Identification of Infarct Core and Penumbra in Acute Stroke Using CT Perfusion Source Images

Patient Selection

We prospectively studied 42 consecutive patients with lesions in the middle cerebral artery (MCA) territory (April 2007-August 2008). Inclusion criteria were the following: arrival <9 hours from onset of stroke symptoms, male or female patients 18–80 years of age, NIHSS score ≥4, no history of stroke or premorbid symptoms, and modified Rankin Scale (mRS) score ≤2. Patients who satisfied these criteria were subsequently excluded for the following reasons: intracranial hemorrhages or brain tumors; NIHSS score ≥25; a history of contrast medium allergy or renal failure; serum glucose level <2.7 mmol/L or >22.2 mmol/L; and degradation of baseline or follow-up imaging examinations by poor imaging quality, including severe motion artifacts. The study protocol was approved by local institutional review board.

Imaging Acquisition

All non-contrast-enhanced CT, CTP, and CT angiography (CTA) examinations were performed on 16-section CT scanners (Somatom Volume Zoom; Siemens, Erlangen, Germany). The imaging protocol included a mandatory admission “one-stop shop” CT examination and a follow-up (2–7 days postadmission) CT/MR imaging of the brain. The admission CT examination included a whole brain contrast-enhanced CT, followed by a 2-level CTP, then a whole-brain CTA. First, the whole-brain contrast-enhanced CT was performed with a section thickness of 9 mm supratentorially and 4.5 mm infratentorially to exclude intracranial hemorrhage. As a follow-up, 2contiguous-section CTP examinations at the level of the basal ganglia/internal capsules and the upper portion of the lateral ventricles were performed, encompassing the anterior, posterior, and MCA territories. A bolus of 40-mL nonionic iodinated contrast agent (iobitridol, 300 mg/mL, Xenetix; Guerbet, Aulnay-sous-Bois, France) was administered into the antecubital vein at an injection rate of 8 mL/s (20-gauge intravenous cannula). Corresponding CT scans were initiated 4 seconds after the injection with the following acquisition parameters: 80 kVp, 209 mA, 1.0 s/rotation, 40-second scanning time, 12 -mm section thickness, and 40 images per section. Finally, a whole-brain CTA was performed with an additional 90 mL of contrast agent, which was injected at a rate of 5 mL/s. The acquisition parameters were 100 kVp, 322 mA, 1.0-mm section thickness, 10-second delay, and 6-second scanning time.

Follow-up imaging was performed on all patients, either with the same CT scanner or by using a 3T MR imaging scanner (Magnetom Trio; Siemens). Axial fluid-attenuated inversion recovery images (TR, 9000 ms; TE, 110 ms; TI, 2500 ms) were used to assess the presence and extent of infarction.

Image Postprocessing

CTP-SIs were loaded from the PACS of our institution onto an off-line workstation (5i3D; Neusoft Medical Systems, Shenyang, China) equipped with postprocessing software, which is a research version of the CTP software of Neusoft Medical Systems used to select arterial phase and venous phase CTP source images and generate CTP color maps.

Postprocessing was performed by 2 neuroradiologists (with 5 years' experience in generating and interpreting CTP) from our institution, and they had no other involvement in the study. The raters input the CTP-Sis into the CTP software. After the rater selects 1 section from the 2 axial sections on the CTP-SI, the software will automatically determine the maximum intensity projection (MIP) image (Fig 1A). On the MIP image, the time attenuation curve (TAC) (Fig 1B) was obtained from the anterior cerebral artery and superior sagittal sinuses. The scanning time was 40 seconds, so there were 40 seconds in the horizontal axis in the TAC, the number of time points corresponding to 1 image in each section. The arterial phase CTP-SI and venous phase CTP-SI were respectively obtained at peak points of the arterial (red) and venous TAC (blue) curves (Fig 1C). The other section would be processed the same way. The CTP color map was postprocessed with the same software, which produced CBF, CBV, mean transit time (MTT), and time-to-peak (TTP) maps.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

The arterial phase CT perfusion source image (CTP-SI) and the venous phase CTP-SI are respectively obtained at the peak points of the arterial (B, red curve) and venous time attenuation curves (TACs) (B, blue curve). On the maximum-intensity-projection image (A), the region of interest is on the area of the anterior cerebral artery (A, red circle); for the arterial TAC curve (B, red curve), the region of interest is on the area of the venous sinus (A, blue circle) for the venous TAC (B, blue curve). On the TAC (B), the horizontal axis is time (seconds) and the vertical axis is attenuation (Hounsfield units). There are 40 seconds in the horizontal axis, every second corresponding to 1 image in each section. With this patient as an example, the peak point of the arterial phase is 15 seconds after scanning begins (at the peak point of the red curve, B).The arterial phase CTP-SI is the fifteenth image in each section (C, left side); the peak point of the venous phase is 23 seconds after scanning begins at the peak point of blue curve (B). The venous phase CTP-SI is the twenty-third image in each section (C, right side).

Imaging Analysis

The arterial/venous phase source images can be reviewed directly at the CT workstation. Regions of relatively diminished contrast enhancement were scored as abnormal on the arterial phase and venous phase CTP-SIs. For the CBF and CBV maps, the raters visually evaluated each Alberta Stroke Program Early CT Score (ASPECTS) region for relatively low CBF and CBV compared with the mirror region in the contralateral hemisphere.

For quantification of the affected territory of the MCA, the methods described for the ASPECTS were used. In ASPECTS topography, the MCA territory was divided into 10 regions (caudate, lentiform nucleus, internal capsule, insula, and M1-M6).^13–15 The ASPECTSs of arterial phase and venous phase CTP-SIs were reviewed first, and CBF and CBV were interpreted at least 1 week later to limit potential recall bias. The raters assessed the imaging independently, blinded to clinical information, and did not know follow-up imaging results.

Statistical Analyses

The unit of analysis was the mean of the 2 raters' ASPECTSs. The relationship between the different imaging parameters (arterial phase CTP-SI and CBF ASPECTS, venous phase CTP-SI and CBV ASPECTS) was analyzed by the Pearson correlation coefficient, the concordance correlation coefficient, and Bland-Altman plots. The Wilcoxon signed rank test was used to assess the differences between the arterial phase CTP-SI and the venous phase CTP-SI. The relationship between the CTP-SI and follow-up imaging was analyzed by the Pearson correlation coefficient. To visually evaluate concordance between raters in assessing ASPECTS in different imaging parameters, we created scatterplots comparing the 2 raters' assessments, with regression lines and 95% limits of agreement; and we also calculated the concordance correlation coefficient by using Pearson statistics. Intraobserver agreements were analyzed by the Pearson correlation coefficient and the concordance correlation coefficient. A P value <.05 was statistically significant. The Statistical Package for the Social Sciences (Version 16.0, SPSS, Chicago, Illinois) was used for statistical analyses.