69], and in multiple sclerosis (MS) plaques [70]. DWI hyperintensity is also seen along the margins of acute intracerebral hemorrhage and in brain abscesses.

Stroke symptoms with initially negative DWI can be explained by symptomatic hypoperfusion above 12 ml/100 g/min, by a very small lesion below the resolution of the MRI, or are initially considered negative due to noise or motion artifacts. A normal DWI exam in a patient with suspected stroke may also indicate a stroke imitator such as epileptic seizures, hypoglycemia, and migraine with aura. Such patients presenting with stroke-like symptoms but showing neither infarction/ischemia nor vessel obstruction are unlikely to benefit from thrombolysis [71].

Lesions on DWI are also a good marker for the core volume in acute ischemic stroke, because only about 10% of it is reversible [72]. This percentage may be higher in very early imaging such as within 3 hours [73]. Several studies have identified an acute DWI lesion volume above 70–100 ml to be a critical size beyond which favorable outcome is highly unlikely [74–76].

At first sight, FLAIR imaging does not seem to be of particular value in acute ischemic stroke, given its lower sensitivity and less contrast between affected healthy tissue compared to DWI. On DWI, there is a much stronger contrast between affected and healthy tissue compared to conventional FLAIR images. As opposed to DWI, FLAIR abnormalities develop more slowly over several hours in acute stroke. Using this knowledge, it has been shown that the so-called DWI-FLAIR mismatch with FLAIR-negative DWI lesions identified patients within 4.5 hours of symptom onset with 62% sensitivity and 83% positive predictive value [77]. About 14% of acute stroke symptoms are diagnosed at awakening [78], and a few others may have unwitnessed stroke onset while awake. Those patients could not be treated with thrombolysis until now as the actual time of symptom onset is unclear. The randomized WAKE-UP study has now convincingly shown the safety and efficacy of thrombolysis in such unknown-onset stroke patients presenting with DWI-FLAIR mismatch within 4.5 hours after awakening [79]. Further use of FLAIR is discussed in the MRA section below.

Magnetic Resonance Angiography (MRA)

MRA directly reveals the location of the vessel occlusion. The size of the thrombus – related to the site of vessel occlusion in MRA – is an important determinant of vessel recanalization rates. It may be overestimated with MRA because stasis around a thrombus appears as occlusion.

Patients presenting with a major vessel occlusion or severe stenosis seem more likely to benefit from thrombolytic treatment [71, 80]. Still, many patients with more proximal occlusions on MRA reveal a considerable lesion growth and a poor outcome [81], especially if not rapidly revascularized.

Without MRA, signal intensity changes within vessels (“vessel signs”) in FLAIR images can be helpful in diagnosing the site of vessel occlusion (Figure 3B.3). However, the hypointense vessel sign in T2*-weighted MRI and the hyperintense vessel sign in FLAIR does not independently predict recanalization, risk for secondary intracerebral hemorrhage, or clinical outcome [82].

Figure 3B.3 Value of FLAIR and T2* images in vessel pathology. In a patient with distal MCA occlusion there is a hyperintense vessel sign of the MCA and its branches on FLAIR (white circles on two left images) indicating slow flow. There is a hypointense thrombus sign on a T2*-weighted image in the distal MCA. The thrombus causes a blooming artifact that is larger than the diameter of the affected vessel (white circle on the right).

Perfusion Imaging (PWI) and the Mismatch Concept

PWI maps are derived from the signal intensity change caused by the passage of contrast agent through the capillary bed and reflect several aspects of cerebral perfusion. Dynamic susceptibility contrast (DSC)-PWI is based on repetitive T2*-weighted acquisitions typically performed every 1.5 seconds.

The signal drop during contrast passage can be post-processed to maps of cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), time to bolus arrival or bolus peak, Tmax using delay corrected or uncorrected algorithms. Results can be expressed as relative (to unaffected hemisphere) or absolute values and different software provide different results for the same parameter maps [83]. Both the volume and the severity of the initial perfusion deficit are associated with the growth of the initial DWI lesion at follow-up imaging.

The penumbra in acute stroke patients has been defined as brain tissue with loss of electric activity and potential recovery after timely recanalization of the occluded artery. It is widely accepted that extension of hypoperfusion on PWI beyond the corresponding DWI boundary represents penumbra [84, 85]. This PWI > DWI mismatch has been used, validated, and refined in several studies [85–87]. Such baseline MRI findings can identify patients that are likely to benefit from reperfusion therapies and can potentially identify subgroups that are unlikely to benefit or may be harmed [88, 89]. This concept is now being used for patient selection into clinical trials, both within and beyond established time windows.

Systematic analysis of DWI and PWI patterns in the non-randomized DEFUSE study [88] and post-hoc analyses [76] have allowed the identification of patients presenting at high risk for bleeding and poor outcome despite thrombolysis (“malignant” profile): presence of >100 ml DWI lesion and/or Tmax 8 seconds delay perfusion. Similarly, the initial mismatch definition of a perfusion deficit of 10 ml or more and 120% or more of the DWI lesion was then refined to “target mismatch,” i.e. a core <70 ml, a significant hypoperfusion of <100 ml, and a mismatch ratio of 1.8. In patients who had reperfusion after thrombolysis, only the ones with an initial “target mismatch” had a more favorable clinical outcome. Such a pattern is also labeled a “favorable TRAIT profile,” TRAIT standing for “treatment-related acute imaging target” [90].

In the post-hoc analysis of the randomized EPITHET thrombolysis trial, target mismatch patients who were thrombolyzed had less infarct growth [91]. Identifying DWI-PWI mismatch has now become simpler with standardized thresholded software such as RAPID®, used in the DAWN and DEFUSE-2 and DEFUSE-3 studies [38, 39, 92].

The PWI > DWI mismatch concept has also been challenged [73, 93] on the grounds that the PWI lesion cannot discriminate reliably between benign oligemia and true penumbra, and because of noted overestimation of the extent of infarction seen at follow-up [94].

Arterial spin labeling (ASL) has recently been introduced as new perfusion imaging technology not requiring contrast agent. Blood labeled with a radiofrequency pulse can be used as an endogenous contrast agent. Zaharchuk et al. [95] and Bokkers et al. [96] compared ASL-DWI mismatch to DSL perfusion-DWI mismatch. They found moderate agreement between the two methods, and called for further studies. Niibo et al. [97] found good correlations of ASL values with traditional MRI core and penumbra thresholds.

Ongoing clinical development of ASL technology and further improvement concerning robustness and accuracy in stroke imaging can be anticipated.

Susceptibility Weighted Imaging (SWI) and Intracerebral Hemorrhage

For the evaluation of intracerebral hemorrhage, clinicians have traditionally relied on CT, in fear of missing or misdiagnosing an intracerebral hemorrhage by utilizing MRI only. Thrombolysis as the most effective treatment of ischemic stroke requires a rapid and reliable imaging assessment to exclude hemorrhage. Systematic studies suggest that MRI identifies intracranial hemorrhages rapidly and reliably, in particular if appropriate sequences are performed such as SWI (or gradient echo [GRE] or T2*) for intracerebral hemorrhage [98] and FLAIR for subdural hematomas (SDHs) and subarachnoid hemorrhage (SAH).

MRI in fact may be superior to CT, especially for the detection of small chronic hemorrhages, the cerebral microbleeds (CMBs). CMBs in the brain parenchyma diagnosed in T2*-weighted MRI should be interpreted in the light of the patient’s history as well as the location, number, and distribution of the lesions and associated imaging findings. The retrospective BRASIL study does not support the hypothesis that CMBs are associated with a higher risk for a clinically relevant intracerebral hemorrhage after anti-coagulation/anti-aggregation therapy or after thrombolytic therapy in stroke patients, and thus does not support the general exclusion of patients from therapy based on the presence of CMBs [99, 100].

SDHs can also be identified reliably with MRI. In the hyperacute setting, SDHs are best demonstrated on FLAIR sequences since FLAIR imaging nulls the effect of cerebrospinal fluid. On DWI, SDHs appear hyperintense and on T2*-weighted images they tend to be hypointense. The presence of mixed signal intensity within the SDH may indicate the presence of blood with different ages and MRI may emerge as a tool in selecting the therapeutic approach to SDHs [101]. Proton density-weighted images may add further value to the diagnosis of SDH [102].

For SAH detection, the best imaging sequences on MRI are FLAIR and proton density-weighted images.

Comparison of MR- and CT-Based Acute Stroke Imaging

The original PWI-DWI mismatch concept is based on MR imaging and postulates that critical hypoperfusion on PWI exceeding the borders of the DWI lesion indicates salvageable tissue which may justify reperfusion treatment [103]. This concept has now been refined as described below, and CT-based perfusion imaging has become an important focus of research to determine core and salvageable tissue, given its wide availability.

With regard to information about brain perfusion, PCT appears at least equivalent to MRI [14, 60, 104–106]. Significant correlation has been demonstrated between PCT-CBV and DWI, between PCT-MTT and PWI-TTP, and between CTA source images and DWI [14, 60]. If threshold models are used, the PCT core correlates well with DWI and PCT total ischemia with PWI-MTT [12, 104]. It has to be cautioned, however, that both imaging methods still need better standardization of terminology, better knowledge about thresholds, further validation of their independent prediction of clinical outcome, and testing in phase III clinical studies [107].

DWI as a marker of early infarct has the advantage of widespread availability and relative robustness. On the other hand, the linear relationship between contrast concentration and signal intensity may be an advantage of CT perfusion imaging over gadolinium-based MR perfusion imaging, allowing a more quantitative estimation of cerebral blood flow [108].

Both CTA and MRA detect significant stenosis and vascular malformations quite reliably [46, 109]. Whereas CTA better quantifies the degree of stenosis and identifies arterial calcifications, MRA is probably more specific in diagnosing cervical artery dissections [110].

Valid criticisms of MR imaging include its cost, limited availability, more difficult patient monitoring, pace maker incompatibility, and the longer time required for scanning [111]. Although exposure to radiation seems acceptable [112], it prohibits frequent use of PCT and CTA because of a probable risk of inducing cancer by radiation, particularly in younger subjects and with multiple exposures [113]. There are reports of radiation overdosing from wrongly calibrated scanners. Iodinated contrast can occasionally be associated with allergy, hyperthyroidism, or renal failure, although the last seems to occur rarely [114]. These drawbacks are counterbalanced by the availability of CT in most emergency rooms, its easy accessibility, and the easy monitoring of patients. Patients with known severe renal failure (creatinine clearance <30 ml/min) should probably receive neither iodinated contrast agents nor gadolinium [115].

Both CT- and MR-based perfusion and arterial imaging are feasible in the emergency setting. Both detect intracranial hemorrhage with high accuracy, and exclude the major contraindications for thrombolysis. Decisions for endovascular therapy based on non-invasive arterial imaging can be obtained by both methods with a sufficient degree of confidence. Advantages of MRI of a higher sensitivity to detect acute ischemic stroke (DWI) have to be balanced with the somewhat easier accessibility of CT. Both methods have certain limitations in patients with agitation and renal failure, as described below. Ideally, a stroke-receiving hospital offers both CT- and MRI-based imaging on an emergency basis, allowing physicians to choose the most appropriate one for a clinical question and situation. More realistically, most hospitals will choose one method as their first-line imaging, where indications and contraindications to multimodal sequences should be integrated in standard operating procedures of hyperacute stroke care.

Treatment Decisions Based on Multimodal Imaging

The concept of using advanced imaging for treatment decisions is similar for MR- and CT-based imaging. Through the above-mentioned systematic work, several important markers for benefit from acute revascularization treatment have been identified: low core volumes, large penumbra volumes accompanied by reperfusion, reperfusion (mostly through recanalization of large arteries), good collateral blood flow. In general, there are important relationships and co-dependence between such multimodal imaging parameters [116], and the most reliable and simple combination of variables for each clinical question still needs to be determined. In addition, generally applied thresholds in perfusion imaging might be inappropriate because the same degree of perfusion impairment might have a different impact on the tissue depending on patient age, the anatomic location, and time from stroke onset.

Analysis of multimodal MR-based treatment trials assessing the predictive value of multimodal imaging has shown its added value in some [88, 92], but not in others [89, 117], or only after reanalysis [80, 118]. The combination of a large artery occlusion, small core, large core-perfusion mismatch, and good cerebral collaterals has been labeled “TRAIT” (treatment-related acute imaging target [90]). Some [119, 120] but not all [80, 117, 121] retrospective analyses of recanalization treatments incorporating multimodal CT show a potential benefit.

Given the proven benefits in the first 2–3 hours after stroke onset and the high likelihood of substantial penumbra and limited infarct size very early on, intravenous thrombolytics should be given immediately after exclusion of major contraindications. Multimodal imaging should then be added during thrombolysis and guide further treatment decisions such as:

• 
  

  adding or not mechanical revascularization treatment

  
  
• 
  

  treating patients arriving late or with unknown stroke onset

  
  
• 
  

  selecting the best treatment modality to achieve rapid reperfusion

  
  
• 
  

  avoiding overtreatment that is likely futile or harmful.

Patient Selection for Clinical Trials Based on Multimodal Imaging

Several important markers for benefit from acute revascularization treatment can be identified by multimodal CT and MRI investigations, as described above. Such knowledge was gained in randomized and non-randomized trials applying multimodal imaging systematically before revascularization trials, with various results:

• 
  

  DEFUSE-1: non-randomized trial of intravenous rtPA at 0–6 hours. Finding: if mismatch pattern was present and reperfusion was achieved, clinical outcome was improved [88].

  
  
• 
  

  EPITHET: randomized trial using intravenous rtPA vs. placebo at 3–6 hours. Findings: initial analysis negative [89]. Post-hoc analysis of target mismatch patients, using coregistration techniques: less infarct growth was observed with rtPA [91].

  
  
• 
  

  DEFUSE-2: non-randomized trial of standard treatment vs. endovascular reperfusion treatment. Findings: if mismatch pattern was present and reperfusion was achieved, clinical outcome was improved [92].

  
  
• 
  

  MR RESCUE: randomized trial using standard treatment (including intravenous rtPA) vs. endovascular recanalization up to 8 hours. Findings: mismatch pattern was not associated with outcome [117].

Several RCTs using advanced neuroimaging for intravenous treatment selection have been completed:

• 
  

  DIAS-2: intravenous desmoteplase vs. placebo in the 3–9-hour window if 20% mismatch on PWI or PCT:

  
  
  
  
  • 
    

    Overall result negative regarding 3 months handicap

    
    
  • 
    

    Subgroup of occlusion patients had benefit [80]

    
    
  • 
    

    Subgroup of patients with 60% mismatch had benefit [122]

    
    
  • 
    

    Intravenous tenecteplase vs. rtPA in the 0–6-hour window if 20% mismatch on PCT tenecteplase was superior to rtPA regarding 3 months handicap.

  
  

RCTs have now successfully used multimodal imaging to select appropriate patients:

• 
  

  ESCAPE: standard treatment vs. endovascular recanalization up to 12 hours if favorable multiphasic CTA imaging [59]

  
  
• 
  

  EXTEND-IA: intravenous rPA vs. added endovascular recanalization up to 6 hours if favorable PCT [37]

  
  
• 
  

  DAWN: standard treatment vs. endovascular recanalization 6–24 hours after last proof of good health if small core on PCT or DWI, and important clinical deficit [38]

  
  
• 
  

  DEFUSE-3: standard treatment vs. endovascular recanalization 6–16 hours after last proof of good health if small core and important penumbra on PCT or DWI/PWI [39].

Lessons learned from such trials and large case series for endovascular treatment can be summarized as follows:

• 
  

  Advanced multimodal imaging can replace time as the main selection criterion, in particular in patients presenting with unknown stroke onset or late after onset. Once a patient is considered for revascularization, “time is brain” applies also for endovascular treatment [123].

  
  
• 
  

  Completeness of recanalization matters, in addition to speed.

Related posts:

Chapter 8 – Cardiac Diseases Relevant to Stroke Chapter 7 – Common Risk Factors and Prevention Chapter 16 – Stroke and Dementia Chapter 19 – Acute Therapies for Stroke Chapter 22 – Management of Acute Ischemic Stroke and its Late Complications Chapter 21 – Intensive Care of Stroke

Stay updated, free articles. Join our Telegram channel

Join