Adjusting input arterial function to improve the accuracy of hypoperfusion assessment in computed tomography perfusion (CTP) for stroke: a preliminary study

Introduction

Despite advancements in prevention, ischemic stroke remains the second leading cause of death and the primary culprit behind long-term injury (1). Timely rescue of the ischemic penumbra can mitigate brain damage following acute ischemic stroke (AIS); otherwise, there is a risk of acute cerebral infarction. The significant mismatch of perfusion images determines the feasibility of endovascular treatment. However, the radiological characterization of the penumbra region tends to be overestimated due to the inaccurate inclusion of a large extent of tissue at negligible infarction risk, now commonly referred to as benign oligemia (2-4). This exaggeration in the degree of mismatch theoretically leads to overtreatment and unexpected complications.

Recent advancements in neuroimaging have positioned accurate penumbra characterization as a critical determinant in optimizing reperfusion treatment decision-making. Although salvageable penumbral tissue remains the primary biomarker for thrombolytic intervention eligibility, current clinical practice lacks standardized quantification protocols. Multi-parametric imaging protocols now enable penumbral evaluation, including magnetic resonance imaging (MRI)-based diffusion-weighted imaging/perfusion-weighted imaging (DWI/PWI) mismatch, time until the residue function reaches its peak (Tmax), or mean transit time (MTT) in computed tomography perfusion (CTP), and positron emission tomography (PET) oxygen extraction fraction mapping. Emerging modalities, including arterial spin labeling (ASL), diffusion-derived vessel density (DDVD) quantification (5), and machine learning-based dynamic threshold algorithms, have enhanced the spatiotemporal resolution of penumbral evolution monitoring. Each neuroimaging modality exhibits characteristic trade-offs in penumbral evaluation. Although MRI has the best accuracy in defining ischemic penumbra, CTP has gradually become the core tool for the screening of reperfusion therapy for AIS due to its high scanning efficiency, quantitative and intuitive parameters, and compatibility with emergency procedures.

CTP employs bolus-tracking technology to enable the quantitative assessment of cerebral hemodynamics. The arterial input function (AIF), which represents the concentration of contrast medium in the brain over time, is crucial for calculating hemodynamic parameters (6-9). It is imperative to identify the appropriate input artery and output vein in post-processing CTP images to derive AIF and venous output function (VOF) by monitoring the initial entry of contrast medium into the brain. Application of deconvolution algorithms to CTP data enables generation of high-resolution parametric maps, including cerebral blood volume (CBV), cerebral blood flow (CBF), and Tmax (10,11). These quantitative outputs facilitate differentiation between salvageable penumbral tissue and irreversibly damaged ischemic cores. Radiologically, the ischemic penumbra manifests as a “mismatch” region between hypoperfused tissue and the infarction core, thus CTP measurement of hypoperfused tissue delineates the outer boundary of the ischemic penumbra. Despite CTP’s established reliability in infarct core delineation, there has been limited focus within most studies and post-processing tools on the precision of hypoperfusion assessment.

Many factors affect the accuracy of CTP in assessing hypoperfused tissues. First, acquisition protocols balancing temporal resolution and scan duration critically affect hemodynamic modeling. Although increased sampling rates reduce motion artifacts, abbreviated scan windows risk truncating the time-density curve (TDC). Second, post-processing platforms introduce variability through divergent processing strategies, including computational models and perfusion threshold selections, systematically influencing hypoperfusion volumetry and spatial localization (8,12,13). Third, the choice of AIF position affects the contrast medium transit time, which in turn profoundly impacts parameter quantification. Different input arteries can be chosen as the AIF to obtain various quantitative perfusion results, but the optimal location of the input artery is still a matter of debate (14). Notably, the Tmax value exhibits significant variability depending on the anatomical positioning of the input artery.

Contemporary CTP post-processing platforms conventionally position the AIF within contralateral middle cerebral artery (MCA) or internal carotid artery (ICA) segments. However, this approach has deviations when it involves hypoperfusion tissue with collateral circulation compensation. Specifically, contrast medium delivery through collateral pathways inherently prolongs the Tmax measurement due to delayed and attenuated blood flow dynamics. Consequently, this limitation may lead to an overestimation of Tmax values and subsequent ischemic penumbra delineation. From a hemodynamic perspective, substantial discrepancies exist between the TDC of the MCA and adjacent parenchymal arterioles supplying hypoperfused tissues. Emerging evidence suggests that utilizing proximal arterial inputs better characterizes regional cerebral perfusion patterns (regardless of collateral circulation status), as these vessels demonstrate more direct hemodynamic coupling with local tissue perfusion demands (15).

This investigation pioneers an AIF selecting strategy, targeting arterioles adjacent to hypoperfused tissue during CTP post-processing. We aimed to modify the AIF during the post-processing of CTP images to enhance the reliability of hypoperfusion area assessment. The study assessed the reliability and accuracy of distinguishing hypoperfused tissue from benign oligemia by analyzing the volume and spatial consistency of hypoperfused tissue. We endeavored to improve the precision of cerebral perfusion image analysis to augment its efficacy in guiding clinical treatment decisions. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1920/rc).

Methods

Patients

Patients diagnosed with AIS between November 2020 and July 2022 in Fujian Medical University Union Hospital were retrospectively collected. The patient inclusion criteria were as follows: (I) clinical diagnosis of AIS; (II) absence of hemorrhage presented on non-contrast computed tomography (NCCT); (III) unilateral occlusion of the ICA terminus or MCA confirmed by CTA; (IV) MRI examination within 10 days after CTP examination; and (V) presence of tissue with Tmax >6 seconds in CTP maps. The exclusion criteria were as follows: (I) motion or metal artifacts; (II) distinctly abnormal arterial and venous peak values; (III) receipt of thrombolytic or endovascular treatment; and (IV) severe hemorrhage in follow-up MRI. Baseline clinical characteristics and demographic information were retrieved from the medical records.

Standard protocol approvals, registrations, and patient consent

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Fujian Medical University Union Hospital Medical Ethics Committee (approval No. 2022KY170). The requirement for informed consent was waived for this retrospective study.

Image acquisition

All computed tomography (CT) scans were performed by a 256-slice CT scanner equipped with a 160-mm wide detector (Revolution, GE Healthcare, Milwaukee, WI, USA). All patients underwent NCCT, head/neck computed tomography angiography (CTA), and CTP examinations. The scanning parameters for helical NCCT were as follows: tube voltage 120 kVp, tube current 350 mAs, thickness 5 mm. The CTP protocol was as follows: tube voltage 80 kVp, tube current 200 mAs, slice thickness 5 mm, 256×0.625 mm detector collimation, and whole-brain coverage in the Z-axis. CTP scanning began after a delay of 5 seconds from contrast injection: first, 12 consecutive phases were acquired with a temporal resolution of 2 seconds, after 2.5 seconds of delays, 8 consecutive phases were acquired with a temporal resolution of 3 seconds. A total of 40 mL of iodinated contrast agent (400 mg I/mL Iomeron, Bracco Imaging, Shanghai, China) was intravenously injected with a flow rate of 4 mL/s, followed by a 30-mL saline flush with the same flow rate.

AIF selection

The selection of AIFs was automatically performed and placed in the (I) normal large artery (AIF_NLA); and (II) collateral circulation arteriole of MCA adjacent to the ischemic lesion (AIF_AIL). The threshold technique was used for segmenting blood vessels, and two rounds of automatic selection of input arteries were conducted for different purposes. In the first round, AIF_NLA was automatically designated as normal intracranial arteries, then Tmax was computed. The perfusion region where Tmax >6 seconds was identified as the region of interest (ROI). In the next round, collateral circulation arterioles adjacent to the ROI were automatically selected as AIF_AIL. To maintain consistency in quantitative values and prevent variations in venous ROI selection from causing secondary changes, we kept the VOF fixed at a specific point on the inferior aspect of the superior sagittal sinus or sinus confluence.

Final infarct volume (FIV) analysis

The ITK-SNAP software (version 4.6.1, www.itksnap.org) was utilized to outline the regions exhibiting high signal intensity in each axial slice of follow-up MRI DWI B=800 s/mm², which served as a reference for the final infarction and subsequently for the calculation of its volume (16).

Data postprocessing

The CTP images underwent post-processing using the Brain Perfusion application (CTPDoc, ShuKun Technology, Beijing, China), which employed a delay-insensitive algorithm and automatically processed CTP data. This resulted in the generation of concentration-time curves for all potential artery candidates in the ROI, with the selection of the optimal AIF based on criteria such as peak moment, peak value, number of crests, and half-height width. Singular value decomposition was utilized to derive CBF, CBV, MTT, Tmax, and other parameters (Appendix 1). Ischemic core was diagnosed when relative CBF was <30% compared to the contralateral hemisphere for both AIF_NLA and AIF_AIL. Hypoperfusion was defined as a relative Tmax delay of >6 seconds for AIF_NLA. For AIF_AIL, a comparison of relative Tmax delays >4 seconds, >6 seconds, and >8 seconds was conducted to determine the optimal threshold value for identifying hypoperfusion regions.

Two observers, one of whom had at least 5 years of experience in neurovascular imaging, were not provided with any additional radiologic or clinical information. If there were any disagreements, the majority opinion was determined after consulting a third stroke neurologist. The brain was extracted from CT or MRI data using the skull and scalp surface extraction functions of Brain Extraction Tool (BET, University of Oxford, Oxford, UK) (17). The DWI image was then co-registered to the baseline CTP using the registration function from ANTsPy (18).

Statistical analysis

The median and interquartile ranges were served to summarize the univariate distribution of metric variables. Descriptive analysis was conducted to identify baseline population, clinical, and imaging factors. Statistical analyses were performed using the statistical software SPSS 26.0 (IBM Corp., Armonk, NY, USA). All statistical tests were two-sided, with significance defined as P<0.05.

The comparison between the hypoperfusion regions of Tmax and the DWI lesions was conducted because, in the absence of reperfusion therapy, cellular necrosis in the ischemic penumbra can occur at a rapid rate of up to 10.1 mL/h, ultimately leading to the complete transformation of the penumbra into an irreversible infarct (19,20). Therefore, theoretically, the volume of hypoperfused tissue should correspond to the volume of the final infarction focus.

The mean difference, the absolute difference, and mean absolute errors (MAEs) were calculated and defined as FIV minus hypoperfusion volume. Bland-Altman analysis and Spearman’s rho analysis were conducted to assess the degree of agreement between FIVs and hypoperfused volumes for AIF_NLA and AIF_AIL, respectively. The Dice similarity coefficient, the mean Hausdorff distance (HD), the positive predictive value (PPV), the true negative rate (TNR), the false negative rate (FNR), and accuracy were calculated to assess spatial agreement between Tmax and DWI lesions. The mean HD represents the average of all minimum distances between two segmentations. A histogram was constructed for each perfusion parameter and patient to illustrate the distribution of perfusion parameter values for infarcted and non-infarcted voxels. The DWI lesion was classified as “positive”, whereas other voxels were categorized as “negative”. The PPV was utilized to evaluate the proportion of initial Tmax hypoperfused tissue that ultimately fell within the DWI lesion. The TNR represented the proportion of tissue labeled as non-infarcted by the threshold that did not eventually infarct. The FNR was defined as the proportion of tissue labeled by the threshold as benign oligemia, which eventually infarcted. Accuracy was defined as the total number of true “positive” and true “negative” voxels from Tmax divided by the summation of all voxels in the Tmax hypoperfusion lesions. The overlap between Tmax and DWI lesions was calculated using SciPy (21), and the spatial agreement was evaluated using both SciPy and scikit-image (22).

Data availability statement

The data are available from the corresponding author on reasonable request.

Results

We identified 157 acute stroke patients with unilateral occlusion of the ICA and MCA during the study period. Among them, 122 cases were excluded, mainly due to thrombolytic or endovascular treatment (n=78), low-quality CTP (n=3), no follow-up MRI (n=22), and severe hemorrhage in follow-up MRI (n=19). Therefore, analysis was conducted on 35 eligible patients, with 17 (48%) of them being male (Figure 1). The mean age of the patients was 73±11 years, with a median National Institutes of Health Stroke Scale (NIHSS) of 10 [interquartile range (IQR), 5–15] and a collateral circulation score of 3 (IQR, 2–4). There were 15 (42.86%) occlusions in the M1 segment, 12 (34.29%) in the M2 segment, and 8 (22.86%) in the ICA. Figure 2 displays summary maps generated with AIF_NLA, AIF_AIL, and follow-up MRI-DWI in a representative case of left hemisphere stroke.

Figure 1 Flowchart of the enrolled patients. AIS, acute ischemic stroke; CTP, computed tomography perfusion; ICA, internal carotid artery; MCA, middle cerebral artery; MRI, magnetic resonance imaging.

Figure 2 An illustration of a patient suffering from a left hemisphere stroke is provided. The CTP analysis results were calculated using the AIF_NLA and AIF_AIL protocols, respectively. The FIV was obtained from DWI. CTPDoc perfusion analysis revealed that the infarct tissue appeared red and the penumbra appeared green. The fusion image comprises the FIV (red) and hypoperfused volume calculated by AIF_NLA (green), and AIF_AIL (yellow). Hypoperfusion volume in AIF_NLA was 168.2 mL and in AIF_AIL was 95.3 mL, respectively. The FIV was 102.3 mL. Spatial consistency analysis indicated that the Dice coefficients of FIV and hypoperfusion calculated by AIF_NLA and AIF_AIL were 0.50 and 0.60, respectively. AIF, arterial input functions; AIL, adjacent to the ischemic lesion; CTP, computed tomography perfusion; DWI, diffusion-weighted imaging; FIV, final infarct volume; NLA, normal large artery.

Volumetric analysis

The median volumes of hypoperfusion in AIF_NLA were 117.90 mL (IQR, 59.63–157.80 mL), whereas those of 4-, 6-, and 8-seconds in the AIF_AIL were 52.38 mL (IQR, 27.35–98.20 mL), 39.89 mL (IQR, 15.05–75.81 mL), and 24.65 mL (IQR, 6.83–57.53 mL), respectively. The median FIVs were 60.20 mL (IQR, 28.73–120.70 mL).

Mean differences, absolute differences, and MAEs between FIVs and hypoperfusion volumes for AIF_NLA and AIF_AIL are indicated in Table 1. There were statistically significant differences in the mean and absolute difference among the four AIF groups (P<0.001). The MAEs with AIF_NLA were 55.66, and with AIF_AIL-4 s, AIF_AIL-6 s, and AIF_AIL-8 s were 20.26, 34.69, and 44.06, respectively. The median hypoperfusion volume was substantially greater in AIF_NLA than in AIF_AIL (P<0.001), and the distributions showed differences (Figure 3). The volumes of hypoperfusion calculated by AIF_AIL-4 s did not differ significantly from the FIVs (P=0.43), whereas the hypoperfusion volumes measured by three other AIF groups were found to be statistically different from the FIVs (P<0.001). AIF_AIL-4 s reduced the degree of hypoperfusion overestimation on admission CTP maps by approximately 30%.

Figure 3 Violin diagram of the FIVs on DWI and hypoperfusion volumes on CTP. ns, P>0.05; ***, P≤0.001. AIF, arterial input functions; AIL, adjacent to the ischemic lesion; CTP, computed tomography perfusion; DWI, diffusion-weighted imaging; FIV, final infarct volume; NLA, normal large artery.

Spearman plots demonstrated the agreement analysis between the FIVs and the volumes of hypoperfusion calculated by the four AIF groups, respectively (Figure 4). AIF_AIL-4 s exhibited the most excellent agreement [rho=0.90; 95% confidence interval (CI): 0.80–0.95], whereas AIF_NLA showed the poorest agreement (rho=0.49; 95% CI: 0.11–0.68). Based on Bland-Altman analysis, comparing the mean bias between the FIVs and the volumes of hypoperfusion calculated by the four AIF groups, respectively, it was found that the AIF_NLA had the highest mean bias (−42.14±55.60 mL), whereas the mean bias was smallest in the AIF_AIL-4 s (5.18±29.35 mL) (Figure 5).

Figure 4 The agreement analysis between the FIVs on DWI and hypoperfusion volumes on CTP. Spearman correlation analysis was performed between the FIVs and hypoperfusion volumes calculated by AIF_NLA (A), AIF_AIL-4 s (B), AIF_AIL-6 s (C), and AIF_AIL-8 s (D), respectively. Solid lines represent the fitted curve obtained by regression analysis between these two variables. The dotted lines represent 95% of the limits of agreement. AIF, arterial input functions; AIL, adjacent to the ischemic lesion; CTP, computed tomography perfusion; DWI, diffusion-weighted imaging; FIV, final infarct volume; NLA, normal large artery.

Figure 5 The difference analysis between the FIVs on DWI and hypoperfusion volumes on CTP. Bland-Altman plots were performed between the FIVs and hypoperfusion volumes calculated by AIF_NLA (A), AIF_AIL-4 s (B), AIF_AIL-6 s (C), and AIF_AIL-8 s (D), respectively. Solid lines demonstrate the mean difference between these two variables. The dotted lines represent 95% of the limits of agreement. AIF, arterial input functions; AIL, adjacent to the ischemic lesion; CTP, computed tomography perfusion; DWI, diffusion-weighted imaging; FIV, final infarct volume; NLA, normal large artery.

Spatial agreement analysis

The comparisons of Dice coefficients, mean HD, PPV, TNR, and FNR between FIVs and hypoperfusion volumes for AIF_NLA and AIF_AIL are displayed in Table 1. It can be seen that 4 seconds had the best spatial agreement in the AIF_AIL. The accuracy was 0.93 (95% CI: 0.91–0.96) for AIF_NLA, whereas it reached 0.96 (95% CI: 0.94–0.98) for AIF_AIL-4 s.

Discussion

This exploratory study sought to improve the precision of hypoperfusion detection in CTP image post-processing by optimizing the selection of AIF. Our findings suggest that utilizing AIF_AIL-4 s, rather than the traditional choice AIF_NLA, led to a reduction of approximately 30% in the overestimation rate of hypoperfusion volume, thus demonstrating enhanced accuracy. These results suggest that employing AIF_AIL for distinguishing between benign oligemia and hypoperfusion may offer significant reliability, potentially assisting in avoiding unnecessary endovascular interventions, optimizing resource allocation, and reducing overtreatment linked to adverse clinical outcomes. As Goyal (23) recommends, “Let us improve the science before changing clinical practice.”

Prolongation of Tmax is a well-established indicator of cerebral hypoperfusion, with the recommended threshold for clinical application being Tmax >6 seconds (24). However, insights from the DEFUSE trial (25) and further studies (26) suggest that establishing the Tmax threshold at approximately 4 seconds effectively captures the distal blood supply or watershed area of the tissue, aligning with our study’s objective of selecting AIF in the collateral circulation arteriole of the MCA adjacent to the ischemic lesion.

The findings of this study revealed that both AIF protocols overestimated the volume of CTP hypoperfused tissue, consistent with previous research (12,27,28). The overestimation of penumbra can be attributed to the overestimation of the hypoperfused region, mainly due to the lack of robust post-processing features in automated software (29). Previous studies have primarily focused on mathematical or statistical calculations to distinguish benign oligemia from salvageable penumbra. Peretz et al. analyzed the ischemic penumbra evaluated by relative mean transit time (rMTT) and concluded that relative cerebral blood flow (rCBF) was the best parameter with a threshold of 0.65 (28). We propose an advanced and clinically significant approach to automate the selection of AIF points based on the results of the original AIF protocol. The initial localization of AIF is primarily guided by referencing the normal large artery, typically the contralateral MCA to the occluded artery. Subsequently, automatic identification of collateral circulation arterioles adjacent to ischemic lesions is performed to optimize AIF.

Another potential reason for the overestimation of hypoperfusion volume could be attributed to findings from perfusion imaging studies, which suggest that oligemia represents the brain tissue that survives continuous artery occlusion. However, in clinical practice, approximately 90% of ischemic stroke lesions exhibit spontaneous recirculation within 7 days (30). Therefore, our study results based on continuous arterial occlusion calculations reflect the worst-case scenario of the lesion rather than the general clinical course. This may also partly explain why the AIF_AIL protocol still overestimates hypoperfusion volume.

The role of AIF in quantitative perfusion parameters is crucial, yet there remains a lack of consensus on the selection of input arteries. The results of this study found that the arterioles in the collateral circulation as input arteries could obtain more accurate perfusion imaging, aligning with the perspectives put forth by Calamante et al. and Ebinger et al. Calamante et al. proposed positioning the input artery as close to the ischemic lesion as possible to minimize contrast injection delay and spread between the measured area and the lesion (31). Ebinger et al. compared different MCA branches as input arteries on MRI before contrast agent injection and found that AIF based on MCA distal branches yielded higher SNR and better imaging quality for perfusion results (32). Niesten et al. concluded that ICA was the optimal choice for patients with severe MCA stenosis or occlusive stroke, although this study only focused on the intracranial arteries and ICA (11). According to van Osch et al., it is crucial to take into account the potential artifacts caused by the partial volume effect, as they could result in alterations to the AIF and subsequently have an impact on the measurement outcomes when utilizing small blood vessels as input arteries (33). The impact of the partial volume effect on the variability of AIF is undeniable; however, a 256-slice spiral Revolution CT was employed in this study. This CT offers broader perfusion coverage, significantly enhances imaging quality, and substantially mitigates the influence of the partial volume effect on the selection of small arteries as input arteries. Our study encompassed collateral circulation arterioles, and follow-up DWI lesions were used as references to compare the diagnostic efficacy of different AIFs by quantifying the accuracy of hypoperfusion lesion volume and location. Furthermore, previous studies have predominantly focused on perfusion parameters such as CBF, CBV, and TTP with limited evidence regarding whether different input arteries are accurate and effective in quantitatively measuring Tmax parameters. Our research is at the forefront of utilizing cerebral arterioles as input arteries in post-processing CTP for evaluating hypoperfusion tissue imaging, pushing the boundaries of exploration into the field of Tmax perfusion parameters, and enhancing research on ischemic penumbra accuracy in CTP images. Another benefit of this study lies in the complete automation of CTP post-processing, diminishing the need for reliance on post-processing technicians and enhancing the reproducibility of CTP.

Accurate assessment of hypoperfusion tissue volume and location is crucial for determining eligibility for vascular interventional procedures. Imaging findings, such as the ratio of hypoperfusion tissue to core infarction (referred to as “mismatch rate”) and penumbra volume, play a vital role in assisting neurologists in promptly assessing patient suitability for intravascular therapy. In the DEFUSE clinical study, the mismatch rate was used as the standard to screen AIS patients for intravenous thrombolysis or arterial interventional thrombolysis, with examples being 1.2 for DEFUSE 1 (25) and 1.8 for DEFUSE 2–3 (24,34). Meanwhile, an excessive treatment resulting from overestimation of the ischemic penumbra can lead to adverse complications such as secondary haemorrhagic transformation (HT), neurological dysfunction, and cerebral hyperperfusion syndrome. There has been widespread acknowledgment of the fact that the utilization of anticoagulants, thrombolytic agents, and endovascular procedures amplifies the incidence and intensity of HT. The majority of cases involving parenchymal hemorrhage after AIS leading to neurological deterioration in the acute phase are associated with the formation of a dense hematoma and its related mass effect—one of the feared complications stemming from excessive treatment (35). The incidence of HT in AIS patients following intravascular therapy ranges from 46% to 49.5% and is correlated with negative outcomes (disability and death) (36-38). Previous randomized controlled studies have demonstrated an increase in the occurrence of secondary symptomatic HT by approximately 2–7% following intravenous thrombolysis with r-PA compared to controls (39). Advanced non-invasive diagnostic imaging techniques provide more precise screening for AIS patients eligible for reperfusion therapy (40,41). Clinicians must carefully assess the potential hazards and benefits of treatment to enhance the proportion of patients receiving treatment who attain favorable results.

Our research is limited by several constraints. Firstly, we used MRI-DWI lesions within 10 days of onset as a reference for FIV. However, the potential increase in DWI lesion volume in patients without reperfusion or due to vasogenic edema at 3–5 days may lead to overestimation or underestimation of results. Furthermore, inherent misalignments in the registration of DWI with CTP, similar to echo planar image distortion, may lead to inaccurate spatial analysis in specific scenarios (Figure S1). Lastly, the sample size of our study is inadequate because most AIS patients with large artery occlusion have received reperfusion therapy. Future research endeavors will prioritize addressing these limitations through expanding the sample size and implementing prospective analysis.

Conclusions

The reliability of automatic evaluation of the hypoperfusion region by CTP image post-processing is considerably increased when using the distal branch of the MCA adjacent to the ischemic lesion as AIF, and the outside boundary of the ischemic penumbra is more precisely detected. As neuroimaging technology advances, it will be able to precisely pinpoint which AIS patients are candidates for endovascular therapies by switching from “clock time” to “tissue time” to correctly monitor blood flow in brain tissue in real-time.

Acknowledgments

The authors would like to express their gratitude to Fujian Medical University Union Hospital (Department of Radiology and Neurology) for granting permission, as well as to the hardworking medical officers who provided support during the study.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1920/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1920/dss

Funding: This work was supported by Fujian Provincial Health Technology Project, China (No. 2020QNB018).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1920/coif). Y.H. is from Shukun (Beijing) Technology Co., Ltd. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Medical Ethics Committee of Fujian Medical University Union Hospital (No. 2022KY170). Informed consent was waived for this retrospective study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Neurol 2021;20:795-820. [Crossref] [PubMed]
2. Desai SM, Ortega-Gutierrez S, Sheth SA, Farooqui M, Lopez-Rivera V, Zevallos C, Salazar-Marioni S, Quispe-Orozco D, Abdelkhaliq R, Tonetti DA, Jovin TG, Jadhav AP. Clinically Approximated Hypoperfused Tissue in Large Vessel Occlusion Stroke. Stroke 2021;52:2109-14. [Crossref] [PubMed]
3. Wu L, Wu W, Tali ET, Yuh WT. Oligemia, Penumbra, Infarction: Understanding Hypoperfusion with Neuroimaging. Neuroimaging Clin N Am 2018;28:599-609. [Crossref] [PubMed]
4. Nomani AZ, Kamtchum Tatuene J, Rempel JL, Jeerakathil T, Winship IR, Khan KA, Buck BH, Shuaib A, Jickling GC. Association of CT-Based Hypoperfusion Index With Ischemic Core Enlargement in Patients With Medium and Large Vessel Stroke. Neurology 2021;97:e2079-87. [Crossref] [PubMed]
5. Chen JQ, Li CY, Wang W, Yao DQ, Jiang RF, Wáng YXJ. Diffusion-derived vessel density (DDVD) for penumbra delineation in brain acute ischemic stroke: initial proof-of-concept results using single NEX DWI. Quant Imaging Med Surg 2024;14:9533-42. [Crossref] [PubMed]
6. Calamante F. Arterial input function in perfusion MRI: a comprehensive review. Prog Nucl Magn Reson Spectrosc 2013;74:1-32. [Crossref] [PubMed]
7. Campbell BC, Christensen S, Levi CR, Desmond PM, Donnan GA, Davis SM, Parsons MW. Cerebral blood flow is the optimal CT perfusion parameter for assessing infarct core. Stroke 2011;42:3435-40. [Crossref] [PubMed]
8. Lin L, Bivard A, Krishnamurthy V, Levi CR, Parsons MW, Whole-Brain CT. Perfusion to Quantify Acute Ischemic Penumbra and Core. Radiology 2016;279:876-87. [Crossref] [PubMed]
9. Soltanpour M, Boulanger P, Buck B. CT Perfusion Map Synthesis from CTP Dynamic Images Using a Learned LSTM Generative Adversarial Network for Acute Ischemic Stroke Assessment. J Med Syst 2024;48:37. [Crossref] [PubMed]
10. Thijs VN, Somford DM, Bammer R, Robberecht W, Moseley ME, Albers GW. Influence of arterial input function on hypoperfusion volumes measured with perfusion-weighted imaging. Stroke 2004;35:94-8. [Crossref] [PubMed]
11. Niesten JM, van der Schaaf IC, Riordan AJ, de Jong HW, Mali WP, Velthuis BK. Optimisation of vascular input and output functions in CT-perfusion imaging using 256(or more)-slice multidetector CT. Eur Radiol 2013;23:1242-9. [Crossref] [PubMed]
12. Rava RA, Snyder KV, Mokin M, Waqas M, Zhang X, Podgorsak AR, Allman AB, Senko J, Shiraz Bhurwani MM, Hoi Y, Davies JM, Levy EI, Siddiqui AH, Ionita CN. Assessment of computed tomography perfusion software in predicting spatial location and volume of infarct in acute ischemic stroke patients: a comparison of Sphere, Vitrea, and RAPID. J Neurointerv Surg 2021;13:130-5. [Crossref] [PubMed]
13. Pennig L, Thiele F, Goertz L, Laukamp KR, Perkuhn M, Kabbasch C, Schlamann M, Fink GR, Borggrefe J. Comparison of Accuracy of Arrival-Time-Insensitive and Arrival-Time-Sensitive CTP Algorithms for Prediction of Infarct Tissue Volumes. Sci Rep 2020;10:9252. [Crossref] [PubMed]
14. Calamante F, Gadian DG, Connelly A. Quantification of perfusion using bolus tracking magnetic resonance imaging in stroke: assumptions, limitations, and potential implications for clinical use. Stroke 2002;33:1146-51. [Crossref] [PubMed]
15. Konstas AA, Goldmakher GV, Lee TY, Lev MH. Theoretic basis and technical implementations of CT perfusion in acute ischemic stroke, part 1: Theoretic basis. AJNR Am J Neuroradiol 2009;30:662-8. [Crossref] [PubMed]
16. Cereda CW, Christensen S, Campbell BCV, Mishra NK, Mlynash M, Levi C, Straka M, Wintermark M, Bammer R, Albers GW, Parsons MW, Lansberg MG. A benchmarking tool to evaluate computer tomography perfusion infarct core predictions against a DWI standard. J Cereb Blood Flow Metab 2016;36:1780-9. [Crossref] [PubMed]
17. Smith SM. Fast robust automated brain extraction. Hum Brain Mapp 2002;17:143-55. [Crossref] [PubMed]
18. Avants BB, Tustison N, Song G. Advanced normalization tools (ANTS). Insight J 2009;2:1-35.
19. Kaufmann AM, Firlik AD, Fukui MB, Wechsler LR, Jungries CA, Yonas H. Ischemic core and penumbra in human stroke. Stroke 1999;30:93-9. [Crossref] [PubMed]
20. Guenego A, Mlynash M, Christensen S, Kemp S, Heit JJ, Lansberg MG, Albers GW. Hypoperfusion ratio predicts infarct growth during transfer for thrombectomy. Ann Neurol 2018;84:616-20. [Crossref] [PubMed]
21. Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods 2020;17:261-72. [Crossref] [PubMed]
22. van der Walt S, Schönberger JL, Nunez-Iglesias J, Boulogne F, Warner JD, Yager N, Gouillart E, Yu T. scikit-image contributors. scikit-image: image processing in Python. PeerJ 2014;2:e453. [Crossref] [PubMed]
23. Goyal M, Menon BK, Derdeyn CP. Perfusion imaging in acute ischemic stroke: let us improve the science before changing clinical practice. Radiology 2013;266:16-21. [Crossref] [PubMed]
24. Albers GW, Marks MP, Kemp S, Christensen S, Tsai JP, Ortega-Gutierrez S, et al. Thrombectomy for Stroke at 6 to 16 Hours with Selection by Perfusion Imaging. N Engl J Med 2018;378:708-18. [Crossref] [PubMed]
25. Albers GW, Thijs VN, Wechsler L, Kemp S, Schlaug G, Skalabrin E, Bammer R, Kakuda W, Lansberg MG, Shuaib A, Coplin W, Hamilton S, Moseley M, Marks MP. Magnetic resonance imaging profiles predict clinical response to early reperfusion: the diffusion and perfusion imaging evaluation for understanding stroke evolution (DEFUSE) study. Ann Neurol 2006;60:508-17. [Crossref] [PubMed]
26. Straka M, Albers GW, Bammer R. Real-time diffusion-perfusion mismatch analysis in acute stroke. J Magn Reson Imaging 2010;32:1024-37. [Crossref] [PubMed]
27. Sasaki M, Kudo K, Honjo K, Hu JQ, Wang HB, Shintaku K. Prediction of infarct volume and neurologic outcome by using automated multiparametric perfusion-weighted magnetic resonance imaging in a primate model of permanent middle cerebral artery occlusion. J Cereb Blood Flow Metab 2011;31:448-56. [Crossref] [PubMed]
28. Peretz S, Orion D, Last D, Mardor Y, Kimmel Y, Yehezkely S, Lotan E, Itsekson-Hayosh Z, Koton S, Guez D, Tanne D. Incorporation of relative cerebral blood flow into CT perfusion maps reduces false 'at risk' penumbra. J Neurointerv Surg 2018;10:657-62. [Crossref] [PubMed]
29. Scalzo F, Nour M, Liebeskind DS. Data science of stroke imaging and enlightenment of the penumbra. Front Neurol 2015;6:8. [Crossref] [PubMed]
30. Kidwell CS, Jahan R, Gornbein J, Alger JR, Nenov V, Ajani Z, Feng L, Meyer BC, Olson S, Schwamm LH, Yoo AJ, Marshall RS, Meyers PM, Yavagal DR, Wintermark M, Guzy J, Starkman S, Saver JL. MR RESCUE Investigators. A trial of imaging selection and endovascular treatment for ischemic stroke. N Engl J Med 2013;368:914-23. [Crossref] [PubMed]
31. Calamante F, Gadian DG, Connelly A. Delay and dispersion effects in dynamic susceptibility contrast MRI: simulations using singular value decomposition. Magn Reson Med 2000;44:466-73. [Crossref] [PubMed]
32. Ebinger M, Brunecker P, Jungehülsing GJ, Malzahn U, Kunze C, Endres M, Fiebach JB. Reliable perfusion maps in stroke MRI using arterial input functions derived from distal middle cerebral artery branches. Stroke 2010;41:95-101. [Crossref] [PubMed]
33. van Osch MJ, Vonken EJ, Bakker CJ, Viergever MA. Correcting partial volume artifacts of the arterial input function in quantitative cerebral perfusion MRI. Magn Reson Med 2001;45:477-85. [Crossref] [PubMed]
34. Lansberg MG, Straka M, Kemp S, Mlynash M, Wechsler LR, Jovin TG, et al. MRI profile and response to endovascular reperfusion after stroke (DEFUSE 2): a prospective cohort study. Lancet Neurol 2012;11:860-7. [Crossref] [PubMed]
35. Álvarez-Sabín J, Maisterra O, Santamarina E, Kase CS. Factors influencing haemorrhagic transformation in ischaemic stroke. Lancet Neurol 2013;12:689-705. [Crossref] [PubMed]
36. Strbian D, Sairanen T, Meretoja A, Pitkäniemi J, Putaala J, Salonen O, Silvennoinen H, Kaste M, Tatlisumak TFor the Helsinki Stroke Thrombolysis Registry Group. Patient outcomes from symptomatic intracerebral hemorrhage after stroke thrombolysis. Neurology 2011;77:341-8. [Crossref] [PubMed]
37. Bracard S, Ducrocq X, Mas JL, Soudant M, Oppenheim C, Moulin T, Guillemin FTHRACE investigators. Mechanical thrombectomy after intravenous alteplase versus alteplase alone after stroke (THRACE): a randomised controlled trial. Lancet Neurol 2016;15:1138-47. [Crossref] [PubMed]
38. Hao Y, Yang D, Wang H, Zi W, Zhang M, Geng Y, Zhou Z, Wang W, Xu H, Tian X, Lv P, Liu Y, Xiong Y, Liu X, Xu G. ACTUAL Investigators (Endovascular Treatment for Acute Anterior Circulation Ischemic Stroke Registry). Predictors for Symptomatic Intracranial Hemorrhage After Endovascular Treatment of Acute Ischemic Stroke. Stroke 2017;48:1203-9. [Crossref] [PubMed]
39. Emberson J, Lees KR, Lyden P, Blackwell L, Albers G, Bluhmki E, et al. Effect of treatment delay, age, and stroke severity on the effects of intravenous thrombolysis with alteplase for acute ischaemic stroke: a meta-analysis of individual patient data from randomised trials. Lancet 2014;384:1929-35. [Crossref] [PubMed]
40. Campbell BCV, Majoie CBLM, Albers GW, Menon BK, Yassi N, Sharma G, et al. Penumbral imaging and functional outcome in patients with anterior circulation ischaemic stroke treated with endovascular thrombectomy versus medical therapy: a meta-analysis of individual patient-level data. Lancet Neurol 2019;18:46-55. [Crossref] [PubMed]
41. Wassélius J, Arnberg F, von Euler M, Wester P, Ullberg T. Endovascular thrombectomy for acute ischemic stroke. J Intern Med 2022;291:303-16. [Crossref] [PubMed]