You are here

  • Home
  • Archive
  • Volume 13, Issue 2
  • 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

Article Text

Article menu

Download PDFPDF

Ischemic Stroke
Original research
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
  1. Ryan A Rava1,2,
  2. Kenneth V Snyder2,3,
  3. Maxim Mokin4,
  4. Muhammad Waqas2,3,
  5. Xiaoliang Zhang1,
  6. Alexander R Podgorsak1,2,5,
  7. Ariana B Allman1,2,
  8. Jillian Senko1,2,
  9. Mohammad Mahdi Shiraz Bhurwani1,2,
  10. Yiemeng Hoi6,
  11. Jason M Davies2,3,7,
  12. Elad I Levy2,3,
  13. Adnan H Siddiqui2,3,
  14. Ciprian N Ionita1,2,3,5
  1. 1 Biomedical Engineering, University at Buffalo-The State University of New York, Buffalo, New York, USA
  2. 2 Canon Stroke and Vascular Research Center, Buffalo, New York, USA
  3. 3 Neurosurgery, University at Buffalo School of Medicine and Biomedical Sciences, Buffalo, New York, USA
  4. 4 Neurosurgery, University of South Florida, Tampa, Florida, USA
  5. 5 Medical Physics, University at Buffalo - The State University of New York, Buffalo, New York, USA
  6. 6 Canon Medical Systems USA Inc, Tustin, California, USA
  7. 7 Biomedical Informatics, University at Buffalo,The State University of New York, Buffalo, New York, USA
  1. Correspondence to Dr Ciprian N Ionita, Neurosurgery and Biomedical Engineering, Toshiba Stroke and Vascular Research Institute, University at Buffalo School of Engineering and Applied Sciences, Buffalo, NY 14206, USA; cnionita{at}buffalo.edu

Abstract

Background CT perfusion (CTP) infarct and penumbra estimations determine the eligibility of patients with acute ischemic stroke (AIS) for endovascular intervention. This study aimed to determine volumetric and spatial agreement of predicted RAPID, Vitrea, and Sphere CTP infarct with follow-up fluid attenuation inversion recovery (FLAIR) MRI infarct.

Methods 108 consecutive patients with AIS and large vessel occlusion were included in the study between April 2019 and January 2020 . Patients were divided into two groups: endovascular intervention (n=58) and conservative treatment (n=50). Intervention patients were treated with mechanical thrombectomy and achieved successful reperfusion (Thrombolysis in Cerebral Infarction 2b/2 c/3) while patients in the conservative treatment group did not receive mechanical thrombectomy or intravenous thrombolysis. Intervention and conservative treatment patients were included to assess infarct and penumbra estimations, respectively. It was assumed that in all patients treated conservatively, penumbra converted to infarct. CTP infarct and penumbra volumes were segmented from RAPID, Vitrea, and Sphere to assess volumetric and spatial agreement with follow-up FLAIR MRI.

Results Mean infarct differences (95% CIs) between each CTP software and FLAIR MRI for each cohort were: intervention cohort: RAPID=9.0±7.7 mL, Sphere=−0.2±8.7 mL, Vitrea=−7.9±8.9 mL; conservative treatment cohort: RAPID=−31.9±21.6 mL, Sphere=−26.8±17.4 mL, Vitrea=−15.3±13.7 mL. Overlap and Dice coefficients for predicted infarct were (overlap, Dice): intervention cohort: RAPID=(0.57, 0.44), Sphere=(0.68, 0.60), Vitrea=(0.70, 0.60); conservative treatment cohort: RAPID=(0.71, 0.56), Sphere=(0.73, 0.60), Vitrea=(0.72, 0.64).

Conclusions Sphere proved the most accurate in patients who had intervention infarct assessment as Vitrea and RAPID overestimated and underestimated infarct, respectively. Vitrea proved the most accurate in penumbra assessment for patients treated conservatively although all software overestimated penumbra.

  • brain
  • CT perfusion
  • stroke
  • MRI

https://doi.org/10.1136/neurintsurg-2020-015966

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Timely reperfusion of patients with acute ischemic stroke (AIS) reduces mortality and functional dependence.1 CT perfusion (CTP) is commonly used to assess patient eligible for endovascular intervention. Although CTP has been shown to benefit patients when symptom onset is >6 hours, this time limit is not dichotomous.2 3 Multiple studies have shown high agreement of infarct prediction between initial CTP imaging and follow-up diffusion weighted imaging (DWI) or fluid attenuation inversion recovery (FLAIR) MRI.4 5 Volumes and spatial locations of infarct have differed between various software programs, based on the perfusion processing algorithms and thresholds used.6 7 Validation of predicted penumbra tissue using CTP has been elusive due to the absence of a standalone penumbra validation method.

    Bayesian post-processing algorithms have been used by Sphere (Olea Medical, La Ciotat, France) and Vitrea (Vital Images, Minnetonka, MN) CTP software to segment ischemic tissue. However, different perfusion parameters and thresholds are used by these software programs as Sphere relies on cerebral blood flow (CBF) with time to peak (TTP) while Vitrea relies on cerebral blood volume with TTP or mean transit time (MTT), respectively, to segment ischemic tissue.8 9 Clinically studied RAPID (iSchemaView, Menlo Park, California, USA) CTP software used different perfusion parameter thresholds (CBF with time to reach the maximum residue function (Tmax)) to segment infarct and penumbra.4 Although various studies have been conducted validating CTP segmented infarct, few studies have addressed the accuracy of spatial location of CTP infarct.3 Additionally, reliability on the amount of penumbra and spatial location has to be answered.

    The aim of this study was to investigate the agreement of spatial overlap between final infarct volumes (FIVs) from FLAIR, and predicted infarct volumes from commonly used, and commercially available, CTP software programs, Sphere, Vitrea, and RAPID. Also, performances in quantifying and spatially locating penumbra were compared across Sphere, Vitrea, and RAPID.

    Methods

    Patient selection

    For this Health Insurance Portability and Accountability retrospective study, institutional review board approval was obtained and informed consent was waived. This study included patients with AIS and large vessel occlusion who underwent CTP evaluation on arrival at a comprehensive stroke center and FLAIR MRI within 48 hours following CTP imaging. Patients with evidence of hemorrhage were excluded using non-contrast CT. A total of 108 consecutive patients between April 2019 and January 2020 meeting the inclusion criteria were included and divided into endovascular intervention (n=58) and conservative treatment (n=50) groups. Intervention patients were required to have undergone successful mechanical thrombectomy, defined as Thrombolysis in Cerebral Infarction (TICI) 2b/2 c/3. Conservative treatment patients did not undergo thrombectomy or receive intravenous thrombolysis. Attendings determined which patients were eligible for thrombectomy based on a comprehensive assessment of the patient’s National Institutes of Health Stroke Scale score (>6), time since onset of symptoms, and baseline CTP penumbra to infarct ratio (ratio >1). TICI scores were measured as a consensus by two endovascular fellows and attendings not involved in data collection. Intervention patients were included to assess infarct as penumbra is salvaged for these patients while conservative treatment patients were included to assess penumbra based on the assumption that all penumbra converted to infarct for these patients.10

    CTP analysis

    Two 2012 Aquilion ONE CT units (Canon Medical Systems Corporation, Otawara, Japan) were used to obtain CTP data using the Neuro ONE protocol. Nineteen scan volumes, each containing 320 slices of 0.5 mm resolution, were obtained between 1.5 and 3.2 s apart over a period of 49.3 s. Omnipaque 350 (50 mL) was injected at a rate of 5 mL/s. Tube voltage of 80 kVp, CT dose indices from 15.3 to 44.4 mGy, and dose length products from 244.5 to 709.8 mGy×cm were used. Reconstructed CTP volumes were available for post-processing within 5 min from the start of the scan.

    Infarct and penumbra measurements and segmentations were conducted using automated analysis with RAPID, Sphere, and Vitrea software. Also, a recommended Vitrea workflow involving user segmentation of the stroke region throughout perfusion volume was conducted as supplementary material. For CTP processing, Sphere downsamples to a slice thickness of 5 mm while RAPID downsamples to a slice thickness of 10 mm for noise reduction. Vitrea uses preprocessing, such as smoothing filters, resulting in an effective CTP processing slice thickness of 2 mm. RAPID CTP analysis was conducted by iSchemaView, identifying infarct as tissue with a relative CBF <30% compared with the contralateral hemisphere, and penumbra as tissue with a Tmax >6 s. Sphere (3.0-SP20) analysis identified infarct as tissue with a relative CBF <25% and >5 s difference in TTP of healthy brain, and penumbra as tissue with >5 s difference in TTP compared with healthy brain. Vitrea software (V.7.10) detects infarct as tissue with relative CBF <38% (with 5.3 s increase in TTP or relative MTT<55%) compared with healthy brain, and penumbra as tissue with 5.3 s increase in TTP or 5.8 s increase in delay time or relative CBF <58% of healthy brain. In rare instances, reduced MTT can occur when the CT scan starts too early, resulting in time density curve truncation.11 Figure 1 compares all three CTP software outputs with FLAIR in predicting FIVs for a patient treated conservatively. Online supplementary figures 1 and 2 compare all three software programs for bilateral infarction and erroneous infarct segmentation cases, respectively. Following ischemic tissue segmentation, infarct and penumbra volumes were exported as digital imaging and communications files from Sphere and Vitrea using built-in features and obtained from the hospital’s picture archiving and communications system for RAPID for registration with FLAIR and spatial agreement analysis.

    Figure 1
    Figure 1

    Ground truth infarct label with the hyper-intensified regions in the fluid attenuation inversion recovery (FLAIR) MRI identifying infarct. RAPID perfusion analysis demonstrates infarct as pink and penumbra as green, while Sphere and Vitrea perfusion analysis demonstrate infarct as red and penumbra as yellow, with Vitrea indicating ventricles in purple. For this patient treated conservatively, there is agreement between all three software programs regarding infarct volume and their respective spatial locations within FLAIR MRI. Final infarct volumes for FLAIR and each CT perfusion software are: FLAIR=214.5 mL, RAPID=272.0 mL, Sphere=252.8 mL, and Vitrea=224.5 mL.

    FLAIR MRI analysis

    FLAIR data were obtained using a Vantage 1.5 T MRI unit (Canon Medical Systems Corporation, Otawara, Japan). The FLAIR protocol involved an echo time of 120 ms, repetition time of 10 000 ms, inversion time of 2500 ms, slice thickness of 5 mm, and in-plane resolution of 0.95 mm. FLAIR was used rather than DWI to calculate FIVs due to instances of DWI showing lesion reversal in the event of rapid complete reperfusion.12

    Infarct volumes from FLAIR were segmented using a technique described in a previous study where a minimum image intensity difference of 120% across hemispheres along with a minimum 120% increase in image intensity from baseline healthy white matter was required to identify infarct tissue. Spatial overlap of segmented infarct using this method was compared with manually segmented infarct for 15 patients and found to have a Dice coefficient of 0.82±0.04 and sensitivity of 0.81±0.060.13 Segmented FIVs were compared with the summation of infarct and penumbra for patients who received conservative treatment.

    Data registration

    CTP volumes were registered with FLAIR using MATLAB’s geometric and intensity based registration tool, as used in previous studies.14–16 Affine transformation, linear interpolation, and multimodal Mattes method similarity metric were used for registration along with a one-plus-one evolutionary optimizer.16 Mattes similarity metric is designed for registration across different imaging modalities through joint probability distribution calculation in both volumes for proper mapping between volumes.17 The transformation used to register CTP and FLAIR volumes was saved and used to register the FLAIR infarct and exported CTP infarct volumes from each software aprogram.

    To assess the accuracy of this registration, ventricles were manually segmented within the CTP and FLAIR volumes prior to registration. Following registration, Dice coefficients were calculated to determine the degree of ventricle spatial overlap.3

    Statistical analysis

    Summary statistics for continuous variables and frequency distributions for categorical data were tabulated for all data. For volume comparisons, mean infarct differences and mean absolute errors (MAEs) were calculated based on infarct growth, defined as final FLAIR infarct minus each CTP software’s predicted infarct, for patients who received the intervention and conservative treatment separately. Subgroup analysis was conducted on each reperfusion status for the intervention group. Predicted infarct for the conservative treatment group was calculated as a summation of infarct and penumbra assuming all penumbra converted to infarct. Percentage of patients with infarct overestimation was calculated for each cohort. Regression analysis and Spearman correlation coefficients between FLAIR FIVs and CTP predicted infarct were calculated and are represented in scatterplots for each cohort. ANOVA analysis was conducted on mean infarct differences across the three software programs for each study cohort. CTP and FLAIR infarct spatial agreement for the intervention and conservative treatment groups were expressed using Dice coefficients, sensitivity, specificity, accuracy, and overlap coefficients. Sensitivity was calculated as the amount of true positive infarct voxels, based on FLAIR, from CTP divided by the summation of true positive and false negative infarct voxels from CTP. Specificity was calculated as the amount of true negative, or healthy tissue, voxels divided by the summation of true negative and false positive infarct voxels from CTP. Accuracy was defined as the summation of true positive infarct and true negative voxels from CTP divided by the summation of all voxels in the CTP volume. Overlap coefficients were determined using predicted CTP infarct as the reference volume to determine the amount of predicted lesion within the FLAIR FIV. This metric was calculated by dividing the number of overlapping infarct voxels between the CTP scan and FLAIR by the total number of infarct voxel from CTP imaging.

    Results

    Patient characteristics and infarct volumes are shown in table 1 along with analysis after separation into intervention and conservative treatment groups. A significant difference (p<0.05) was only seen between the two cohorts for time from stroke onset to time of initial imaging. Of the 108 patients, 70 (65%) were imaged within 6 hours from onset of symptoms and 47/58 (81%) in the intervention group were treated before the 6 hour mark. Twenty-eight of 58 (48%) intervention patients received intravenous thrombolysis at a median of 124.5 min (IQR 84.8–195.5 min) after symptom onset.

    View this table:
    Table 1

    Characteristics and outcomes of patients with ischemic stroke

    Volumetric and spatial overlap analysis

    Mean differences and MAEs between FLAIR FIVs and predicted CTP infarct for each software program are indicated in table 2 along with the percentage of patients whose infarct was overestimated, for the intervention and conservative treatment groups. Subgroup analysis based on reperfusion status for these metrics is available in online supplementary table 1. ANOVA analysis indicated significant evidence for mean infarct difference between the three software programs (p=0.023) for the intervention group. For the conservative treatment group, p>0.05 using ANOVA analysis indicated no significance difference between the three CTP software programs when calculating average differences between predicted and FIVs. Figure 2A–C shows regression plots for each CTP software program, comparing predicted infarct with FIVs along with Spearman correlation coefficients for the endovascular intervention group. Figure 2D–F shows similar regression analysis for the conservative treatment group.

    Figure 2
    Figure 2

    (A–C) For the endovascular intervention group, correlation between final fluid attenuation inversion recovery (FLAIR) MRI infarct and predicted CT perfusion (CTP) infarct volumes for the (A) Sphere, (B) Vitrea, and RAPID (C) software programs. (D–F) For the conservative treatment group, correlation between final FLAIR MRI infarct and predicted CTP infarct volumes for the (D) Sphere, (E) Vitrea, and (F) RAPID software programs. Linear regression equations and Spearman correlation coefficients are represented in each plot.

    View this table:
    Table 2

    Mean differences (95% CIs), mean absolute errors, percentage of patients with infarct overestimation, and degree of spatial overlap for calculated infarct between each CT perfusion software program and fluid attenuation inversion recovery

    Spatial agreement analysis indicated a Dice of 0.78±0.02 in assessing ventricle overlap for registration accuracy. Table 2 shows the degree of spatial overlap between the predicted infarct and FIVs for the three CTP software programs and for each patient category. For the intervention group, ANOVA analysis p=0.002 for overlap coefficients indicated a significant difference between the amount of spatial overlap with FLAIR FIVs for the three CTP software programs. For intervention group, Dice coefficient calculation, ANOVA p<0.0001, showed a similar trend regarding significant discrepancies between the infarct spatial overlap of the three CTP software programs. Online supplementary table 2 shows the spatial overlap metrics for each reperfusion status in the intervention group. ANOVA analysis indicated p values >0.05 and <0.05 for overlap and Dice coefficients, respectively, for the conservative treatment group when comparing spatial overlap of predicted and FIV for the three CTP software programs.

    The recommended workflow for Vitrea v7.10 encourages user interaction to check outputs (such as artery and vein curves) and segment the stroke region throughout perfusion volumes. All volume and spatial overlap metrics for the recommended Vitrea workflow are shown in online supplementary table 3. Online supplementary table 4 shows the reperfusion status analysis for the recommended Vitrea workflow.

    Discussion

    This study compared volume estimation of infarct and penumbra between three available CTP software programs with follow-up FLAIR FIVs. These volumes of infarct and penumbra were quantified using perfusion map thresholds deemed most efficient by previously conducted studies and manufactures of the software to determine the agreement of ischemic tissue spatial overlap.4 9 Although accurate quantification of infarct tissue has been shown to impact the clinical outcome of patients with AIS, spatial agreement of infarct between initial and follow-up imaging is important as it allows clinicians to determine precise regions affected by occlusions along with embolus location.18 19 Accurate volume and spatial estimation of penumbra tissue is of great importance as it impacts on eligibility for reperfusion procedures. Typically, a penumbra to infarct ratio of at least 1 is needed for enrollment in endovascular procedures.20 In practice, underestimation of infarct is preferred to allow for inclusion of patients in endovascular procedures to regain lost neurological function but this comes with the potential risk of reperfusion hemorrhage.4

    Results for patients who received endovascular intervention indicated that Sphere had the best infarct volume estimation, whereas slightly higher Dice and overlap coefficients were seen for Sphere and Vitrea compared with RAPID. Subgroup analyses based on reperfusion status indicated a similar trend and an increase in infarct segmentation accuracy with TICI 3. However, RAPID and recommended Vitrea workflow had fewer instances of infarct overestimation, indicating they are more likely to deem patients eligible for thrombectomy. This indicates that RAPID and recommended Vitrea workflow may be more clinically practical as they provide fewer unnecessary exclusions from reperfusion procedures. Discrepancies are viewed between Dice and overlap coefficients due to evolution of infarct following onset which continues to occur between initial imaging, endovascular procedures, and follow-up imaging. Because of this, overlap coefficients may be more efficient as they describe the amount of the CTP lesion encompassed within the FLAIR lesion. Furthermore, spatial agreement results for intervention patients appeared lower than for conservative treatment patients because these patients had smaller regions of infarct. Twenty-two intervention patients had infarct volumes <15 mL, which leads to an increased probability in lower spatial overlap agreement.

    The results for patients who received conservative treatment indicated an overestimation of predicted infarct volumes compared with FIVs for all three software programs. This overestimation of predicted infarct can be attributed to an overestimation of penumbra as predicted infarct is considered as the summation of infarct with penumbra, and infarct estimation was shown to be accurate on its own in the intervention group. Use of the recommended Vitrea workflow with user segmentation of ischemic regions can however eliminate penumbra overestimation. RAPID’s large overestimation of penumbra can be attributed to its trend of underestimating infarct, as indicated by the percentage of patients with infarct overestimation in table 2. Even with this penumbra overestimation, overlay coefficients increased for all three software programs as the baseline CTP lesion was now substantially larger leading to a higher probability of it being contained within the FIV lesion.

    Although this study provides an estimate of volumetric and spatial overlap of penumbra tissue, there is no independently feasible way to validate the accuracy of this estimation. The assumption that all penumbra converts to infarct in conservative treatment patients can provide a method for this validation, although it has limitations. Spontaneous recanalization is the main issue associated with this assumption as it is possible that some patients will reperfuse themselves by dislodging the embolus based on vessel flow conditions.21 This spontaneous reperfusion occurs in 5–30% of cases, as indicated in studies by Paciaroni et al and Meves et al.22–24 This would lead to penumbra being salvaged as opposed to turning to infarct. It is possible that spontaneous reperfusion played a role in the overestimation of penumbra for all three software programs, although any bias is eliminated by testing each software program on the same group of patients.

    Mean infarct differences and spatial agreement metrics between Vitrea and Sphere with RAPID are of great relevance as RAPID has been shown to accurately assess FIVs in previous studies compared with other CTP software programs.25 26 RAPID has been used in the DEFUSE3 (Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke), DAWN (DWI or CTP Assessment with Clinical Mismatch in the Triage of Wake-Up and Late Presenting Strokes Undergoing Neurointervention with Trevo), and SWIFT PRIME (Solitaire with Intention for Thrombectomy as Primary Endovascular Treatment) trials, showing good infarct volume estimation for the purpose of determining patient eligibility for endovascular procedures.5 20 27 Conclusions from these studies along with results from this study indicate that Vitrea and Sphere appear to perform similar to RAPID and can be clinically applied for segmentation of infarct tissue.

    Limitations of the study include the time from initial perfusion imaging to follow-up FLAIR imaging. Due to penumbra converting to infarct at a rate up to 10.1 mL/hour, any predicted penumbra could potentially appear as infarct on follow-up imaging for intervention cases if not all penumbra is salvaged (ie, TICI 2b).28 Also, it has been shown that patients who achieve TICI 2b/2 c/3 reperfusion within 100 min following CTP imaging have lower degrees of minimum infarct growth, which would increase the spatial overlap of infarct volumes.29 The presence of edema is a further limitation as this can inflate the FLAIR infarct volume, as seen in figure 1. Another limitation is that the segmentation algorithm for infarct from FLAIR is imperfect and although it does have a Dice coefficient of 0.82, there is potential for erroneous segmentation of infarct, which lowered the results from the study. Similarly, the accuracy of the registration algorithm may have the impacted spatial overlap results as inaccuracies, similar to those seen by Hoving et al, due to registration occurred between the imaging modalities.3 Additionally, head movement during the CTP scan was not quantified for these patients, which could have impacted the accuracy of infarct volume estimations.30 Comparing only three CTP software programs is a further limitation as there are many more (eg, syngo.via (Siemens Healthcare, Erlangen, Germany) and IntelliSpace Portal (Philips Corporation, Amsterdam, Noord Holland, the Netherlands)). A final limitation is that all patients came from one comprehensive stroke center and the study did not include patients outside of the institution.

    Conclusions

    The Sphere software program performed best overall in assessing infarct volume in the intervention group based on the lowest mean infarct difference metric, with similar overlap and Dice coefficients to Vitrea. Also, the Vitrea software program appeared to estimate penumbra volumes more accurately that the other software programs, as indicated by the lower mean infarct differences and similar spatial agreement metrics for the conservative treatment group. This indicates multiple perfusion parameter combinations and thresholds have the ability to detect infarct accurately. However, based on volumetric and spatial agreement metrics, there still appears to be room for improvement regarding automated penumbra estimation using CTP.

    References

    1. ↵
      2. Albers GW ,
      3. Thijs VN ,
      4. Wechsler L , et al
      . 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.doi:10.1002/ana.20976 pmid:http://www.ncbi.nlm.nih.gov/pubmed/17066483
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      2. Powers WJ ,
      3. Rabinstein AA ,
      4. Ackerson T , et al
      . 2018 guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2018;49:46–99.doi:10.1161/STR.0000000000000158
      OpenUrlAbstract/FREE Full Text
    3. ↵
      2. Hoving JW ,
      3. Marquering HA ,
      4. Majoie CBLM , et al
      . Volumetric and spatial accuracy of computed tomography perfusion estimated ischemic core volume in patients with acute ischemic stroke. Stroke 2018;49:2368–75.doi:10.1161/STROKEAHA.118.020846 pmid:http://www.ncbi.nlm.nih.gov/pubmed/30355095
      OpenUrlPubMed
    4. ↵
      2. Mokin M ,
      3. Levy EI ,
      4. Saver JL , et al
      . Predictive value of RAPID assessed perfusion thresholds on final infarct volume in SWIFT PRIME (solitaire with the intention for thrombectomy as primary endovascular treatment). Stroke 2017;48:932–8.doi:10.1161/STROKEAHA.116.015472 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28283606
      OpenUrlAbstract/FREE Full Text
    5. ↵
      2. Saver JL ,
      3. Goyal M ,
      4. Bonafe A , et al
      . Solitaire with the Intention for Thrombectomy as Primary Endovascular Treatment for Acute Ischemic Stroke (SWIFT PRIME) trial: protocol for a randomized, controlled, multicenter study comparing the Solitaire revascularization device with IV tPA with IV tPA alone in acute ischemic stroke. Int J Stroke 2015;10:439–48.doi:10.1111/ijs.12459 pmid:25777831
      OpenUrlCrossRefPubMed
    6. ↵
      2. Bathla G ,
      3. Limaye K ,
      4. Policeni B , et al
      . Achieving comparable perfusion results across vendors. the next step in standardizing stroke care: a technical report. J Neurointerv Surg 2019;11:1257–60.doi:10.1136/neurintsurg-2019-014810 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31118267
      OpenUrlAbstract/FREE Full Text
    7. ↵
      2. Koopman MS ,
      3. Berkhemer OA ,
      4. Geuskens RREG , et al
      . Comparison of three commonly used CT perfusion software packages in patients with acute ischemic stroke. J Neurointerv Surg 2019;11:1249–56.doi:10.1136/neurintsurg-2019-014822 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31203208
      OpenUrlAbstract/FREE Full Text
    8. ↵
      2. Xiong Y ,
      3. Huang CC ,
      4. Fisher M , et al
      . Comparison of automated CT perfusion softwares in evaluation of acute ischemic stroke. J Stroke Cerebrovasc Dis 2019;28:104392.doi:10.1016/j.jstrokecerebrovasdis.2019.104392 pmid:31562038
      OpenUrlCrossRefPubMed
    9. ↵
      2. Rava RA ,
      3. Snyder KV ,
      4. Mokin M , et al
      . Assessment of a Bayesian Vitrea CT perfusion analysis to predict final infarct and penumbra volumes in patients with acute ischemic stroke: a comparison with rapid. AJNR Am J Neuroradiol 2020;41:206–12.doi:10.3174/ajnr.A6395 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31948951
      OpenUrlAbstract/FREE Full Text
    10. ↵
      2. Chavez JC ,
      3. Hurko O ,
      4. Barone FC , et al
      . Pharmacologic interventions for stroke: looking beyond the thrombolysis time window into the penumbra with biomarkers, not a stopwatch. Stroke 2009;40:558–63.
      OpenUrlCrossRefPubMed
    11. ↵
      2. Kudo K ,
      3. Sasaki M ,
      4. Yamada K , et al
      . Differences in CT perfusion maps generated by different commercial software: quantitative analysis by using identical source data of acute stroke patients. Radiology 2010;254:200–9.doi:10.1148/radiol.254082000 pmid:http://www.ncbi.nlm.nih.gov/pubmed/20032153
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      2. Inoue M ,
      3. Mlynash M ,
      4. Christensen S , et al
      . Early DWI reversal following endovascular reperfusion is typically transient in patients imaged 3-6 hours after onset. Stroke 2014;45:1024.
    13. ↵
      2. Rava RA ,
      3. Waqas M ,
      4. Snyder KV , et al
      . Comparison of automated cerebral infarct segmentation techniques using DWI and FLAIR MRI. The International Society for Magnetic Resonance in Medicine; August 8-13, 2020.
    14. ↵
      2. Hu HH ,
      3. Li Z ,
      4. Pokorney AL , et al
      . Assessment of cerebral blood perfusion reserve with acetazolamide using 3D spiral ASL MRI: preliminary experience in pediatric patients. Magn Reson Imaging 2017;35:132–40.doi:10.1016/j.mri.2016.08.019 pmid:http://www.ncbi.nlm.nih.gov/pubmed/27580517
      OpenUrlPubMed
    15. ↵
      2. Maknojia S ,
      3. Tam F ,
      4. Das S , et al
      . Visualization of brain shift corrected functional magnetic resonance imaging data for intraoperative brain mapping. World Neurosurg X 2019;2:100021. doi:10.1016/j.wnsx.2019.100021 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31218295
      OpenUrlPubMed
    16. ↵
      2. Muthukumaran D ,
      3. Sivakumar M
      . Medical image registration: a Matlab based approach. Int J Sci Res Comp Sci Engineer Info Technol 2017;2:29–34.
      OpenUrl
    17. ↵
      2. Mattes D ,
      3. Haynor DR ,
      4. Vesselle H , et al
      . PET-CT image registration in the chest using free-form deformations. IEEE Trans Med Imaging 2003;22:120–8.doi:10.1109/TMI.2003.809072 pmid:http://www.ncbi.nlm.nih.gov/pubmed/12703765
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      2. Gasparotti R ,
      3. Grassi M ,
      4. Mardighian D , et al
      . Perfusion CT in patients with acute ischemic stroke treated with intra-arterial thrombolysis: predictive value of infarct core size on clinical outcome. AJNR Am J Neuroradiol 2009;30:722–7.doi:10.3174/ajnr.A1439 pmid:http://www.ncbi.nlm.nih.gov/pubmed/19164437
      OpenUrlAbstract/FREE Full Text
    19. ↵
      2. Cullen SP ,
      3. Symons SP ,
      4. Hunter G , et al
      . Dynamic contrast-enhanced computed tomography of acute ischemic stroke: CTA and CTP. Semin Roentgenol 2002;37:192–205.doi:10.1016/S0037-198X(02)80019-0 pmid:12226898
      OpenUrlCrossRefPubMed
    20. ↵
      2. Albers GW ,
      3. Lansberg MG ,
      4. Kemp S , et al
      . A multicenter randomized controlled trial of endovascular therapy following imaging evaluation for ischemic stroke (DEFUSE 3). Int J Stroke 2017;12:896–905.doi:10.1177/1747493017701147 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28946832
      OpenUrlCrossRefPubMed
    21. ↵
      2. Barber PA ,
      3. Davis SM ,
      4. Infeld B , et al
      . Spontaneous reperfusion after ischemic stroke is associated with improved outcome. Stroke 1998;29:2522–8.doi:10.1161/01.STR.29.12.2522 pmid:http://www.ncbi.nlm.nih.gov/pubmed/9836763
      OpenUrlAbstract/FREE Full Text
    22. ↵
      2. Shah QA
      . Spontaneous recanalization after complete occlusion of the common carotid artery with subsequent embolic ischemic stroke. J Vasc Interv Neurol 2009;2:147–51.pmid:http://www.ncbi.nlm.nih.gov/pubmed/22518244
      OpenUrlPubMed
    23. ↵
      2. Paciaroni M ,
      3. Caso V ,
      4. Venti M , et al
      . Outcome in patients with stroke associated with internal carotid artery occlusion. Cerebrovasc Dis 2005;20:108–13.doi:10.1159/000086800 pmid:http://www.ncbi.nlm.nih.gov/pubmed/16006758
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      2. Meves SH ,
      3. Muhs A ,
      4. Federlein J , et al
      . Recanalization of acute symptomatic occlusions of the internal carotid artery. J Neurol 2002;249:188–92.doi:10.1007/PL00007863 pmid:http://www.ncbi.nlm.nih.gov/pubmed/11985385
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      2. Austein F ,
      3. Riedel C ,
      4. Kerby T , et al
      . Comparison of perfusion CT software to predict the final infarct volume after thrombectomy. Stroke 2016;47:2311–7.doi:10.1161/STROKEAHA.116.013147 pmid:http://www.ncbi.nlm.nih.gov/pubmed/27507864
      OpenUrlAbstract/FREE Full Text
    26. ↵
      2. Kudo K ,
      3. Christensen S ,
      4. Sasaki M , et al
      . Accuracy and reliability assessment of CT and MR perfusion analysis software using a digital phantom. Radiology 2013;267:201–11.doi:10.1148/radiol.12112618 pmid:http://www.ncbi.nlm.nih.gov/pubmed/23220899
      OpenUrlCrossRefPubMed
    27. ↵
      2. Jovin TG ,
      3. Saver JL ,
      4. Ribo M , et al
      . Diffusion-weighted imaging or computerized tomography perfusion assessment with clinical mismatch in the triage of wake up and late presenting strokes undergoing neurointervention with Trevo (DAWN) trial methods. Int J Stroke 2017;12:641–52.doi:10.1177/1747493017710341 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28569123
      OpenUrlCrossRefPubMed
    28. ↵
      2. Guenego A ,
      3. Mlynash M ,
      4. Christensen S , et al
      . Hypoperfusion ratio predicts infarct growth during transfer for thrombectomy. Ann Neurol 2018;84:616–20.doi:10.1002/ana.25320 pmid:http://www.ncbi.nlm.nih.gov/pubmed/30168180
      OpenUrlCrossRefPubMed
    29. ↵
      2. Albers GW ,
      3. Goyal M ,
      4. Jahan R , et al
      . Ischemic core and hypoperfusion volumes predict infarct size in SWIFT PRIME. Ann Neurol 2016;79:76–89.doi:10.1002/ana.24543 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26476022
      OpenUrlCrossRefPubMed
    30. ↵
      2. Fahmi F ,
      3. Beenen LFM ,
      4. Streekstra GJ , et al
      . Head movement during CT brain perfusion acquisition of patients with suspected acute ischemic stroke. Eur J Radiol 2013;82:2334–41.doi:10.1016/j.ejrad.2013.08.039 pmid:http://www.ncbi.nlm.nih.gov/pubmed/24041432
      OpenUrlCrossRefPubMed

    Footnotes

    • Contributors RAR, KVS, MM, MW, XZ, ARP, YH, and CNI conceived and designed the research. RAR, KVS, MW, ARP, ABA, JS, MMSB, JMD, EIL, and AHS collected and reviewed the data. RAR analyzed the data and performed the statistical analysis. CNI and JS handled the funding and supervision of the research. RAR drafted the manuscript. All authors made revisions to the manuscript and reviewed the final version.

    • Funding This research was funded by Canon Medical Systems USA, Inc.

    • Competing interests KVS: consulting for Canon Medical Systems Corporation, Penumbra Inc, Medtronic, Jacobs Institute; and co-founder of Neurovascular Diagnostics Inc. MM: consultant for Canon, Cerebrotech, Imperative care; and NIH grant support R21NS109575. YH: employee of Canon Medical Systems USA Inc. AHS: financial interest/investor/stock options/ownership in Amnis Therapeutics, Apama Medical, Blink TBI Inc, Buffalo Technology Partners Inc, Cardinal Consultants, Cerebrotech Medical Systems Inc, Cognition Medical, Endostream Medical Ltd, Imperative Care, International Medical Distribution Partners, Neurovascular Diagnostics Inc, Q’Apel Medical Inc, Rebound Therapeutics Corp, Rist Neurovascular Inc, Serenity Medical Inc, Silk Road Medical, StimMed, Synchron, Three Rivers Medical Inc, Viseon Spine Inc; consultant/advisory board for Amnis Therapeutics, Boston Scientific, Canon Medical Systems USA Inc, Cerebrotech Medical Systems Inc, Cerenovus, Corindus Inc, Endostream Medical Ltd, Guidepoint Global Consulting, Imperative Care, Integra LifeSciences Corp, Medtronic, MicroVention, Northwest University–DSMB Chair for HEAT Trial, Penumbra, Q’Apel Medical Inc, Rapid Medical, Rebound Therapeutics Corp, Serenity Medical Inc, Silk Road Medical, StimMed, Stryker, Three Rivers Medical Inc, VasSol WL Gore and associates; principal investigator/steering comment of the following trials: Cerenovus NAPA and ARISE II; Medtronic SWIFT PRIME and SWIFT DIRECT; MicroVention FRED & CONFIDENCE; MUSC POSITIVE; Penumbra 3D Separator, COMPASS, INVEST; Davies-Research grant: National Center for Advancing Translational Sciences of the National Institutes of Health under award number KL2TR001413 to the University at Buffalo; speakers’ bureau for Penumbra; honoraria for Neurotrauma Science, LLC; and shareholder/ownership interests in RIST Neurovascular. EIL: shareholder/ownership interests in NeXtGen Biologics, RAPID Medical, Claret Medical, Cognition Medical, Imperative Care (formerly the Stroke Project), Rebound Therapeutics, StimMed, Three Rivers Medical; national principal investigator/steering committees for Medtronic (merged with Covidien Neurovascular) SWIFT Prime and SWIFT Direct Trials; honoraria from Medtronic (training); consultant for Claret Medical, GLG Consulting, Guidepoint Global, Imperative Care, Medtronic, Rebound, StimMed; advisory board for Stryker (AIS Clinical Advisory Board), NeXtGen Biologics, MEDX, Cognition Medical, Endostream Medical; and site principal investigator for CONFIDENCE study (MicroVention), STRATIS Study—Sub I (Medtronic). CNI: equipment grant from Canon Medical Systems and Cummings Foundation support.

    • Patient consent for publication Not required.

    • Ethics approval University at Buffalo Institutional Review Board approval was obtained and informed consent was waived for this Health Insurance Portability and Accountability Act compliant retrospective study. IRB study No: MOD00005807.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data availability statement Data are available upon reasonable request. The data is composed of deidentified participant data and can be obtained upon request from Dr. Ciprian N. Ionita at cnionita@buffalo.edu.