Exclude Intracranial Hemorrhage

Any acute intracranial hemorrhage (ICH) is a contraindication for acute reperfusion therapy ( Fig. 11-1 ). This may be intracerebral hemorrhage, subarachnoid hemorrhage, and subdural or extradural hematoma. Although NCCT is used as the standard-of-care test to exclude intracranial hemorrhage, apart from a small number of patients with autopsy correlation, there are no level I data to definitively evaluate the sensitivity of NCCT for ICH detection. Regardless, CT has been used in multiple stroke trials for detection of ICH.

Intracranial hemorrhage on CT in a 50-year-old woman who presented with headache and left hemiplegia; CT was performed at 90 minutes post onset. Axial NCCT reveals large right intracerebral hemorrhage (circled area) .

The appearance of ICH on CT depends on the time elapsed since the hemorrhage. In general there is a linear relationship between CT attenuation and hemorrhage hemoglobin protein content. In the hyperacute phase, the ICH consists of red and white blood cells, platelets, and serum that confer an attenuation of approximately 30 to 60 Hounsfield units (HU). Over the next few hours, clot retraction occurs with increasing attenuation, generally to 60 to 80 HU, but potentially up to 100 HU. The overall CT appearance is dependent on the plasma hemoglobin content, patient coagulation mechanisms, and slice thickness of the CT imaging technique. As further time elapses, the attenuation of the ICH on NCCT reduces, and by 3 weeks is generally less than the surrounding brain tissue.

MRI is considered equivalent to CT for detection of intracranial hemorrhage. The appearance of ICH on MRI is more complex, and the signal characteristics depend on the state of the iron atoms within the hemoglobin of the ICH and the red cell membranes. Intraparenchymal hemorrhage evolves through five distinct stages depending on time from onset. This can be divided into the hyperacute (first few hours), acute (hours to days), early subacute (days to weeks), late subacute (weeks to months), and chronic phases (months). Typical findings are detailed in Box 11-2 and seen in Figure 11-2 . Acute subarachnoid hemorrhage and intraventricular hemorrhage are best detected with a combination of fluid-attenuated inversion recovery (FLAIR) or double inversion recovery (DIR) MR images, and susceptibility weighted imaging (SWI).

Evolution of intracranial hemorrhage on MRI. A, MRI day 1 post intracerebral hemorrhage; B, 4 weeks post hemorrhage; and C, 4 months post hemorrhage. Axial T1 (top row), axial T2 (middle row), and axial SWI (bottom row) reveal typical signal and morphologic changes with evolution to a hemosiderin-lined slitlike cavity by 4 months.

Is It an Ischemic Stroke?

The presence of radiologic changes on imaging tests that are concordant with the clinical symptoms and signs is helpful in the diagnosis of a true ischemic stroke and excluding a stroke mimic. Ultimately, however, the diagnosis of ischemic stroke is clinical and not based on any sole radiologic test. Moreover, even with sensitive MRI techniques, imaging can be negative in cases of true ischemic stroke, particularly if symptoms are transient or clinically localized to the brainstem or subcortical territory.

NCCT findings of parenchymal hypoattenuation or swelling in a topographic distribution concordant with the clinical symptoms can confirm the clinical diagnosis of stroke ( Fig. 11-3 ). However, NCCT findings are time dependent and may be normal or near normal even in the setting of established or large infarct, particularly within the first 3 hours of stroke onset ( Fig. 11-4 ). Additional NCCT detection of a hyperdense thrombosed intracranial vessel can also confirm the stroke diagnosis. However, even using modern CT scanner technology for patients who may be eligible for reperfusion therapies, the diagnostic sensitivity for NCCT is approximately 50%. The addition of CT angiography (CTA) for detection of vessel occlusion increases sensitivity by another 10% to 15%. In addition, if the typical topographic distribution of infarction is discernable on the source images of the CTA, this too confirms the diagnosis of ischemic stroke rather than stroke mimic. Overall, a multimodal imaging incorporating CT perfusion (CTP) provides the highest CT-based imaging sensitivity for confirming diagnosis of ischemic stroke ( Fig. 11-5 ). Postcontrast CT may be helpful to aid rapid recognition of enhancing tumors that may mimic stroke by presenting with postictal neurologic deficit ( Fig. 11-6 ). Nonetheless, CT-based imaging strategies remain limited in their sensitivity for detection of subcortical and brainstem stroke.

Features of acute infarction on CT and MRI in an 82-year-old woman who presented with speech disturbance with unclear time of onset. A, Axial NCCT reveals parenchymal hypoattenuation and loss of gray-white differentiation in a vascular topographic distribution, concordant with the clinical symptoms of established stroke. MRI FLAIR ( B ), DWI ( C ), and ADC maps ( D ) are concordant with CT findings demonstrating parenchymal swelling, FLAIR hyperintensity, and diffusion restriction.

Large infarcts may be difficult to discern on NCCT. This 39-year-old woman presented with a right MCA syndrome with hemiplegia and neglect. A, Axial NCCT at 1 hour 40 minutes displayed at standard window width and level; the infarct is just appreciable. B, Axial MRI DWI reveals faint DWI changes in the right MCA territory. C, Axial MRI ADC map confirms diffusion restriction and large established right MCA infarct. CT acquisition to MRI acquisition time was 40 minutes.

Utility of CTP for detection of small distal territory strokes. This 59-year-old man presented with mild dysarthria, NIHSS 1. A, No definitive stroke is evident on NCCT at 2 hours post onset. B, Concurrent CTP reveals perfusion abnormality in the left frontal lobe, confirming the diagnosis of stroke. C, Subsequent MRI DWI confirms infarct at corresponding location as CTP abnormality.

Utility of contrast-enhanced CT in a 55-year-old woman who presented with speech disturbance, NIHSS 8. CT was performed 3.5 hours after onset. A, Axial NCCT reveals faint hyperattenuating area (oval) without intracerebral hemorrhage. B, Axial contrast-enhanced CT clearly shows abnormal parenchymal enhancement. Final diagnosis was seizure activity from glioblastoma.

MRI findings of parenchymal FLAIR and T2 hyperintensity and swelling in a topographic distribution concordant with the clinical symptoms also confirm the clinical diagnosis of stroke (see Fig. 11-3 ). However, similar to the parenchymal findings on CT, FLAIR may be negative, particularly within the first 3 hours of stroke onset, and the infarct only detectible on diffusion-weighted imaging (DWI). Thus it has been proposed that a rapid stroke imaging protocol using DWI scans alone may be of benefit in confirming the diagnosis of ischemic stroke and excluding stroke mimics. However, there remain issues with detection of small subcortical and brainstem strokes, particularly when DWI is performed within the first few hours of stroke onset. The use of MR angiography (MRA) and MR perfusion techniques may also increase diagnostic confidence for ischemic stroke if vessel occlusions or typical ischemic perfusion patterns are discernible. Moreover, by performing further MR sequences, an alternate cause for symptoms may be discernible ( Fig. 11-7 ).

Hemiplegic migraine mimicked acute stroke in a 5-year-old boy who presented with headache, right-sided hemiparesis, and altered conscious state. Axial T2 MRI ( A ) and DWI ( B ) do not reveal any stroke, nor intracranial vessel occlusion on ( C ) MR angiography. D, There is asymmetric cortical venous signal on SWI, a phenomenon reported in pediatric hemiplegic migraine. The patient made a full recovery.

Is There an Intracranial Vessel Occlusion?

Detection of a hyperdense thrombosed intracranial vessel on NCCT can provide confirmation of ischemic stroke diagnosis before parenchymal changes are discernible. Common locations include the horizontal or M1 segment of the MCA, the sylvian or M2 segment of the MCA, the terminal internal carotid artery, and the basilar artery. Identification of the hyperdense vessel on NCCT provides information about thrombus burden and has prognostic and therapeutic information. In particular, patients with a hyperdense vessel have greater thrombus burden, greater likelihood of early ischemic parenchymal changes, more severe neurologic presentation, and less likelihood of early neurologic recovery even after intravenous thrombolysis. Importantly, use of thin sections (≤2.5 mm) allows greater detection of hyperdense vessels ( Fig. 11-8 ). CTA is highly accurate for detection of central large vessel occlusions, with a “flow-gap” or vessel “cutoff” being diagnostic for a vessel occlusion. However, CTA only provides a snapshot in time and thus may not allow sufficient time for collateral flow to reach the distal end of the thrombus. This can result in overestimation of thrombus length on CTA, and the use of NCCT may be more accurate ( Fig. 11-9 ). On MRI, vessel occlusion may result in altered arterial T1, T2, or FLAIR signal, or T2* imaging, but greater sensitivity is achieved with SWI ( Fig. 11-10 ). Similar to CTA, MRA can be used for detection of vessel occlusion; however, MRA is prone to overestimation of vessel stenosis and typically limited to the more central vasculature compared to CTA.

Use of thin-section NCCT increases intravascular thrombus detection. This 85-year-old woman presented with right MCA syndrome; image shows NCCT at 90 minutes post onset. A, Axial NCCT at 5-mm-thick reconstruction fails to convincingly demonstrate right MCA intravascular thrombus (arrow). B, Axial NCCT at 2.5-mm-thick reconstruction shows the hyperdense right MCA (arrow).

Clot length measurement with CT demonstrating overestimation of clot length that can occur with CTA. This 47-year-old woman presented with rapidly improving right MCA symptoms; CT was performed 2 hours post onset. A, NCCT reconstructed in oblique plane measures clot in the right MCA branch at 8 mm. B, CTA measures segment of nonopacification at 20 mm, overestimating true clot length. C, Delayed postcontrast CT provides additional time for the collateral circulation to reach the distal end of the clot and measures segment of nonopacification at 6 mm.

Utility of SWI for clot imaging in a 52-year-old man who presented with ataxia. Axial T2 MRI ( A ) and DWI ( B ) reveal an acute left lateral medullary infarct and a tiny cerebellar infarct (arrows). The abnormal signal of the thrombus (circled) on the axial SWI ( C ) is easier to discern than the lack of flow signal on the MR angiogram ( D ).

How Big Is the Infarct Core?

The term infarct core refers to nonviable tissue that is irreversibly injured or with such critically reduced tissue perfusion that cellular death is inevitable. DWI is considered the gold standard for established and irreversible infarct, because DWI reversal is considered uncommon. The DWI sequence is sensitive to disruption of the normal random movement (diffusion) of water molecules. Proposed mechanisms for reduced water diffusivity in cerebral infarction include cell wall dysfunction with associated intracellular/extracellular water shift, increased intracellular viscosity, and increase in extracellular space tortuosity. DWI is obtained by using gradient pulses incorporated into a T2-weighted sequence. Both T2 and diffusion contribute to image contrast, with the magnitude of the diffusion weighting known as the b value. The b value is dependent on gradient duration, amplitude, and interval, with higher b values reflecting greater diffusion weighting. At high b values, diffusion properties are the main determinant of image contrast; however, the T2 component may remain visible, potentially leading to an effect known as T2 shine-through. The DWI image should be interpreted with the apparent diffusion coefficient (ADC) map, a postprocessed image where tissue contrast is dependent only on diffusion. Restricted diffusion, confirming infarct core, is characterized by increased DWI signal with typical corresponding decrease in ADC signal. In contrast, T2 shine-through is characterized by increased DWI signal with corresponding increase in signal on the ADC map. Changes of restricted diffusion manifest within minutes after acute infarction and peak between 24 and 72 hours. However, occasionally, early DWI may result in only faint or minimal changes on DWI, and thus both DWI and ADC maps should be carefully scrutinized in cases of hyperacute stroke (see Fig. 11-4 ).

Although some institutions use MRI as their primary stroke-imaging strategy because of the increased sensitivity of DWI over CT for established infarct, other institutions have been searching for a reliable and valid alternative using CT, given the rapidity and greater availability of CT imaging techniques. NCCT signs of established infarct are parenchymal hypoattenuation and swelling. Specifically in the MCA territory, loss of gray-white attenuation of the insular ribbon and hypoattenuation in the lentiform nuclei are considered early signs of MCA infarction. Adjusting the window width and level of the CT image increases detection of early ischemic changes ( Fig. 11-11 ). More standardized methods of assessment for early ischemic change is achieved by using standardized reporting templates, such as the Alberta Stroke Program Early CT Score (ASPECTS). ASPECTS was proposed as a more accurate and practical alternate method for clinicians and radiologists to assess the extent of parenchymal injury. Moreover, since an increase in the rates of death and dependency occurred with patients treated with intravenous thrombolysis if pretreatment ASPECTS was 7 or less, it has been suggested as an alternate to the “1/3 MCA rule.” However, there is only moderate interrater reliability for using ASPECTS in such a dichotomized fashion, which may have an impact on utility. As an alternate to NCCT, the source image from CTA provides additional information for the detection of ischemic change. Although initially considered to be as reliable as DWI in assessing the size of the infarct core, CTA source images on faster modern CT scanners commonly overestimate infarct core size.

Utility of variable CT window width and level. This 39-year-old woman presented with a right MCA syndrome with hemiplegia and neglect. Axial NCCT at 1 hour 40 minutes displayed at ( A ) standard window width and level (W:75 L:35) and ( B ) varied window width and level (W:35 L:35). The large right MCA infarct (oval) is easier to discern using the varied “Stroke” window width and level.

CTP techniques show increasing promise in the accurate detection of infarct core. CTP data are obtained by monitoring the passage of a bolus of iodinated contrast media through the cerebral parenchyma via serial CT scans. Complex mathematical equations are used to calculate the average time for the contrast bolus to pass through each part of the imaged brain parenchyma (mean transit time [MTT]), the time to peak enhancement (time to peak [TTP]), the volume of blood within each part of the imaged brain parenchyma (cerebral blood volume [CBV]), and the volume of blood passing through each part of the brain parenchyma over time (cerebral blood flow [CBF]). Each CT image is usually obtained in the axial plane, typically every 1 to 1.5 seconds for approximately 60 seconds. Interpretation is typically by visual inspection of color-coded parametric maps displaying MTT, TTP, CBV, and CBF data. Modern multislice scanners are able to assess the whole brain in a single examination, with the potential to display perfusion data in multiple planes ( Fig. 11-12 ).

Modern CT scanners may be able to assess perfusion of the whole brain and display CTP data as a brain volume reconstruction. This 80-year-old man presented with left hemisensory changes, NIHSS 2. A, Superior view of whole brain volume CTP demonstrating elevated MTT in the right MCA territory. B, Oblique whole brain volume reconstruction.

There is debate on which is the best CTP parameter to delineate infarct core. Early data suggested that CBV provided the most accurate measure of core volume. More recent data suggest that relative CBF may be the most robust parameter. Importantly, there is significant variation in perfusion maps depending on the CT scanner and postprocessing software used, and thus each local institution should gain their own experience using their own postprocessing software. Current research trials are investigating the feasibility of automated software to improve reproducibility and provide standardization, and these may be available for routine clinical use in the near future.

How Big Is the Penumbra?

The term penumbra refers to tissue that is potentially viable (i.e., with reduced perfusion that results in cellular dysfunction but not inevitable cell death). Thus penumbral tissue is at risk of infarction but may recover with adequate and timely reperfusion. Importantly the distinction of infarct core and penumbra were not used in the NINDS or ECASS trials, and thus intravenous thrombolysis is generally administered after initial NCCT imaging at most centers.

Within 3 hours of stroke onset, penumbral information may be evident on NCCT. It has been postulated that areas of brain parenchyma that are isoattenuating and slightly swollen may represent penumbral tissue rather than established infarct. However, the majority of ischemic penumbra appears normal on NCCT. CT and MR perfusion MTT or TTP maps are most commonly used in comparison to the CBF or CBV maps to estimate the size of the penumbra. In addition TMax maps may also be obtained based on further mathematical equations. MTT and TTP abnormalities may be discernible immediately after symptom onset, making CTP a valuable tool to confirm the suspicion of ischemic stroke. Large cortical perfusion abnormalities, such as those that occur with proximal vessel occlusion, are easier to detect than smaller or subcortical perfusion abnormalities. MTT and TTP abnormalities are difficult to detect in the posterior fossa, because this area is prone to artifact on CT and may not be adequately included during CTP. Importantly, for clinical use it must be noted that there is variation in the appearance of the perfusion maps depending on the CT scanner and postprocessing software used. For ischemic stroke trial use, more reproducible information is required, and thus various CTP parameters have been validated against MRI, with CT and MR TMax greater than 6 seconds considered reliable in the assessment of penumbra.

The great interest in core and penumbral assessment is in the advancement of the mismatch hypothesis. The mismatch hypothesis is based on identification of the infarct core and the penumbra, with the volumetric difference representing the mismatch (i.e., tissue at risk of infarction without timely reperfusion) ( Fig. 11-13 ). Restoration of perfusion to the penumbral tissue results in clinical recovery, and thus patients with mismatch could be selected for reperfusion therapies. This promising concept is being investigated in a number of trials, particularly those outside traditional treatment windows (wake-up strokes, >4.5 hours since presentation), because mismatch may persist for at least 48 hours in selected cases. The mismatch concept could facilitate treatment of a larger number of stroke patients, as the majority of stroke patients present to hospital outside the traditional intravenous thrombolysis treatment window of 4.5 hours.

Mismatch on CTP in a 24-year-old man who presented with left MCA syndrome, NIHSS 11. CT perfusion at 4.5 hours post initial TIA and 3 hours post recurrence of hemiparesis. A, Axial MTT map reveals large area of MTT abnormality. B, Near-normal CBV in the corresponding territory, indicating perfusion mismatch.

Is There an Etiology Evident?

There are a number of risk factors for ischemic stroke, such as age, male sex, hypertension, diabetes, hypercholesterolemia, smoking, and family history of cerebrovascular disease. However, the exact stroke trigger that initiated the ischemic event often remains unclear. Over 20 years ago, the Trial of Org 10172 in Acute Stroke Treatment (TOAST) investigators created a clinical classification system of inferred stroke etiology that has been widely adopted in stroke registries and trials. The TOAST classification recognizes 5 etiologies: (1) large-artery atherosclerosis, (2) cardioembolism, (3) small-artery occlusion, (4) stroke of other determined etiology, and (5) stroke of undetermined etiology. Diagnosis is based on clinical features, laboratory tests, and imaging of the brain, heart, and cervicocranial vessels. For a large-artery atherosclerosis mechanism, more than 50% stenosis of the relevant intra- or extracranial vessel is required, as is exclusion of a cardiac embolic cause. However, over the last 20 years, there have been advances in imaging, and in particular, use of DWI has become routine in the evaluation of ischemic stroke. This has shed further light on the vascular mechanisms of stroke and facilitates more accurate, consistent, and evidence-based classification of the cause of ischemic stroke in the individual patient.

The use of modern CT and CTA or MR and MRA techniques can help determine ischemic stroke etiology. A large-artery mechanism is most likely in the setting of high-grade stenoocclusive disease with an infarct in the relevant vascular territory and the absence of acute infarction in other vascular territories. However, while the TOAST classification would consider greater than 50% stenosis of the relevant vessel for a large-artery atherosclerosis mechanism, complex plaque with lesser degrees of stenosis may also provide an embolic source. Thus the degree of stenosis is best used to determine the need for carotid endarterectomy rather than necessarily determine the stroke etiology.

MRI with DWI provides the greatest sensitivity for detection of small acute infarcts. Similarly, detection of multiple acute infarcts closely related in time in both the right or left anterior and/or posterior circulations makes a central embolic cause most likely. In certain cases, the intramural hemorrhage from arterial dissection and secondary occlusion or embolic infarction may be evident on CT or MRI ( Fig. 11-14 ). In particular, cervicocranial arterial dissection should be considered in patients who do not have typical stroke risk factors, when there is a history of neck trauma or manipulation, connective tissue disorders, or where there are supporting clinical features such as neck pain.

Utility of fat-suppressed T1 MRI for diagnosis of arterial dissection. This 46-year-old man presented with headache, neck pain, and Horner’s syndrome. A, Contrast-enhanced MR angiogram reveals severe narrowing (arrow) of the distal cervical ICA. Sagittal ( B ) and coronal ( C ) reconstructions of 3D volumetric T1 fat-suppressed MRI clearly show the intramural hematoma (arrows).

Prognostication

Although multiple factors such as age and severity of neurologic deficit affect the clinical outcome of a patient with ischemic stroke, there are imaging factors that also provide early prognostic information. The size of the infarct core prior to acute reperfusion therapy is a predictor of clinical outcome and response to therapy and can be assessed using CT and MRI techniques as described previously. In particular, patients with NCCT ASPECTS less than 7 and DWI volumes of more than 70 to 100 cc are more likely to have poor functional outcome regardless of reperfusion and may incur harm with acute reperfusion therapies. In addition to the core volume, infarct location in more eloquent areas—in particular the internal capsule or corona radiata—adversely affects outcome. The presence of proximal intracranial vessel occlusion is another strong predictor of clinical outcome, with greater thrombus burden leading to poorer clinical outcome. Moreover, greater thrombus burden may be less responsive to intravenous reperfusion therapies and require alternate endovascular approaches.

Complications of Acute Stroke Treatment

The major concern after acute reperfusion therapy for ischemic stroke is intracranial hemorrhage. In the NINDS trial, as assessed by CT scan performed at 24 hours or with clinical suspicion for hemorrhage before 36 hours, asymptomatic hemorrhage occurred in 11% of patients treated with intravenous thrombolysis; 6% overall were symptomatic. In the ECASS III trial, as assessed by CT or MRI performed before 36 hours, 27% of treated patients had any intracranial hemorrhage; overall 2% were symptomatic. The rates of any intracranial hemorrhage after endovascular reperfusion approximates 20% to 30%, with rates of symptomatic hemorrhage ranging between 2% and 7%. However, multiple variables have an impact on the likelihood of symptomatic hemorrhage, including age, severity of neurologic deficit, infarct size, and serum glucose levels.

To accurately characterize complications, it is important to distinguish between hemorrhagic infarction and parenchymal hematoma formation. For accuracy of interpretation it is useful to consider the definitions used in stroke trials, and in particular the ECASS trials. This defined hemorrhagic events on CT as hemorrhagic infarction (HI) types I and II and parenchymal hematoma (PH) types I and II. HI I was defined as small petechiae along the infarct margins of the infarct; HI II, confluent petechiae within the infarct but without mass effect; PH I as focal blood clot in less than 30% of the infarct area, with mild mass effect; and PH II as focal blood clot greater than 30% of the infarct volume, with significant mass effect. PH II was a significant predictor of poor clinical outcome; other types may have less impact on clinical outcome. Of note, these hemorrhagic events were typically assessed with CT, and the advent of new MRI techniques such as SWI can increase the detection of intracranial hemorrhage ( Fig. 11-15 ). This makes it even more important to distinguish between hemorrhagic infarct and parenchymal hematoma, because HI is common after ischemic stroke and particularly after reperfusion therapies. In addition, little is known about the normal temporal evolution of ischemic stroke on SWI, which may include development of minor HI that is not discernible on CT or conventional MRI. This may more closely reflect the higher incidence of HI evident on autopsy studies as compared to CT studies.

Hemorrhagic infarction on MRI in a 69-year-old man who presented with right hemiparesis outside the time frame for treatment; MRI was performed at 9 days post onset. Axial T1 ( A ), T2 ( B ), FLAIR ( C ), and SWI ( D ) reveal large left subacute MCA territory infarct, with cortical blood products easiest to identify on the SWI sequences.

Hemorrhagic events should also be distinguished from the common parenchymal or subarachnoid contrast staining that can mimic hemorrhage after endovascular stroke therapy. Although this may be distinguishable with delayed interval CT or MRI, the dual energy CT techniques provide a useful alternative for immediate distinction on posttreatment CT scans. Endovascular stroke therapy may also lead to procedural embolic complications and infarcts in remote territories. Although this occurs in fewer than 15% of cases, it may adversely affect clinical outcomes. Thus interpretation of posttreatment imaging should assess for the presence of hemorrhagic complications, contrast staining, and new embolic infarcts.