Non-contrast CT of the brain remains the mainstay of imaging in the setting of an acute stroke. It is fast, inexpensive, and readily available. Its main limitation, however, is the limited sensitivity in the acute setting. Detection depends on the territory, the experience of the interpreting radiologist and of course the time of the scan from the onset of symptoms. Whether tissue is supplied by end arteries (e.g. lenticulostriate arteries) or has collateral supply (much of the cerebral cortex) will influence how quickly cytotoxic edema develops 6. For example, detection of MCA territory infarct is approximately 60-70% in the first 6 hours 3, although changes in the deep grey matter nuclei (especially lentiform nucleus) can be visible within 1 hour of occlusion in up to 60% of patients 6.
The goals of CT in the acute setting are:
- exclude intracranial hemorrhage, which would preclude thrombolysis
- look for any "early" features of ischemia
- exclude other intracranial pathologies that may mimic a stroke, such as a tumor
Non-contrast CT has also been used historically to exclude patients from receiving thrombolysis based on the extent of hypoattenuation at presentation. This criterion has, however, been removed from the 2018 American Heart Association guidelines 18. Nonetheless, finding large areas of established infarction on acute non-contrast CT continues to play an important role in patient selection and management.
The earliest CT sign visible is the hyperdense vessel sign, representing direct visualization of the intravascular thrombus/embolus and as such is visible immediately 7,21. Although this can be seen in any vessel, it is most often observed in the middle cerebral artery (see hyperdense middle cerebral artery sign and middle cerebral artery dot sign) 21. It may be of therapeutic and prognostic value to differentiate this hyperdense 'regular' thromboembolic focus from a calcified cerebral embolus. In very rare instances of fat macroembolism, a hypodense vessel sign may be seen instead 22.
Within the first few hours, several signs are visible depending on the site of occlusion and the presence of collateral flow. Early features include:
- loss of grey-white matter differentiation, and hypoattenuation of deep nuclei:
- cortical hypodensity with associated parenchymal swelling with resultant gyral effacement
- cortex which has poor collateral supply (e.g. insular ribbon) is more vulnerable 6
Visualization of loss of grey-white matter differentiation is aided by using a stroke window with a narrow width (8-40 HU) and a slightly lower center than routine brain window 18, 41-43.
The hypoattenuation and swelling become more marked with time, resulting in a significant mass effect. This is a major cause of secondary damage in large infarcts.
As time goes on, the swelling starts to subside, and small amounts of cortical petechial hemorrhages (not to be confused with hemorrhagic transformation) result in elevation of the attenuation of the cortex. This is known as the CT fogging phenomenon 5. Imaging a stroke at this time can be misleading as the affected cortex will appear near normal.
Later still, the residual swelling passes, and gliosis sets in, eventually appearing as a region of low density with a negative mass effect. Cortical mineralization can also sometimes be seen appearing hyperdense.
CT perfusion has emerged as a critical tool in selecting patients for reperfusion therapy as well as increasing the accurate diagnosis of ischemic stroke among non-expert readers four-fold compared to routine non-contrast CT 9.
It allows both the core of the infarct (that part destined to never recover regardless of reperfusion) to be identified as well as the surrounding penumbra (the region which, although ischemic, has yet to go on to infarct and can be potentially salvaged). CT perfusion may also demonstrate early evidence of associated crossed cerebellar diaschisis.
The key to interpretation is understanding several perfusion parameters:
A simplified conceptual model that is still in use is that areas demonstrating matched defects in CBV and MTT represent the unsalvageable infarct core. In contrast, areas with prolonged MTT but preserved CBV are considered to be the ischemic penumbra 9.
However, contemporary automated software uses CBF reduction thresholds to estimate the infarct core and time-to-maximum (Tmax) to estimate the penumbra 45.
It is of note that CT perfusion may overestimate infarct core on admission, especially in the early time window of a stroke, by predicting lesion in areas that will not show infarct on follow-up imaging, a phenomenon known as a ghost infarct core 29.
These factors are discussed further separately, see CT perfusion.
- may identify thrombus within an intracranial vessel, and may guide endovascular clot retrieval
- evaluation of the carotid and vertebral arteries in the neck
- may be necessary before thrombolysis in pediatric stroke cases
- assess collateral vessels using single-phase CTA
- may identify cardioaortic thrombus as etiology of stroke, if the standard protocol is extended 30
Multiphase or delayed CT angiography is showing benefit either replacing CT perfusion or as an additional 4th step in the stroke CT protocol, as it guides patient selection for endovascular therapy by assessing collateral blood flow in ischemic and infarct tissue 24.
MRI is more time-consuming and less available than CT but has significantly higher sensitivity and specificity in the diagnosis of acute ischemic infarction in the first few hours after onset.
Within minutes of arterial occlusion, DWI demonstrates increased signal and reduced ADC values 4,10. This correlates well with infarct core (for a detailed discussion of DWI and ADC in stroke see diffusion-weighted MRI in acute stroke). At this stage, the affected parenchyma appears normal on other sequences, although changes in flow will be detected (occlusion on MRA) and the thromboembolism may be detected (e.g. the susceptibility vessel sign on SWI). Slow or stagnant flow in vessels may also be detected as a loss of normal flow void and high signal on T2/FLAIR and T1 C+ (intravascular enhancement), and the presence of the prominent vessel sign on SWI may indicate poor collateralisation 23.
If infarction is incomplete, then cortical contrast enhancement may be seen as early as 2 to 4 hours 10.
In a minority of cases, DWI may be normal (please refer to DWI-negative acute ischemic stroke for more details).
Generally, after 6 hours, high T2 signal will be detected, initially more easily seen on FLAIR than conventional fast spin-echo T2 10. This change continues to increase over the next day or two.
T1 hypointensity is only seen after 16 hours 10 and persists.
During the first week, the infarcted parenchyma continues to demonstrate high DWI signal and low ADC signal, although by the end of the first week, ADC values have started to increase. The infarct remains hyperintense on T2 and FLAIR, with T2 signal progressively increasing during the first 4 days. T1 signal remains low, although some cortical intrinsic high T1 signal may be seen as early as 3 days after infarction 10. After day 5, the cortex usually demonstrates contrast enhancement on T1 C+ 10. Less common patterns of enhancement include arterial enhancement, encountered in approximately half of strokes and becomes evident after 3 days, and meningeal enhancement, which is uncommon and is usually seen between 2 and 6 days 10.
Hemorrhage, most easily seen on susceptibility-weighted imaging (SWI), is not a good indicator of age. Although most commonly seen after 12 hours and within the first few days, it may occur earlier or as late as 5 days 10.
ADC demonstrates pseudonormalization, typically occurring at 10-15 days 10. As ADC values continue to rise, infarcted tissue progressively gets brighter than normal parenchyma. In contrast, DWI remains elevated due to persistent high T2/FLAIR signal (T2 shine through), unless hemorrhage (T2 blackout) or cystic encephalomalacia 10. T2 fogging is also encountered typically between 1 and 5 weeks, most commonly around week 2 10,11. Cortical enhancement is usually present throughout the subacute period.
T1-weighted sequences continue to show hypointensity throughout the area of infarct with cortical intrinsic high T1 signal due to the liquefactive necrosis and influx of monocytes as a response. The terms "cortical laminar necrosis" or "pseudolaminar necrosis" are occasionally, but incorrectly, used to describe this appearance in the context of thromboembolic stroke, but should be restricted to use in cases of isolated cortical necrosis. See the article on cortical laminar necrosis for a fuller discussion of this.
T1 signal remains low with intrinsic high T1 in the cortex if cortical necrosis is present 10. T2 signal is high. Cortical contrast enhancement usually persists for 2 to 4 months 10. Importantly, if parenchymal enhancement persists for more than 12 weeks, the presence of an underlying lesion should be considered 10.
ADC values are high. DWI signal is variable, but as time goes on, the signal progressively decreases.
Often described as an emerging application of point-of-care ultrasonography, use of transcranial Doppler (TCD) sonography has been utilized for the diagnosis of intracranial vessel occlusion, as well as the differentiation between ischemic and hemorrhagic stroke 14.
In the context of a CT negative for intracerebral hemorrhage and a clinically suspicious patient presentation, diagnostic criteria for occlusion of an isolated vessel are as follows 12:
- complete absence of color flow Doppler signals
- absence of pulsed-wave Doppler signals
- concurrent adequate visualization of surrounding parenchyma and vessels
- color flow and pulsed wave Doppler signals must be demonstrated adequately in the remainder of the circle of Willis 16
Sonographic monitoring of the complications of ischemic stroke is also possible, including the detection of: