Beyond ASPECTS: A Practical Guide to CT Perfusion Interpretation

Abstract

Computed tomography perfusion (CTP) imaging plays a pivotal role in the early evaluation of patients presenting with acute ischemic stroke (AIS), particularly by identifying candidates for endovascular thrombectomy. Accurate interpretation of CTP requires a structured approach that integrates technical understanding, clinical judgment, and recognition of the modality’s limitations. This review was prompted by real clinical challenges faced by the senior author and aims to provide both a theoretical foundation and practical guidance for interpreting CTP. Key concepts are illustrated through real clinical scenarios and corresponding annotated images. In addition to reviewing current AHA/ASA guidelines, we discuss institutional best practices and highlight challenging clinical scenarios in which CTP can significantly influence treatment decisions. This article aims to equip clinicians with the knowledge and tools needed for consistent and effective use of CTP.

Introduction

Indication for CTP

CT perfusion (CTP) imaging plays a critical role in the evaluation of patients with suspected acute ischemic stroke (AIS), particularly for identifying candidates for potentially curative mechanical thrombectomy (MT) within the extended time window of 6 to 24 h from the last known well. The major randomized controlled trials that demonstrated the efficacy of MT in this extended time frame, DEFUSE 3 (6–16 h) ¹ and DAWN (6–24 h), ² used perfusion-based imaging criteria to select patients for intervention (see Table 1). These criteria have since been incorporated into the most recent American Stroke Association/American Heart Association (ASA/AHA) guidelines published in 2019. ⁹ As a result, accurate assessment and interpretation of CTP are essential for optimal management of AIS patients with large vessel occlusion presenting in the extended time window. Despite its clinical importance, CTP remains underutilized, primarily due to limitations in resources and infrastructure. ¹⁰

Table 1.

Perfusion-based inclusion criteria for DEFUSE3 and DAWN.

Trial name/Inclusion criteria	DEFUSE3	DAWN
LVO definition	ICA or MCA M1	ICA or MCA M1
Time from LKW	6–16 h	6–24 h
Penumbra definition	Mismatch ratio between total critical hypoperfusion (Tmax >6 s) and core infarct (rCBF <30) >1.8	Mismatch between clinical deficit and core infarct
Penumbra size	Penumbra volume >15 mL	N/A
Core infarct definition	rCBF <30% as compared to contralateral hemisphere	rCBF <30% as compared to contralateral hemisphere
Core infarct size	<70 mL	< One third of MCA territory
Postprocessing CTP software	RAPID	RAPID

Perfusion parameters

In principle, CTP maps the contrast attenuation of brain tissue over time as a surrogate for tissue perfusion. The three most important CTP parameters in clinical practice are CBV (cerebral blood volume), CBF (cerebral blood flow), and Tmax (time to maximum). CBV is defined as the volume of blood in a given region of brain tissue, CBF is the volume of blood passing through a given region of brain tissue per unit time, ¹¹ and Tmax represents the contrast arrival delay to a tissue of interest. Ischemia will cause delayed contrast bolus arrival, as blood flow is rerouted from occluded vessels to vasodilated roundabout leptomeningeal collaterals. The result is elevated Tmax (Figure 1) and decreased CBF. By contrast, as autoregulatory mechanisms decrease cerebrovascular resistance to maintain CBF in early ischemia (i.e., critically hypoperfused tissue that has not yet infarcted), CBV is preserved or increased (Figure 2). This distinction is clinically relevant during qualitative evaluation of CTP maps, as decreased CBF is a better marker of tissue hypoperfusion, while decreased CBV more accurately represents the core infarct.

Figure 3.

A large field of view that includes the torcula Herophili and as much supratentorial brain as possible and excludes the ocular lens is selected for scanning.

Figure 1.

This patient presented with a left MCA occlusion. No contrast is seen in the left MCA on arterial phase (left); however, there is late filling of the left MCA territory on capillary phase (right) due to circuitous leptomeningeal collateralization to perfuse the left MCA territory, as the rest of the arteries have already washed out. In fact, there was continued delayed bolus arrival in this territory well into the venous phase (not shown). This explains why Tmax is elevated in the left MCA territory.

Figure 2.

In this patient, there is increased CBV in the left frontoparietal lobe and decreased CBV in the left posterior basal ganglia (left). The region of increased CBV is associated with Tmax elevation (right) and corresponds to the penumbra, while the region of decreased CBV represents the core infarct. Penumbral increase in CBV is attributable to autoregulatory mechanisms, however, CBV can also be normal in the penumbra. By contrast, CBV is always decreased in the core infarct.

Core and penumbra

A core principle of CTP imaging is the identification of the ischemic penumbra—a region of hypoperfused but potentially salvageable tissue that surrounds the core infarct. The penumbra is thought to correspond to CBF values between 10 and 20 mL/100 g/min, reflecting tissue with impaired function but temporarily preserved viability. ¹² This zone represents the primary target for reperfusion therapies, making CTP essential for guiding MT decisions.

On CTP, the penumbra is defined indirectly as the mismatch between the volume of total critically hypoperfused tissue and the volume of core infarct. Typically, the former is identified as tissue with Tmax >6 s, and the latter as tissue with relative CBF (rCBF) <30% compared to the contralateral hemisphere. ¹³ These thresholds were established by the landmark randomized controlled trials DEFUSE 3 and DAWN, which demonstrated the benefit of MT in the extended treatment window.

However, it is important to note that threshold values can vary depending on the CTP postprocessing software used. Therefore, radiologists should be familiar with their institution’s specific software parameters and default thresholds to ensure accurate interpretation and clinical decision-making.

Acquisition

Optimizing the scan

There are many technical parameters involved in CTP acquisition, such as z-axis coverage, imaging technique, temporal resolution (frequency of imaging), and length of the scan. ¹⁴ As optimal parameters vary somewhat between CTs, the overseeing physician should be familiar with their vendor’s recommendations. Nevertheless, ACR, ASNR, and SPR have published joint practice guidelines that include general acquisition recommendations listed below (Table 2). ¹⁴ Also, a sample technical setup from our institution is described in the appendix (Appendix 1) and a series of recommendations from the literature and vendors is provided below (Table 3). One additional recommendation from the experience of our senior author is that IV contrast should be injected via right upper extremity veins, as opposed to left upper extremity veins, to avoid potential outflow narrowing of the left brachiocephalic vein as it’s compressed between the aorta and manubrium. This would cause dense IV contrast reflux into the venous system, stretch out the bolus and decrease CTP image quality (Figure 4).

Table 2.

Select CTP acquisition recommendations per ACR-ASNR-SPR practice guidelines.

CTP step	Recommendation
Contrast injection	Power injector with dual-bore saline-chase injection pumps
	Contrast volume at least 40 mL
	Injection rate of at least 4 mL/s in adults (smaller in children)
	Saline chase of at least 15–20 mL
Acquisition time	50–70 s
Radiation dose	70–90 kVp and 100–200 mAs
Imaging	At least two baseline images before arrival of contrast, normally achieved with starting the imaging 4 seconds after contrast injection (should be delayed in patients with poor cardiac output)

Table 3.

Select CTP acquisition recommendations from the literature and vendors.

Recommendation	Source
Temporal resolution should be no greater than 3 seconds	3
Scan duration should be between 60 and 73 s	According to a study of 70 CTP scans, a scan duration of 60.7 s was sufficient for 90% of scans and 72.9 s for 100% of scans 4
Scan duration should be at least 65 s	RAPID 5
The highest available contrast concentration should be used	6
Saline chaser should be 40 mL	7
The patient’s head should be tucked prior to imaging to avoid ocular lens exposure (Figure 3)	8

Figure 4.

Contrast reflux (left) as a result of the left brachiocephalic vein being pinched off between the aorta and manubrium (right). The contrast bolus is stretched, and the CTA/CTP image is degraded. Injecting contrast through a right upper extremity vein circumvents this problem.

Postprocessing

After the CTP scan is performed, the software selects an artery for the arterial input function (AIF), usually the A2 branch of the ACA or the M1 branch of MCA, and a vein for the venous output function (VOF), usually the superior sagittal sinus or the torcula Herophili. The VOF is used to normalize the AIF data, which would otherwise be prone to partial volume effects ¹⁵ , while the AIF serves as a reference for measuring contrast passage through brain tissue. Postprocessing algorithms are then applied to generate the time-attenuation curves, parametric CTP color maps, and quantitative determination of core infarct versus total critically hypoperfused tissue volume.

Interpretation

A graphical overview of the following workflow is presented in Figure 5, which summarizes the main steps described in this section.

Figure 5.

Summary of CT perfusion interpretation workflow.

Reviewing the output

After postprocessing software has been applied to appropriately acquired CTP images, the interpreting physician should be provided with graphed time-attenuation curves and CTP parameter colormaps (Figure 6). It is important to review source CT images to evaluate for motion artifact which might alter CTP parameters (Figure 7). Also, the morphology of the time-attenuation curves should be inspected. The AIF curve should peak earlier and lower than the VOF curve and both should downslope and level off until new baselines are established (Figure 8). ¹⁶ Abnormal appearing curves can limit CTP evaluation; sometimes, this can be corrected by using a different postprocessing software or by manually selecting a better AIF and VOF (Figure 9).

Figure 6.

Perfusion parameter colormaps are generated by postprocessing software, in this case demonstrating decreased CBF (revealing hypoperfusion state; top left), decreased CBV (corresponding to core infarct volume; top right), and increased Tmax (corresponding to total critical hypoperfused tissue volume; bottom left) in the left MCA territory. An associated DWI image (bottom right) localized the infarct to the left gangliocapsular region with decreased CBV.

Figure 7.

Motion artifact due to head motion near the end of the scan leading to global perfusion artifact.

Figure 8.

Sufficient length of acquisition time is essential for ensuring that the time-attenuation graph is captured until a new baseline is established. Compare sufficient length (left) to insufficient length (right). Note that the new baseline is nonzero due to the recirculation of contrast.

Figure 9.

Global cerebral artifact on CBF and Tmax due to poor AIF and VOF. Note the faulty TOF curve with lack of normal AIF and VOF curves, generating abnormal appearing colormaps (top). After reprocessing on another software, a normal appearing TOF curve was generated along with diagnostic appearing CBV, CBF, and Tmax color maps (bottom).

Quantitative assessment

In addition to time-attenuation curves and parametric perfusion maps, modern CTP postprocessing software generates axial images that highlight regions of core infarct (CBF less than 30%) and critically hypoperfused tissue (Tmax greater than 6 s). The corresponding volumes of these regions are also calculated. A key output is the mismatch volume and ratio between these two regions (Figure 10), which estimates the ischemic penumbra.

Figure 10.

CTP postprocessing software provides volumetric data for ischemic penumbra and core infarct, along with color coded corresponding images. The core infarct (purple) is defined by neuronal death with irreversible loss of function, corresponding to area of restricted diffusion and rCBF on CTP <30% (per RAPID). The penumbra (yellow) is defined by ischemic but salvageable tissue which corresponds to the difference between the volume of core infarct and the volume of critically hypoperfused tissue (green; defined by RAPID as Tmax > 6 s).

Qualitative assessment

In the absence of reliable quantitative postprocessing software, qualitative interpretation of CTP parametric color maps remains a valuable alternative. Among these, the CBF map is particularly important due to its high sensitivity to perfusion abnormalities. Hypoperfusion appears as areas of decreased CBF, while hyperperfusion, such as luxury perfusion or periictal hyperperfusion (discussed below), presents as increased CBF. However, visual assessment of the CBF map alone cannot distinguish between core infarct and ischemic penumbra, as both demonstrate reduced CBF.

In contrast, the CBV map is more specific: it typically shows decreased values only in regions of core infarction (Figure 2). For this reason, the CBV map can be used to approximate core infarct volume and may be considered a surrogate for diffusion-weighted imaging (DWI), the current gold standard for core infarct estimation. ¹⁷

Conversely, the Tmax map is the preferred parameter for assessing the volume of total critically hypoperfused tissue, defined by markedly delayed perfusion (Tmax >6 s). The ischemic penumbra can then be estimated by the mismatch between the area of elevated Tmax and the area of reduced CBV (Figure 11).

Figure 11.

In this patient presenting with acute left MCA syndrome, the non-contrast CT (top left) was negative, but the concurrent CTA demonstrated a distal left M1 occlusion (top right). CT perfusion demonstrated a small region of core infarct (i.e., decreased CBV) in the left frontoparietal lobe (bottom left) surrounded by a much larger region of critically hypoperfused tissue (i.e., increased Tmax) (bottom right), a mismatch that indicates a large salvageable penumbra.

Pearls and pitfalls

Always review the non-contrast head CT

It is essential to recognize that CTP reflects only the hemodynamic characteristics of contrast bolus arrival and distribution but does not provide direct insight into the structural integrity of brain tissue as MRI does. As a result, therapeutic or spontaneous recanalization of a previously occluded major vessel may normalize perfusion parameters on CTP, thereby masking an established infarct (Figure 12). In such cases, the infarct may go undetected because contrast arrival is no longer delayed. Similarly, delayed recruitment of collaterals in the area of an already formed infarct can render core infarct undetectable by quantitative CTP thresholds (Figure 13). Therefore, always evaluate the head CT to exclude an already formed infarct in the region labeled penumbra, as CT hypodensity more accurately reflects irreversible tissue injury.

Figure 12.

There is a small infarct at the left caudate nucleus (red arrow) and a larger infarct in the left fronto-insular and anterior gangliocapsular regions (blue arrow) on head CT that are not reported in the CTP findings (left). The smaller infarct is likely not detected by RAPID because its size is below the resolution of quantitative analysis, while the larger infarct, corresponding to an area of increased Tmax (>6 s), is likely masked on CTP due to delayed recruitment of collateral vessels.

Figure 13.

There is a large right MCA territory infarction seen on non-contrast CT (right) performed after right M1 reperfusion. However, on CTP this appears as a much smaller region of core infarct on the rCBV map (bottom left). This is because reperfusion has allowed for contrast to recirculate in the right M1 territory, masking the underlying, irreversible tissue infarction. Moreover, there is luxury perfusion of the right operculum and basal ganglia (corresponding to elevated rCBF), superimposed on and obscuring a large portion of the infarct on the CBF colormap. These findings underscore the importance of reviewing the non-contrast head CT alongside CTP findings.

Another important caveat is that some quantitative postprocessing software may fail to detect small infarcts, especially those below the software’s sensitivity threshold. For instance, the RAPID platform only identifies infarcts larger than 3 mL on its quantitative mismatch maps (Figure 12). ¹⁸ Additionally, some software algorithms are intentionally designed to suppress chronic infarcts from the quantitative analysis, in order to prevent unnecessary exclusion of patients from potentially beneficial interventions (Figure 14). Correlation with non-contrast head CT can therefore provide valuable information not already incorporated in the CTP output.

Figure 14.

Bilateral chronic frontal infarcts (left) are not included in CBF (middle) or Tmax color maps (right) due to automatic exclusion of chronic infarcts that approach density of CSF by RAPID algorithm.

Early window strokes

According to the ASA/AHA guidelines, CTP is not recommended in the early time window of AIS (less than 6 h from last known well) ¹² for patient selection for MT. Applying the extended time window threshold (rCBF <30%) during this early period can lead to overestimation of core infarct size, a phenomenon known as the “ghost infarct core” (Figure 15). In one study of 36 CTP scans performed within 90 min of last known well, 17% significantly overestimated core infarct volume when using extended window thresholds. ¹⁹

Figure 15.

CTP performed in the early time window before 6 h of last known well demonstrates a large core infarct. Post thrombectomy (L ICA occlusion) MRI was normal (not shown), indicating ghost core infarct.

Despite these limitations, clinical experience suggests that CTP may still offer value in certain early window scenarios, some of which will be discussed below. However, the uncertain clinical benefit of CTP in the early time window must be carefully balanced against the potential risks of additional radiation exposure and treatment delays.

Subocclusive thrombi

Subocclusive thrombi are often difficult to visualize on CTA, ²⁰ yet they may benefit from MT. ²¹ In equivocal cases, CTP can be diagnostic by revealing a corresponding region of significant hypoperfusion (Figure 16).

Figure 16.

This patient presented with left upper extremity weakness for 3 days and exam was notable for right gaze preference and left upper extremity numbness. There is a subocclusive thrombus in the right distal M1. It is difficult to appreciate on the CTA (bottom right), but perfusional alterations are readily appreciated on CTP (top right). Digital subtraction angiography (DSA) images before (bottom left) and after successful TICI 3 embolectomy (bottom center) are included as well.

Chronic steno-occlusive disease

In patients with chronic steno-occlusive disease, differentiating acute ischemia from baseline perfusional alterations presents a diagnostic challenge. In these cases, Tmax may appear critically elevated (i.e., >6 s), reflecting chronic delayed contrast arrival rather than acute, critically reduced perfusion. A classic example of this pattern is seen in moyamoya disease, where chronic vascular changes can result in prolonged Tmax without impending acute infarction (Figure 17).

Figure 17.

Large right sided Tmax prolongation (right) due to chronic right ICA occlusion (red arrow) in a patient with moyamoya. The patient was asymptomatic. This can sometimes be seen with chronic steno-occlusive disease and doesn’t mean that the patient will develop an infarct in this area. The only way to estimate future infarct risk in this setting is with a cerebrovascular reserve study.

Despite these challenges, CTP can often differentiate acute ischemia from the underlying chronic process by characterizing the distribution of perfusional alterations. Chronic steno-occlusive disease sometimes presents with absent, minimal, or scattered perfusional abnormalities (Tmax >6 s) (Figure 18). By contrast, acute ischemia in the setting of an acute large vessel occlusion (LVO) will present as large, confluent, and severe abnormalities (Figure 11).

Figure 18.

There is a right M1 occlusion seen on CTA (left) with only small and scattered areas of corresponding Tmax elevation (right). An acute occlusion would more likely present as larger MCA territory Tmax prolongation, increasing diagnostic confidence that this represents chronic steno-occlusive disease.

It is important to note, however, that future ischemic risk in the setting of chronic steno-occlusive disease cannot be reliably assessed by static CTP alone and instead requires dedicated cerebrovascular reserve studies.

MeVO

Although MT is typically recommended only for the treatment of LVOs, such as ICA, M1, ⁶ vertebral, and basilar artery ²² occlusions, some centers treat medium vessel occlusions (MeVo), such as M2, M3, A1-A3, and P1-P3, with MT as well. ²³ Although MeVos can be difficult to detect on CTA, localized alterations to CTP parameters can be critical to their diagnosis (Figure 19, Figure 20). However, it should be noted that there is emerging data that questions the efficacy of MT in the setting of MeVo. ²⁴

Figure 19.

This patient has a large right M1 infarct, to which the patient’s clinical presentation was initially attributed. However, there is also critical hypoperfusion with elevated Tmax >6 s in the right ACA territory (left), leading the interpreting radiologist to reexamine the CTA and identify the MeVO involving a branch of the right ACA (right).

Figure 20.

This patient had a subtle left inferior M2 MeVO that was difficult to appreciate on CTA (left) but perfusional alterations were readily appreciated on CTP (right).

Abnormal hyperperfusion states

In contrast to the normally decreased CBF seen in acute ischemia, increased CBF can indicate an abnormal hyperperfusion state within a vascular territory. In the context of recent ischemia, this may represent luxury perfusion or reperfusion-hyperperfusion syndrome, a phenomenon characterized by increased, non-nutritive blood flow to previously ischemic, vasodilated tissue, out of proportion to metabolic demand. In such cases, CBF and CBV are elevated, while Tmax remains unchanged or even decreased (Figure 21, Figure 13). This atypical perfusion pattern can lead to misinterpretation by automated software, which may either suggest contralateral pathology ²⁵ or fail to detect an underlying ipsilateral infarct. A similar imaging appearance of CBF elevation may occur in the setting of peri-ictal hyperperfusion, although this is typically cortically based (Figure 22), in contrast to luxury perfusion, which can also involve the deep parenchyma.

Figure 21.

There is an acute infarct in the left gangliocapsular area (right). In this same region, the CBF is elevated and Tmax is decreased (left), consistent with hyperperfusion related to luxury perfusion.

Figure 22.

Ictal/post-ictal hyperperfusion in the left MCA/PCA territory in the setting of Todd’s paralysis. Incidentally noted is evidence of prior left temporal lobectomy. Importantly, there is a cortically based region of CBF/CBV elevation in the left temporal occipital MCA/PCA territory that demonstrates washed-out cortical veins on SWI (bottom right image). The latter is likely due to relatively decreased deoxyhemoglobin in this region due to periictal hyperperfusion out of proportion to metabolic demand.

Conclusion

While current AHA/ASA guidelines recommend integrating CTP into standard stroke protocols to establish eligibility for mechanical thrombectomy, CTP remains challenging to interpret and underutilized for a variety of scenarios. For example, CTP can clarify underlying perfusional abnormalities in periictal hyperperfusion, luxury perfusion, MeVOs, and subocclusive thrombi. Moreover, for appropriate interpretation of CTP, the reader must be aware of potential pitfalls and how to avoid them. In addition to reviewing the technical background, indications, and interpretive strategies for reading CTP, this review has highlighted its value in challenging cases and underscored how to avoid common pitfalls. Looking ahead, CTP utilization can be further optimized through the integration of artificial intelligence tools, standardization of postprocessing software, and multicenter validation of CTP thresholds.

Appendix 1. Sample Technical CT Perfusion Setup.

At our institution, we use the Somatom x.cite scanner (38.4 mm z-axis coverage, toggle/shuttle technique) and follow the manufacturer’s scanner-specific recommendations. Radiation is set at 200 mAs and 70 kV, and temporal resolution is set at 1.5 s for 27 cycles and then 3 s for 9 cycles, for a total of 67.5 s. Slice thickness and slice interval are both set at 10 mm. 40 mL of contrast is injected at a rate of 5 mL/s via the right upper extremity antecubital vein.

Abbreviation

AIS

acute ischemic stroke

MT

mechanical thrombectomy

CTP

CT perfusion

Tmax

time to maximum

AUC

area under the curve

AIF

arterial input function

VOF

venous output function

LVO

large vessel occlusion

MeVO

medium vessel occlusion

Consent for publication

The clinical images and case information used in this review are fully de-identified and originate from historical cases. The individuals cannot be re-identified, and no personal identifiers or sensitive details are included.

ORCID iDs

Raphael Miller https://orcid.org/0009-0003-8408-7480

Ryan Morasse https://orcid.org/0009-0009-3044-7296