Comparison of Perfusion Computed Tomography and Computed Tomography Angiography Source Images With Perfusion-Weighted Imaging and Diffusion-Weighted Imaging in Patients With Acute Stroke of Less Than 6 Hours’ Duration
Background and Purpose— We aimed to determine the diagnostic value of perfusion computed tomography (PCT) and CT angiography (CTA) including CTA source images (CTA-SI) in comparison with perfusion-weighted magnetic resonance imaging (MRI) (PWI) and diffusion-weighted MRI (DWI) in acute stroke <6 hours.
Methods— Noncontrast-enhanced CT, PCT, CTA, stroke MRI, including PWI and DWI, and MR angiography (MRA), were performed in patients with symptoms of acute stroke lasting <6 hours. We analyzed ischemic lesion volumes on patients’ arrival as shown on NECT, PCT, CTA-SI, DWI, and PWI (Wilcoxon, Spearman, Bland-Altman) and compared them to the infarct extent as shown on day 5 NECT.
Results— Twenty-two stroke patients underwent CT and MRI scanning within 6 hours. PCT time to peak (PCT-TTP) volumes did not differ from PWI-TTP (P=0.686 for patients who did not undergo thrombolysis/P=0.328 for patients who underwent thrombolysis), nor did PCT cerebral blood volume (PCT-CBV) differ from PWI-CBV (P=0.893/P=0.169). CTA-SI volumes did not differ from DWI volumes (P=0.465/P=0.086). Lesion volumes measured in PCT maps significantly correlated with lesion volumes on PWI (P=0.0047, r=1.0/P=0.0019, r=0.897 for TTP; P=0.0054, r=0.983/P=0.0026, r=0.871 for CBV). Also, PCT-CBV lesion volumes significantly correlated with follow-up CT lesion volumes (P=0.0047, r=1.0/P=0.0046, r=0.819).
Conclusions— In hyperacute stroke, the combination of PCT and CTA can render important diagnostic information regarding the infarct extent and the perfusion deficit. Lesions on PCT-TTP and PCT-CBV do not differ from lesions on PWI-TTP and PWI-CBV; lesions on CTA source images do not differ from lesions on DWI. The combination of noncontrast-enhanced CT (NECT), perfusion CT (PCT), and CT angiography (CTA) can render additional information within <15 minutes and may help in therapeutic decision-making if PWI and DWI are not available or cannot be performed on specific patients.
The advent of new magnetic resonance imaging (MRI) techniques such as perfusion-weighted imaging (PWI) and diffusion-weighted imaging (DWI) has improved diagnostic imaging in stroke.1,2 However, computed tomography (CT) scanners are more widely available and less expensive than MRI scanners and are often located in the emergency departments even of smaller community hospitals. Usually, CT is used to exclude intracranial hemorrhage (ICH) or tumor, but it can also be used to detect early signs of an infarct.3 Besides ICH exclusion, the depiction of infarct core and total hypoperfused
See Editorial Comment, page 1657
brain and vessel status is expected to be shown by a comprehensive imaging tool such as stroke MRI.4 The aim of our study was to determine the diagnostic value of perfusion CT (PCT) and CT angiography (CTA) including CTA source images (CTA-SI) in comparison with PWI and DWI in stroke patients <6 hours after symptom onset.
Patients and Methods
Our target group consisted of patients aged 18 to 85 years with symptoms of acute stroke within the first 6 hours. Patients with contraindications for MRI or contrast-enhanced CT were excluded. Exclusion criteria for PCT or CTA were a history of contrast medium allergy or renal failure. All patients received noncontrast-enhanced CT scan to exclude ICH before enrollment in the study. If a patient was enrolled, we continued the examination with PCT and CTA, followed by MRI. The planned time for performing stroke CT and MRI studies was <6 hours after symptom onset, with a time interval between CT and MRI of <2 hours, but as rapidly as possible. We obtained informed consent from all patients or their next of kin. The study protocol was approved by the local institutional review board.
Imaging and Clinical Assessment
All patients were examined with a multislice CT scanner (Somatom Volume Zoom; Siemens) and immediately thereafter with a 1.5-T whole-body MR imager (EDGE; Philips) equipped with enhanced gradient hardware for echo planar imaging. Our noncontrast-enhanced CT (NECT), CTA, CTA-SI, and stroke MRI protocols have been described in detail elsewhere.5,6 The perfusion CT technique in this study has been described by König et al.7,8 In brief, axial dual-section dynamic CT at the level of the basal ganglia was performed to encompass areas of the anterior, posterior, and middle cerebral artery territories. For each study, a bolus of 50 mL nonionic iodinated contrast agent (Ultravist; Schering) was injected into a cubital vein with a power injector at a rate of 10 mL/s. A sequence of 2 * 40 images was then collected at a rate of 1 image per second. To achieve cerebral perfusion data, we used a commercial Perfusion CT software (Siemens), which allows the calculation of cerebral blood flow (CBF), cerebral blood volume (CBV), and time to peak (TTP) maps.9
For CTA, another 65 mL of a nonionic contrast medium were injected at an injection rate of 5 mL/s for 40 mL followed by another 25 mL at 2.5 mL/s. After a delay of 17 seconds, spiral scanning was performed: beam collimation 4 * 1.0 mm, slice width 1.5 mm, normalized spiral pitch 1.0, 120 kV, and 125 mAs. For diagnosis we used the CTA-SI and the 3-dimensional reconstructions of the data sets. Both the NECT and CTA-SI were evaluated by changing the window width and center level during visual review to accentuate potentially subtle foci of low attenuation or enhancement.
PCT imaging is restricted to the subvolume covered by the total detector width of 20 mm (2 adjacent slices of 10 mm each). To allow comparison of PCT images with CTA-SI and MR images, the observers manually outlined the infarct area in the corresponding slices using the same angulation. Regarding the PCT-TTP and PCT-CBV maps, the threshold for manually outlining the lesions was set according to the color code: yellow, orange, and red for PCT-TTP lesions (corresponding to a TTP increase of >≈4 seconds), and violet and dark blue (corresponding to a CBV value of ≈30% to 40% of normal gray matter values) for PCT-CBV lesions (Figures 1E, 1F, and 2E, 2F). This method has proven to be quick, practicable, and robust under clinical conditions (regarding the feasibility in patients with acute stroke, when quick diagnosis is required). The threshold for manually outlining the lesions on PWI and DWI was set according to the hyperintensity on the maps (Figures 1G, 1H, 2G, 2⇓H). The outlined infarct areas were then multiplied with the slice thickness and added up to 20 mm thickness for comparison with PCT lesion volumes. The final infarct volume was determined on day 5, according to CT images. All PCT, CTA, DWI, and PWI lesion volumes were measured in random sequence by independent observers blinded to the patients’ identity and clinical status and to previous measurement results.
For statistical analysis we used a standard software package (StatView 5.0 MedCalc 22.214.171.124). Demographic data and time intervals of examinations and descriptive statistics of scores are given as median values with ranges. We used the nonparametric Wilcoxon test and Bland-Altmann plots to compare CT and MRI measurements. Correlations between lesion volumes from different imaging parameters were analyzed using Spearman-rank correlation. The analysis was performed separately for patients with and without thrombolysis. To further test the hypothesis that CT and MRI measurements yield equivalent results, we additionally used linear regression analysis on the subset of patients who had any kind of anomaly in the imaging procedures.
From February 2002 to April 2003, 1087 stroke patients were treated in our dedicated neurological emergency room. Of these, 71 patients were initially eligible for this study. Eighteen had an ICH and 8 denied consent for participation. Of the remaining 45 patients, 19 could not undergo MRI because of stroke severity or pacemakers, ie, the MR protocol was only performed in 26 of 45 cases (feasibility 58%). Three had a history of contrast agent allergy, ie, the complete stroke CT protocol (NECT, CTA, PCT) was performed in 42 of 45 cases (feasibility 93%). One patient was later excluded because he had experienced recurrent embolisms. Therefore, 22 patients (8 women, 14 men, mean age 66.7 years, range 39 to 84 years) were consecutively enrolled in this prospective study (Table 1). Thirteen patients received standard thrombolytic therapy with recombinant tissue plasminogen activator (rt-PA). The infusion was started during or after stroke MRI in all 13 patients.
PCT, CT, CTA, and MRI were successfully completed in all patients without any side effects. Both stroke CT and stroke MRI protocols were performed within the first 60 minutes to 6 hours after stroke onset (CT: 60 minutes to 5 hours 15 minutes, mean 2.33±1.04 hours; MRI: 1 hour 15 minutes to 6 hours, mean 3.0±1.12 hours). The time interval between CT and MRI ranged from 15 minutes to 2 hours (0.75±0.40 hours). For the CT protocol, scanning took between 10 and 20 minutes (mean 13 minutes). For MRI, the scan time was between 16 and 32 minutes (mean 23 minutes). In no instance was thrombolytic treatment delayed by imaging.
Of the 22 patients, 6 showed early signs of infarction on the NECT images. Three had cortical hypodensity and 3 had a hyperdense artery sign. In these 6 patients the observers also depicted lesions on PCT-TTP and PCT-CBV and on day 5 NECT (Table 2).
Eleven patients had a vessel occlusion seen on both CTA and MRA at the same location. Ten of the 22 patients had an initial vessel occlusion in the anterior or middle cerebral circulation according to CTA and MRA, and 1 patient had occlusion of the distal basilar artery. Six patients each presented with either a proximal or a distal middle cerebral artery (MCA) main stem occlusion, whereas 1 had MCA branch occlusion and 3 had distal internal carotid artery (ICA) occlusion according to the initial CTA. All but 1 of the 11 patients without vessel occlusion on CTA and MRA showed no initial DWI lesion in the 2 adjacent slices.
Comparison of CT and MR Results and Correlation With Follow-Up Infarct Volumes
PCT lesion sizes did not differ significantly from PWI lesion size TTP (P=0.686 for patients who did not undergo thrombolysis/P=0.328 for patients who underwent thrombolysis; P=0.893/P=0.169 for CBV; Wilcoxon) at baseline (Table 3a). Bland-Altmann plots confirmed the equivalence of both modalities (Figure 3). Furthermore, the difference between CTA-SI lesion volumes and DWI lesion volumes did not reach statistical significance (P=0.465/P=0.086; Wilcoxon).
In all patients, the lesion volume measured in PCT maps significantly correlated (Spearman rank) with the lesion volume on PWI (r=1.0/r=0.90 for TTP; r=0.98/r=0.87 for CBV). Significant correlation was also found between CTA-SI and DWI lesion volumes (r=0.98/r=0.92) as well as between PCT-CBV and DWI lesion volumes (r=1.0/r=0.87). PCT-CBV lesion volumes significantly correlated with follow-up CT lesion volumes (r=1.0/r=0.82).
Linear regression analysis emphasized the principal equivalence of CT and MR lesion sizes. To avoid bias by zeros, we restricted the analysis to the subset of patients with imaging lesions (n=16). Table 3b shows that lesion sizes in TTP, CBV, and CTA-SI/DWI can be presented as proportional to each other (intercept=0) with a factor close to 1. Remarkably, this is also true for the lesion size in follow-up as predicted by PCT-CBV (ratio 1.03).
In all cases of vessel occlusion, DWI lesions corresponded to lesions in the PCT maps, even though lesions on PCT-CBV usually were slightly larger than on DWI (ratio 1.35). To estimate the influence of only using 2 adjacent slices in PCT, we also completely outlined the final infarct volume on day 5 in the whole brain. Here, we also found a significant correlation between PCT-CBV (r=0.98/r=0.88) and PCT-CBF (r=0.98/r=0.82) lesion volumes and the final infarct volume regarding not only the corresponding 2 slices on the follow-up but also the complete infarction.
Although DWI seems to be superior in direct comparison to NECT,10 stroke MRI at present is fully (24 hours) available only in major stroke centers.11 In particular, most patients with symptoms of acute hemispheric stroke are treated in general hospitals that do not have this access to MRI.
We performed CT, PCT, CTA, and stroke MRI within an average time interval of 2.33 hours for CT and 3.0 hours for MRI after symptom onset in 22 patients. Although safety and reliability were not investigated according to predefined parameters, the overall performance of the combination of CT, PCT, and CTA was good: All data could be analyzed, and no patient had allergies or renal insufficiency after contrast administration. Because of the short investigation time of CT/PCT/CTA, the scan can be performed rapidly with fewer movement artifacts than in MRI, especially in severely ill patients. Also, patient access is far easier than in the MRI scanner. The higher practical feasibility of CT was also evident in our study: although 93% of patients eligible for the study were eligible for CT and the procedure was successfully performed, this applied to only 58% of the patients with respect to MRI.
In the past, various perfusion imaging techniques have been reported to be useful in the clinical investigation of patients with acute stroke. Recent studies with PCT have shown that various combinations of perfusion parameters can be used to predict outcome after ischemic stroke.12–15 Severely reduced CBV has emerged as an important predictor for essentially nonviable tissue.8,13 In patients with clinical symptoms of ischemic stroke but no pathologic findings in NECT, PCT, and CTA, a large MCA infarction can be excluded. This information may be sufficient for the therapeutic decision to not perform thrombolysis, although smaller ischemic tissue volumes outside the 2-cm section covered by perfusion CT may be missed. Even though dynamic PCT is essentially a 2-dimensional technique, which on current multi-slice CT (MSCT) scanners is restricted to a relatively small volume, we were able to show that this does not seriously limit patient management in acute stroke caused by MCA occlusion. In cases of large MCA infarctions, we found a highly significant correlation between PCT-CBF and PCT-CBV lesion volumes with the final infarct volume, regarding not only the corresponding 2 slices on the follow-up but also the complete infarction.
Even though the reported diagnostic yield of NECT within 6 hours after symptom onset is low (50% to 70%), the detection of X-ray hypoattenuation on NECT is highly specific for irreversible brain damage.3 Still, the sensitivity in diagnosing ischemic stroke is lower than for DWI in the first few hours.10 CTA can increase the diagnostic value of CT.16–18 Furthermore, previous studies have shown that lesion volumes on CTA-SI significantly correlate with the lesion volumes on baseline DWI.5
In MRI, a PWI/DWI mismatch reflects the ischemic penumbra to a certain extent and therefore makes it possible to pragmatically estimate the size of tissue at risk for irreversible infarction.2 We stress that the combination of CT, CTA-SI, and PCT also provides similar information. Here, the volume of the affected brain area that has inadequate blood supply can be estimated by the difference between the CTA-SI lesion volumes and the PCT-CBF lesion volumes. CTA gives information about the vessel occlusion, if present, and can show the irreversibly damaged brain tissue and the status of the collateral vessels at the same time. In our study, PCT-CBV lesions were slightly larger than DWI lesions (which may depend on the visual threshold chosen for the analysis of the PCT-CBV maps), but PCT-CBV lesions correlated well with follow-up CT lesion volumes.
Our results correspond to findings of a recently published study.19 These authors also found a statistically significant correlation between CT and MR abnormalities regarding TTP and CBF. However, in our study we found a higher correlation and a higher level of significance, which is probably because of the fact that we examined all patients within the first 6 hours after stroke onset.
In hyperacute stroke, the combination of PCT and CTA can render important diagnostic information regarding the infarct and the perfusion deficit. Lesions on PCT-TTP and PCT-CBV do not differ from lesions on PWI-TTP and PWI-CBV, and lesions on CTA source images do not differ from lesions on DWI. The combination of NECT, PCT, and CTA can render additional information within <15 minutes and may help in therapeutic decision-making if PWI and DWI are not available or cannot be performed on a specific patient. Lesions on PCT maps and CTA-SI highly correlate with perfusion and diffusion-weighted stroke MRI, allowing estimation of the tissue at risk in a similar fashion.
We express our gratitude to all members of the medical and technical staff in the Department of Neuroradiology and to the members of the Heidelberg neurocritical care, stroke, and intermediate care unit’s medical and nursing staff. This work was supported by University Heidelberg Junior research grant 58/2001. The Deutsche Forschungsgemeinschaft supported our work with 2 research grants for Drs Schellinger and Fiebach (SCHE 613/1-1 and FI 897/1-1).
- Received October 23, 2003.
- Revision received February 11, 2004.
- Accepted March 26, 2004.
von Kummer R, Bourquain H, Bastianello S, Bozzao L, Manelfe C, Meier D, Hacke W. Early prediction of irreversible brain damage after ischemic stroke at CT. Radiology. 2000; 219: 95–100.
Schellinger PD, Fiebach JB, Hacke W. Imaging-based decision making in thrombolytic therapy for ischemic stroke: present status. Stroke. 2003; 34: 575–583.
Schramm P, Schellinger PD, Fiebach JB, Heiland S, Jansen O, Knauth M, Hacke W, Sartor K. Comparison of CT and CT angiography source images with diffusion-weighted imaging in patients with acute stroke within 6 hours after onset. Stroke. 2002; 33: 2426–2432.
König M, Kraus M, Theek C, Klotz E, Gehlen W, Heuser L. Quantitative assessment of the ischemic brain by means of perfusion-related parameters derived from perfusion CT. Stroke. 2001; 32: 431–437.
Fiebach JB, Schellinger PD, Jansen O, Meyer M, Wilde P, Bender J, Schramm P, Jüttler E, Oehler J, Hartmann M, Hahnel S, Knauth M, Hacke W, Sartor K. CT and diffusion-weighted MR imaging in randomized order: diffusion-weighted imaging results in higher accuracy and lower interrater variability in the diagnosis of hyperacute ischemic stroke. Stroke. 2002; 33: 2206–2210.
Handschu R, Garling A, Heuschmann PU, Kolominsky-Rabas PL, Erbguth F, Neundorfer B. Acute stroke management in the local general hospital. Stroke. 2001; 32: 866–870.
Nabavi DG, Kloska SP, Nam EM, Freund M, Gaus CG, Klotz E, Heindel W, Ringelstein EB. MOSAIC: Multimodal Stroke Assessment Using Computed Tomography: novel diagnostic approach for the prediction of infarction size and clinical outcome. Stroke. 2002; 33: 2819–2826.
Wintermark M, Reichhart M, Cuisenaire O, Maeder P, Thiran JP, Schnyder P, Bogousslavsky J, Meuli R. Comparison of admission perfusion computed tomography and qualitative diffusion- and perfusion-weighted magnetic resonance imaging in acute stroke patients. Stroke. 2002; 33: 2025–2031.
Lev MH, Segal AZ, Farkas J, Hossain ST, Putman C, Hunter GJ, Budzik R, Harris GJ, Buonanno FS, Ezzeddine MA, Chang Y, Koroshetz WJ, Gonzalez RG, Schwamm LH. Utility of perfusion-weighted CT imaging in acute middle cerebral artery stroke treated with intra-arterial thrombolysis: prediction of final infarct volume and clinical outcome. Stroke. 2001; 32: 2021–2028.
Wildermuth S, Knauth M, Brandt T, Winter R, Sartor K, Hacke W. Role of CT angiography in patient selection for thrombolytic therapy in acute hemispheric stroke. Stroke. 1998; 29: 935–938.
Eastwood JD, Lev MH, Wintermark M, Fitzek C, Barboriak DP, Delong DM, Lee TY, Azhari T, Herzau M, Chilukuri VR, Provenzale JM. Correlation of early dynamic CT perfusion imaging with whole-brain MR diffusion and perfusion imaging in acute hemispheric stroke. AJNR Am J Neuroradiol. 2003; 24: 1869–1875.