In the context of acute stroke, computed tomography (CT) of the brain is the most widely used imaging method. All patients with suspected acute ischaemic stroke (AIS) must have a native CT scan of the brain (NCCT), which can assess the age and extent of ischaemia and exclude other pathology of sudden neurological deficit that will exclude the patient from recanalization therapy (tumour, haemorrhage) [1]. In the first hours of stroke onset, no changes may be apparent on the NCCT scan. Subsequently, early ischaemic changes develop (developing cytotoxic oedema, white-grey matter de-differentiation, loss of gyrification, flattering of the liquor spaces at the brain convexity, dense artery sign) until a hypodense rim corresponding to cerebral infarct is formed [1]. In clinical practice, we use the Alberta Stroke Program Early CT score (ASPECTS) to assess the extent of early ischaemic changes in the middle cerebral artery (MCA) territory [1]. Nowadays, automatic assessment of early ischaemic changes (Brainomix, iSchemaView RAPID) is being developed and used. In addition, CT perfusion (CTP) imaging is becoming clinically available. CTP is a functional examination of brain tissue that characterises the state of cerebral perfusion and provides information about the current status of the circulation/microcirculation [2]. In 2017, a review of 27 studies by Shen et al. [3] was published that evaluated and compared the diagnostic accuracy of CTP with NCCT and CT angiography (CTA) in the detection of AIS. The pooled sensitivity and specificity of CTP for acute ischaemic stroke were 82% and 96%, respectively. CTP was more sensitive than NCCT and had similar accuracy to CTA in detecting AIS. There were no statistically significant differences in sensitivity and specificity between patients who underwent CTP within 6 hours of symptom onset and 6 hours after the symptom onset [3]. However CTP has limited diagnostic utility in acute ischaemic posterior circulation stroke (PCS) [4]. Other limitations of CTP are lacunar strokes with 50% false negative cases and small cortical and subcortical strokes with a size of infarction under 3 cm${}^2$ and low sensitivity of CTP (29%) [3].

Published studies indicate that perfusion imaging may have a prognostic value in anterior circulation acute ischemic stroke and aid in the selection of patients outside the time window for administration of intravenous thrombolysis (IVT) or mechanical thrombectomy (MT) [5]. Although CTP has value in assessing tissue perfusion status in the hyperacute stroke setting and the long-term clinical prognosis of patients with AIS receiving reperfusion therapy, the prognostic use of CTP requires optimization and further validation [6].

The aim of this review was to give an overview of CTP acquisition and processing, clinical significance of CTP in anterior circulation stroke according to published meta-analysis and randomised controlled trials and its role in diagnostics of stroke mimics.

CTP represents a dynamic acquisition capturing a passing contrast bolus (its wash-in and wash-out) through the brain tissue [7, 8]. In other words, the brain is repeatedly scanned as the contrast medium flows through the region of interest and its relative increase, peak and decrease is measured, and an attenuation-time curve is generated [9].

Based on the derived attenuation-time curves that are obtained for arterial input function (representing arterial flow/wash-in) and venous output function (representing venous flout/wash-out), perfusion parameters are calculated for each voxel on a CTP map representing different hemodynamic properties [8, 9].

The essential hemodynamic parameters in CTP studies:

(1) Cerebral Blood Flow (CBF) refers to the volume of blood flowing in a unit of brain mass during a unit of time, measured in milliliters/100 g/min (mL/100 g/min). CBF is often expressed proportionately (relative CBF) as normalized measure to a presumed normal reference region (in the contralateral hemisphere).

(2) Cerebral Blood Volume (CBV) refers to the fraction of a tissue that is vascularized, expressed in the milliliters/100 g.

(3) Mean Transit Time (MTT) represents the average time that takes a contrast bolus to traverse the capillary bed; MTT is reported as an absolute in seconds.

(4) Time to maximum of the residual function (Tmax) expresses the delay from the start of scan acquisition to the maximum intensity of contrast bolus in each voxel.

CBF, MTT, and CBV are mathematically related by the equation: CBF = CBV/MTT, known as the central volume principle [10]. Therefore, measurement of any 2 of these parameters is sufficient to derive the third parameter [8], Fig. 1 (Ref. [11]).

The contrast passage and subsequent attenuation-time curves are dependent on arterial flow (expressed as arterial input function) and tissue characteristics. Among factors influencing the arterial flow of contrast media through the tissue belong impaired cardiac output, severe carotid stenosis or factors related to the contrast bolus injection (injection rate, saline chase). These factors may result in delay or dispersion of the contrast bolus which may introduce an error in the quantification of CBF [8, 12].

Raw CTP data can be processed with 2 techniques to generate perfusion maps, non-deconvolution and deconvolution methods. Non-deconvolution methods are based on first-pass iodine extraction measurements resulting in a simplified and less computationally intensive processing algorithm. Deconvolution methods account for physiological variations in arterial flow, the collateral flow effect, and venous outflow components of cerebral perfusion [8, 13].

There exist different deconvolution methods: singular value decomposition (SVD) [14], delay- and dispersion-corrected singular value deconvolution (dd-SVD) [15], and Bayesian methodologies [16]. dd-SVD or Bayesian-estimated generate the arterial delay time in contrast Tmax derived from SVD model. Use of different deconvolution methods may result in the measurement variability of the core and penumbra based on the deconvolution algorithm [8].

The goal of perfusion analysis in the clinical use is to quantify tissue with significant hypoperfusion which is likely to infarct if reperfusion is not achieved (ischemic penumbra) and identify tissue that is likely irreversibly infarcted (ischemic core) [5]. Some form of thresholding is usually applied to derived perfusion maps to obtain more consistent and objective result.

It is important to note that the CTP threshold is not absolute. It has been shown that the ischemic core may be overestimated if the perfusion imaging is performed very early after the symptom onset [17] and the use of more strict thresholds should be considered especially if the imaging is performed within the first hour (“golden hour”) [18]. The core overestimation was also demonstrated in those patients with achieved rapid reperfusion [17]. Although it is know that the white matter is more resistant to the hypoperfusion and therefore more strict thresholds should be applied to distinguish grey and white matter at risk of infarction with a higher accuracy [19], this differentiation is not widely used in clinical practice and remains to be so far of a research interest.

Prediction of the tissue fate is also influenced by the severity of hypoperfusion [17, 20]. Severely hypoperfused tissue defined as Tmax delay $>$10 s tends to progress more rapidly that the tissue with better residual perfusion through collateral flow. The severity of hypoperfusion can be quantified by the hypoperfusion intensity ratio (HIR), which represents the proportion of Tmax $>$6 s perfusion lesion with Tmax $>$10 s perfusion lesion. The high HIR was proved to predict the poor collaterals and infarct progression [20].

Multiple software packages generating perfusion maps are available. However, comparison of commercially available software shows significant differences in core and penumbra volume calculation due to different methods of generating perfusion parametres [17, 21, 22]. For instance, one software may generate highly delay-sensitive MTT maps, but other packages may generate delay insensitive maps [17]. Such a differences may subsequently influence patient selection for reperfusion therapy.

However, the automated image analysis offers the potential to improve patient outcomes by reducing triage, transfer and treatment times [23, 24] by identifying the patients with perfusion deficit especially in smaller, rural hospitals where physicians with experience in stroke imaging or manual processing of perfusion scans are not always available [25].

Within a time window of up to 4.5 hours for administration of systemic thrombolysis and up to 6 hours for mechanical thrombectomy, evaluation of native CT is considered sufficient preferably using ASPECTS score and CT angiography to indicate recanalisation therapy. Studies with both endovascular therapy and systemic thrombolysis (EXTEND-IA, DEFUSE3, DAWN, EXTEND, WAKE-UP) have shown benefit in patients treated outside the guideline recommended therapeutic window. The significance of CTP, which allows differentiation between necrotic core and penumbra, lies primarily in the extension of the possibility of treating patients with systemic thrombolysis or endovascular therapy. These patient groups are: wake up stroke, unknown time of onset and time of onset $>$4.5 hours for the indication of intravenous thrombolysis and $>$6 hours for mechanical thrombectomy [5].

A 2017 meta-analysis by Ruy et al. [26] examined the effect of treatment using perfusion imaging compared with CT protocols without CTP. 994 patients from 8 randomised and prospective studies published between 2011–2016 were analysed. In most studies, patients in the perfusion imaging group were treated in extended time window, up to 26 hours or with an unknown time of onset. Patients with both recanalization therapy (intravenous thrombolysis plus mechanical thrombectomy) were 1.9 more likely to achieve good clinical outcome, as expressed by modified Rankin Scale (mRS) of 0–2, when using perfusion imaging.

Another meta-analysis by Katyal in 2020 looked at randomised and retrospective studies with recanalization therapy (intravenous thrombolysis plus mechanical thrombectomy, or a combination of both) conducted from 2005 to 2020). Recanalization therapy was evaluated in patients with unknown time of onset, wake up stroke and stroke after 4.5 h for intravenous thrombolysis and after 6 h for mechanical thrombectomy, who were compared with patients treated within the standard window. From a total of 2226 studies, 15 with good quality methodology were selected. Nevertheless, the limitations of the meta-analysis mentioned that it comprised of prospective studies without a blinded placebo group, it was partly sponsored by commercial entities, and the CTP protocols were heterogeneous. A total of 5687 patients were analysed, 2040 (36%) with CTP guided recanalization therapy compared to 3647 (64%) with standard therapy. The following criteria were evaluated: good clinical outcome at 90 days (mRS 0–2), symptomatic intracranial haemorrhage (SICH), percentage of recanalization, and mortality. The odds of good clinical outcome were 2.3 times greater in the CTP-guided recanalization group (OR = 2.3; 95% CI 1.6–3.21). In the subgroup with endovascular therapy after 6 h compared to patients without recanalisation therapy, the good clinical outcome was even 5 times higher (OR = 5.01; 95% CI 3.07–8.17). There was no significant difference between groups in SICH, 5% in the CTP group versus 4.7% in control group. There was a significant reduction in mortality (OR = 0.75; 95% CI 0.59–0.74) and a higher likelihood of recanalization (OR = 4.37; 95% CI 2.52–7.58) in the CTP group [27].

CTP thus opens up the possibility of therapy for patients who would otherwise not be treated due to failure to meet the time specific criteria. Select 2 study with endovascular treatment and the use of CTP is currently underway looking for efficacy and safety in patients with large core infarcts and extended time windows [27].

Besides the role of CTP in the diagnosis of AIS, its role in prognosis of cerebral ischaemia was studied. Meta-analysis of Campbell showed that increasing ischaemic core volume (CTP time to maximum longer than 6 s) was independently associated with functional independence and functional improvement [28].

The frequency of stroke mimics is variable and depends on many factors. The diagnosis of stroke mimics is made approximately in 20–50% of cases of suspected acute stroke depending on whether the patients are evaluated by the emergency personal or stroke physicians [39].

According to the aetiology, stroke mimics can be divided into two groups: medical stroke mimics and functional stroke mimics. Diseases that may not have a positive finding in NCCT and therefore their diagnosis are difficult are as follows: recrudescence, migraine, epilepsy, vertigo associated with peripheral vestibular syndrome, attacks in multiple sclerosis and functional psychogenic disease. The most common of those is recrudescence which is defined as a sudden worsening of clinical symptoms in patients with neurological deficits after previous ischaemic stroke. The most common underlying cause is infection or worsening of a known cardiac disease [40]. The suspicion of stroke mimics is mainly based on the clinical presentation and neurological symptoms. Neurological symptoms typically do not correspond clearly to vascular territories and often fluctuate in nature. In stroke mimics, the suddenness of the onset is not always evident, fluctuations in symptom severity are common, and systemic signs are also often present [40] (Table 1).

It is not uncommon that patients presenting with stroke mimics have a previous history of epilepsy, migraine, depression or other psychiatric symptoms. The highest accuracy for ischaemic stroke diagnosis is achieved by CTP, whose pooled sensitivity is 82% (95% CI 75–88) and specificity 96% (95% CI 89–99). Sensitivity decreases in the diagnosis of posterior circulation stroke to 76% (70–81) and specificity to 93% (87–97) [39]. Although DWI MRI has a higher sensitivity in the diagnosis of ischaemic stroke, performing a multimodal CT examination in the acute phase in patients with suspected stroke mimics has multiple benefits. First, multimodal CT examination is able to exclude stroke mimics, second, the results are valuable for the treatment strategy and the patient’s prognosis [41, 42].

Some stroke mimics have typical CTP imaging patterns such as migraine and epileptic seizures. Up to 25% of migraine patients may present focal neurologic deficit imitating acute ischaemic stroke [43].

Migraine aura is usually associated with a perfusion deficit not limited to a specific vascular territory, and only a moderate increase of TTP [44].

In epileptic seizures neurologic deficit may last up to 48 hours after the convulsion [45].

Perfusion MRI studies can show a typical CBF increased (hyperperfusion) at epileptogenic area Outside the area, in the postictal phase, the perfusion parameters are normal or reduced but there is the absence of a typical vascular lesion distribution [46]. The CT perfusion study and CT angiography demonstrated a dramatic reduction in the cerebral blood flow and blood volume involving the entire left hemisphere, but with relative symmetry of mean transit time, ruling out a large vessel occlusion [47].

Acute treatment with systemic thrombolysis should be continued even if stroke mimics are suspected based on a negative multimodal CT scan. The systemic thrombolysis in stroke mimics is safe, the rate of intracranial haemorrhage is very low (0–2%), similar to the symptomatic ICH rate in patients treated with IVT for acute myocardial infarction [48]. In the SITS international stroke thrombolysis registry with 10,436 treated subjects, only 429 (4.1%) treated patients were diagnosed with stroke mimics. The most common mimics treated with thrombolysis were functional disorder (30.8%), migraine (17.5%) and seizures (14.2%) [39]. The correct diagnosis in cases where ischaemic stroke is suspected is crucial and there is a need to also improve the diagnostics of stroke mimics.

Education of paramedics and effective prehospital care can improve the overall quality of the provided stroke care. The possibility of teleconsultation increases the correct diagnosis of ischaemic stroke in the prehospital phase. The availability of stroke physician with combination of multimodal CT examination can further improve the rates of correct diagnosis of stroke and distinguished stroke mimics [40]. In case of negative multimodal CT examination on admission, 24 hours follow up of MR is recommended for stroke mimics exclusion [49, 48]. Misdiagnosis of stroke leads to an unnecessary use of stroke prevention medication, psychosocial problems and stigmatisation [40].

Our work is a brief overview of the importance of CT and CTP in stroke, this is not an extensive systematic review or metanalysis. Our own analysis was not performed, only the data from carried out analysis were comprised in our work.

CTP-based patient selection for recanalization treatment has led to extension of crude time windows for intravenous thrombolysis and mechanical thrombectomy. The use of automated software bridges the shortcomings of standard commercial software programs and enables the introduction of CTP into clinical practice. CTP appears to be a useful tool in detecting stroke mimics during the initial CT scan in acute patients with suspected AIS.

Other standardisation of perfusion criteria and imaging processing will help to widespread CTP to clinical practise. More evidence is needed for CTP in indication of reperfusion therapy in patient with large core infarct, treatment beyond 24 hours and futile reperfusion up to standardised time window.

DV has made substantial contributions to conception and design, has given final approval of the version to be published. OV has made substantial contributions to conception and design. PC has made substantial contributions to conception and design. KŠ has made substantial contributions to acquisition of data and interpretation of data. KD has made substantial contributions to acquisition of data and interpretation of data. MB has been involved in drafting the manuscript.

Not applicable.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.