| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Depts of Radiology (J.D.E., D.P.B., J.M.P.) and Community and Family Medicine (D.M.D.), Duke University Medical Center, Box 3808, Durham, NC 27710-3808; Dept of Radiology, Massachusetts General Hospital, Boston (M.H.L.); Institute of Diagnostic and Interventional Radiology, Friedrich-Schiller-University Jena, Germany (T.A., C.F., M.H.); Imaging Research Laboratories, John P. Robarts Research Institute, London, Ontario, Canada (T.Y.L.); Dept of Diagnostic and Interventional Radiology, University Hospital, Lausanne, Switzerland (M.W., R.M.); and Dept of Radiology, Royal North Shore Hospital, St Leonards, Australia (D.B.). From the 2000 RSNA scientific assembly. Received Feb 15, 2001; revision requested Mar 26; revision received May 16; accepted June 20. Address correspondence to J.D.E. (e-mail: eastw004@mc.duke.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
MATERIALS AND METHODS: Twelve patients with acute middle cerebral artery stroke and 12 control patients were examined by using single-section computed tomography (CT) perfusion scanning. Analysis was performed with a deconvolution-based algorithm. Comparisons of mean CBF, cerebral blood volume (CBV), and mean transit time (MTT) were determined between hemispheres in all patients and between low- and normal-attenuation regions in patients with acute stroke. Two independent readers examined the images for extent of visually apparent regional perfusion abnormalities. The data were compared with extent of final infarct in seven patients with acute stroke who underwent follow-up CT or magnetic resonance imaging.
RESULTS: Significant decreases in CBF (-50%, P = .001) were found in the affected hemispheres of patients with acute stroke. Significant changes in CBV (-26%, P = .03) and MTT (+111%, P = .004) were also seen. Significant alterations in perfusion were also seen in low- compared with normal-attenuation areas. Pearson correlation between readers for extent of CBF abnormality was 0.94 (P = .001). Intraobserver variation was 8.9% for CBF abnormalities.
CONCLUSION: Deconvolution analysis of CT perfusion data is a promising method for evaluation of cerebral hemodynamics in patients with acute stroke.
Index terms: Brain, blood flow • Brain, CT, 13.12111, 13.12116, 13.12119, 17.12111, 17.12116, 17.12119 • Brain, infarction, 13.78 • Brain, perfusion, 17.12119
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Despite the substantial promise of dynamic CT perfusion scanning for the assessment of patients with stroke, the practicality of dynamic CT perfusion methods that have been previously reported has been limited in an important way. Prior methods of dynamic CT perfusion scanning have depended on the use of rapid intravenous infusions of contrast material (typically 10–20 mL/sec in an arm vein) (3,4,7–9). With a perfusion imaging method, dependence on such rapid rates of infusion represents a limitation because the use of rapid rates of infusion requires the routine use of large venous cannulas, such as the 14-gauge catheters used in the study by Koenig et al (7). Many patients with acute stroke (eg, elderly patients and patients with small or fragile veins) do not have suitable venous access for such rapid infusions. Use of slower injection rates might also be safer than use of rapid injection rates.
Recently, a new CT perfusion method has been developed that permits infusion rates that are substantially slower (eg, 4 mL/sec) than those required previously. This method is based on the central volume principle of cerebral hemodynamics (10), and a mathematic operation called deconvolution is used (11,12). This method of analysis of CT perfusion scanning has been validated in animal studies (2,13,14). However, to our knowledge, there have been few published data related to use of this method in humans (13,14).
The purpose of this study was to test three basic hypotheses about deconvolution-derived CT perfusion scanning in patients with acute stroke. The first hypothesis was that for patients with acute middle cerebral artery (MCA) stroke, mean cerebral blood flow (CBF) values within affected (ie, ischemic) territories would be lower than mean CBF values in nonischemic territories. The second hypothesis was that mean CBF values within regions of low attenuation on unenhanced CT scans would be lower than mean CBF values within normal-attenuation regions found adjacent to low-attenuation regions within the same affected hemispheres. The third hypothesis was that, as with previously described methods of hemodynamic imaging, CT perfusion scanning with deconvolution-based software would provide greater information about the extent of ischemic brain tissue than unenhanced CT scanning.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Patients with Acute Stroke
Twelve patients (eight women, four men; mean age, 64.8 years ± 16.8 [SD]; age range, 30–87 years) with acute stroke symptoms in the territory of one MCA were retrospectively identified from the group of patients who underwent scanning and who had acute stroke symptoms. Stroke in the territory of the MCA was specifically chosen for study because the high incidence, morbidity, and mortality associated with MCA stroke have established it as a major stroke subtype. For the purposes of this study, acute stroke was defined as stroke symptoms of fewer than 2 days duration, though most patients had a symptom duration of less than 1 day. In all cases, acute MCA stroke was clinically diagnosed. In all patients with acute stroke, regions of abnormality, which were indicated by either hypoattenuation on the initial or follow-up unenhanced CT scan or hyperintensity on follow-up T2-weighted magnetic resonance (MR) images, consistent with MCA infarction were identified within the territories of the affected (ie, symptomatic) MCA.
In all but one case, the described stroke was, according to available medical history and imaging findings, the first-ever large-vessel cerebral ischemic event. In 10 cases, CT perfusion scans were obtained within 24 hours of symptom onset, and in one case, they were obtained at 27 hours after symptom onset. One patient was initially found unresponsive at home, and the time of onset of ischemia was not known with certainty but ranged between 12 and 24 hours. For the 11 patients in whom the time of onset of symptoms was known, mean time from onset of symptoms to CT perfusion scanning was 6.4 hours (median, 5 hours; minimum, 1.5 hours).
Control Patients
Twelve patients (eight women, four men; mean age, 61.8 years ± 18.2; age range, 21–83 years) without symptoms of acute stroke were retrospectively identified from the larger group of patients without acute symptoms. This number of patients was chosen so that the number of control patients would be the same as the number of patients with acute stroke who were examined. Age was the sole criterion for selecting individual control patients for inclusion in the control group. Age was used as a criterion because the objective was to establish a control group that had a similar mean age as the group with acute stroke. (CBF is known to decrease with age, and it was thought that controlling for mean age would help to make the study more accurate.) For each patient with acute MCA stroke, the patient without acute stroke who was closest in age to the patient with acute stroke was selected for inclusion in the control group. Patients with acute stroke and control patients, however, were not paired for the analysis.
In the control patients, the following conditions were diagnosed: chronic stroke (n = 5), asymptomatic carotid artery stenosis (n = 2), transient ischemic attack (TIA) (n = 2), amaurosis fugax (n = 1), carotid artery dissection (n = 1), and recent traumatic subarachnoid hemorrhage. The patient with carotid artery dissection had a history of TIA that was treated with endoluminal stent placement and antiplatelet medications. That patient was asymptomatic in the 6 months preceding CT perfusion scanning. One of the two control patients who had a recent TIA was symptomatic at the time of CT perfusion scanning but did not have evidence of vascular occlusion or of a brain abnormality on initial or follow-up unenhanced CT scans. The other patient with a medical history of TIA was not symptomatic at CT perfusion scanning. In summary, the control group included mostly patients with cerebrovascular disease, and control patients had a mean age similar to that of the group with acute stroke.
Scanning Technique
All patients with acute stroke first underwent unenhanced CT scanning including either 3- or 5-mm-thick transverse sections through the posterior fossa and either 5- or 10-mm-thick transverse sections through the supratentorial region. In nine control patients, unenhanced CT scanning was not performed at CT perfusion scanning, but these patients had undergone unenhanced CT or MR examinations of the brain in the preceding 12 months.
CT angiographic examinations of the circle of Willis were performed in all subjects prior to (typically, 5–10 minutes before) CT perfusion scanning by using helical scanning and a beam collimation of either 1- or 1.5-mm thickness. In six patients with acute stroke and six control patients, CT angiographic examinations included coverage of the cervical vessels and the circle of Willis. The total volume of contrast material infused for the CT angiographic examinations varied between 100 and 150 mL (mean, 113 mL) for patients with stroke and was 150 mL for control patients. Iodine concentration of the contrast material was either 300 mg/dL in eight patients with acute stroke and one control patient or 370 mg/dL in four patients with acute stroke and 11 control patients. For all subjects, the same type of contrast material that was used for the CT angiographic examination was used for the CT perfusion examination.
For CT perfusion scans, repeated scanning of either one 10-mm section (single-section CT scanners) or two adjacent 10-mm sections (multisection CT scanners) was performed. For all scanning, a 25-cm field of view was used, and scans were reconstructed with a matrix of 512 x 512 pixels. Scans were obtained in the transverse plane, and the level examined included the basal ganglia in 10 patients with acute stroke and in all control patients. In one patient with acute stroke, the level chosen was the cerebral convexity above the level of the ventricles; in another patient with acute stroke, the level chosen was immediately below the basal ganglia. Two different x-ray generation techniques were used, though each patient was examined by using only one of the two techniques. With one technique, 140 kVp with 40 mA for 40 seconds was used in three patients with stroke and two control patients. With the other technique, 80 kVp with 200 mA for 45 seconds was used in nine patients with stroke and 10 control patients. Most patients were scanned with 80 kVp because this peak kilovoltage has been shown to provide greater contrast, and the dose is lower when 80 kVp is used instead of 140 kVp (15).
In all but one case, continuous (cine) acquisition was used. In one case, 25 transverse scans (ie, standard noncontinuous scans), each of 1-second scanning duration, were obtained every 2 seconds for a total examination time of 50 seconds. For the case in which the transverse (noncontinuous) method was used, an x-ray technique with 80 kVp and 200 mAs was used. In none of the cases was the table of the CT scanner moved during CT perfusion scanning. The mean volume of infused contrast material was 43 mL (range, 40–50 mL) for patients with acute stroke. A volume of 40 mL of contrast material was used in 11 control patients, and a volume of 45 mL was used in one control patient. In six patients with acute stroke and in all control patients, an infusion rate of 4 mL/sec was used. In two patients with acute stroke, a rate of 5 mL/sec was used, and in four patients with acute stroke, a rate of 10 mL/sec (mean infusion rate for patients with MCA stroke, 6.2 mL/sec) was used.
Analysis of Scans
Unenhanced CT scans.—Unenhanced CT scans were reviewed by two experienced neuroradiologists (J.D.E., D.P.B.) together on an imaging workstation (Advantage Windows; GE Medical Systems, Milwaukee, Wis) by using manually adjusted window width (10 HU) and window level (35 HU) settings to optimize the detection of signs of acute stroke (16). For each case, an agreement was reached concerning the presence and location of the following findings: increased attenuation within a cerebral vessel (eg, MCA), obscuration of the insular ribbon or basal ganglia, and regional decreased tissue attenuation.
CT angiograms.—CT angiograms were evaluated by one neuroradiologist (J.D.E.) with the imaging workstation and commercial software (REFORMAT; GE Medical Systems) that provided three-dimensional reconstruction of the data in the transverse, coronal, and sagittal planes. The interval between reading of the unenhanced CT scans and of the CT angiograms was 4 weeks (an interval used to limit recall). The presence and locations of arterial stenoses were recorded for the following arteries: internal carotid artery, MCA, anterior cerebral artery, basilar artery, and posterior cerebral artery. Arterial stenoses were graded as mild (<50%), moderate (50%–70%), severe (>70%), or occluded.
CT perfusion scans.—All CT perfusion scans were analyzed by using an imaging workstation (Advantage Windows; GE Medical Systems) with commercial CT perfusion analysis software (CT PERFUSION; GE Medical Systems) to create maps of CBF, cerebral blood volume (CBV), and mean transit time (MTT). For the patients in whom two 10-mm-thick sections were used (n = 6), only the more cephalad section was analyzed to maintain a uniform number of sections in all subjects. The software required placement of small regions of interest (ROIs) on one artery and one vein to generate arterial input functions and venous outflow functions, respectively, for the deconvolution analysis. This process is required because the computer algorithm for the deconvolution method is used to compare the shape and height of the time-attenuation curve of each pixel of the CT time series with the shape and height of the arterial and venous time-attenuation curves to determine CBF and CBV. MTT was computed by using the central volume principle that defines CBF as the ratio of CBV to MTT (Appendix).
Both readers who examined the CT perfusion scans (J.D.E., D.P.B.) individually placed ROIs on the arteries and veins at the time that the scans were evaluated by using the same protocol. ROIs were 3–7 mm2 in both the artery and the vein. For most patients, the larger of the two anterior cerebral arteries was chosen for placement of the ROI that provided the arterial input function, and the superior sagittal sinus was chosen for placement of the ROI that provided the venous outflow function. These vessels were chosen because they are reliably identified in most subjects and because their courses run nearly perpendicular to the transverse plane of section used in CT scanning of the brain. It was assumed that by choosing vessels that typically are oriented perpendicular to the transverse plane of section, errors due to volume-averaging artifacts would be decreased.
For those patients (n = 2) and control patients (n = 1) for whom the superior sagittal sinus was not well visualized within the plane of section on the CT scans, another large venous channel, such as the transverse sinus or straight sinus, was selected for placement of the ROI that represented the venous outflow function. In one patient with acute stroke and one control patient, the internal carotid artery on the unaffected (ie, nonischemic) side was chosen to be the arterial input function because the anterior cerebral artery was not well visualized within the plane of section on the CT scan. In one patient with acute stroke, a branch of the MCA on the unaffected side within the sylvian fissure (ie, an M2 branch) was chosen to be the arterial input function.
Because cardiac function can influence measured CBF, the cardiac function of patients in this study was evaluated with measurement of the interval between the time of initiation of injection and the time of bolus arrival in the artery selected for the arterial input function ROI. This time to arrival was recorded, and mean values were computed for both the patients with acute stroke and the control patients.
Part 1: Hypothesis 1—Comparison of Hemispheres
For this part of the investigation, a neuroradiologist (J.D.E.) drew two types of ROIs by hand to delimit regions corresponding to the expected territories of supply of the MCAs bilaterally. The first type of ROI that was drawn included the full expected extent of supply of the MCA territory seen on the CT scan. This first type of ROI included the cortical gray matter, subjacent white matter, and basal ganglia territories typically supplied by the MCA. The second type of ROI included solely the basal ganglia. The basal ganglia regions were prospectively chosen for a separate analysis to provide information about a region within the MCA territory that did not contain large cortical blood vessels.
It was assumed that the inclusion of large cortical blood vessels in the total MCA territories might affect the mean values of the perfusion parameters and that examination of the basal ganglia regions separately could be used to control for this potential source of error. Thus, for this part of the investigation, four ROIs were drawn, two on each side (Fig 1), in every stroke patient and control patient. Next, the perfusion maps were created in the following order: CBV, CBF, and MTT. After each perfusion map was created, the mean value of that perfusion parameter (eg, CBF) contained within each of the four ROIs was recorded.
|
|
|
In each patient, the perfusion maps were always created and examined by the readers in the following order: CBV, CBF, and MTT. This was done to prevent bias caused by seeing the MTT map first (the MTT map has been previously reported [1] to typically show the largest area of abnormality). For each of the three perfusion parameters examined, ranges of values were selected a priori to represent regions with an increased probability of having an ischemic abnormality. The range of CBF values chosen to represent ischemia was 0–10 mL/100 g/min, because CBF values in this range have been shown in previous human and experimental studies (2,17) to be consistent with severe ischemia. The range of CBV values chosen was 0–1.5 mL/100 g. This range was chosen to represent tissue with substantial oligemia (approximately half or less of the mean CBV in unaffected hemispheres). The range of MTT values chosen was all values greater than 6 seconds. This range represented tissue with MTT values at least 3 SDs greater than the mean MTT value in control patients in this study.
The software used in this investigation allowed preselected ranges to be displayed as a single color (eg, blue for CBF values between 0 and 10 mL/100 g/min). In this way, it was possible to visually examine the perfusion maps for regions of abnormality by using a color-coded map that allowed regions of severe ischemia to be conspicuous.
One of the readers (J.D.E.) performed an additional 45 measurements in a subset of patients with acute stroke. The intraobserver variation in extent for each type of perfusion map (eg, CBF) was computed as the mean per-patient value of the SD of the measurements of extent for that parameter divided by the mean value of extent for that parameter.
Determination of Final Infarct Extent
For each patient with acute stroke in whom follow-up CT or MR imaging findings of the brain were available, final infarct extent was determined by one of three radiologists (J.D.E., T.A., M.W.), each of whom used the same protocol and workstation. Final infarct extent was measured in two ways. First, the area of final infarction within the plane of the CT perfusion scan that was obtained during initial evaluation was determined. Second, the total infarct volume, taking into account sections outside the plane of the CT perfusion scan, was determined. For each section examined, the area contained within a hand-drawn ROI that delimited the regions of low attenuation (CT scans) or high signal intensity (MR images) was noted and recorded. For the determination of total infarct volume, the areas of abnormality in each section were multiplied by the thickness of each section plus any intersection gap, except for the lowest section, where no intersection gap was added to the section thickness.
Statistical Analysis
Two-tailed paired t tests were used to compare mean perfusion parameter values in affected hemispheres with those in nonaffected hemispheres in patients with acute stroke. These tests also were used to compare perfusion parameter values within the right MCA territories with those values in the left MCA territories in control patients. Two-tailed paired t tests were also used to compare the sizes of depicted abnormalities seen on different types of perfusion parameter maps (eg, size of the CBF abnormality vs size of the CBV abnormality). Unpaired t tests were used to compare mean parameter values in the patients with acute stroke with values in control patients. Pearson correlation coefficients were used to compare the extents of regional perfusion abnormalities seen by each reader with (a) the extents of perfusion abnormalities seen by the other reader and with (b) the final infarct extent in those patients who underwent follow-up scanning. The 
coefficients for the agreement of the two readers for the presence of visually seen abnormalities were also computed (SAS; SAS Institute, Cary, NC) for each of the perfusion parameters.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
|
|
Analysis of CT Perfusion Scans
Part 1: Comparison of mean perfusion parameter values measured within MCA territories.—All patients with acute stroke had lower CBF values on affected sides than on unaffected sides. Eleven patients had greater MTT values on affected sides than on unaffected sides. Ten patients had lower CBV values on affected sides, and two patients had greater CBV values on affected sides. Considering the whole group, statistically significant decreases in mean CBF and mean CBV values and statistically significant increases in mean MTT values were on the sides affected by acute stroke within the ROIs that contained the full MCA territories together.
Decreases in mean CBF and mean CBV values and increases in mean MTT values were also measured in the ROIs containing solely the basal ganglia. The changes in mean CBF value and mean MTT value that were measured in the basal ganglia regions were also statistically significant. However, the decrease in mean CBV value that was measured in the basal ganglia regions was not statistically significant. The data (mean values and SDs) are summarized in Table 1. The mean time of arrival of the bolus of contrast material in the reference (normal) arteries in patients with acute stroke was 13.4 seconds ± 3.2 (range, 9–19 seconds).
|
Comparison of mean perfusion parameter values in the control patients with those in the unaffected hemispheres in patients with acute stroke revealed no statistically significant differences. No statistically significant differences in mean perfusion parameter values were determined with respect to sex.
Part 2: Comparison of low- with normal-attenuation regions in affected hemispheres of patients with acute stroke.—Seven patients with acute stroke had regions of low attenuation within the plane of section on the CT perfusion scan. The mean CBF value in regions of low attenuation was 13.1 mL/100 g/min ± 8.4 compared with the mean CBF value of 31.6 mL/100 g/min ± 12.4 in adjacent regions of normal attenuation on the same sides (41%, P = .004). The mean CBV value in regions of low attenuation was 0.9 ± 0.4 compared with the mean CBV value of 2.3 ± 0.7 in normal-attenuation areas (40%, P = .003). The mean MTT value in regions of low attenuation was 5.3 seconds ± 1.2 compared with 6.5 seconds ± 2.3 in normal-attenuation areas (P = .10).
Part 3: Analysis of visually apparent regional CT perfusion imaging abnormalities.—The overall quality of perfusion maps was considered to be good by both readers, with no appreciable differences seen between maps obtained in patients with acute stroke and those obtained in control patients. No differences in perfusion map quality were apparent between the maps of those obtained in the four patients in whom an infusion rate of 10 mL/sec was used and those obtained in the remaining patients in whom an infusion rate of either 4 or 5 mL/sec was used.
Regions of CBV abnormality were found by both readers in seven patients with acute stroke. Two patients with acute stroke had no regional CBV abnormality detected by either reader; regional abnormalities of CBV were initially detected by only one of two readers in three patients with acute stroke. In two of these three patients, both readers ultimately agreed that abnormalities were present. In the third case, the readers ultimately agreed that no abnormality was present. No regional CBV abnormality was detected by either of the readers in any of the 12 control patients. In patients with acute stroke, the mean extent of regional CBV abnormalities was 1,117 mm2 ± 1,583.
Regions of CBF abnormality were found by both readers in 10 patients with acute stroke. In one patient with acute stroke, a regional CBF abnormality was not seen by either of the readers, and in one patient with acute stroke, a regional CBF abnormality was initially identified by one of the readers. Both readers ultimately agreed that the abnormality was present in that case. No regional abnormality of CBF was found by either reader in any of the control patients. For patients with acute stroke, the mean extent of regional CBF abnormality was 1,518 mm2 ± 1,669.
Regions of MTT abnormality were found by both readers in all 12 patients with acute stroke. No regional abnormality of MTT was detected by either of the readers in 10 control patients. Regional abnormality of MTT was seen by both readers in one control patient. Regional abnormality of MTT was initially seen by only one reader in a second control patient. For this patient, the readers ultimately agreed that an abnormality was present. For patients with acute stroke, the mean extent of regional MTT abnormality was 3,075 mm2 ± 1,508. In two control patients in whom regional MTT abnormalities were found, the mean extent of abnormalities was 453 mm2.
Inter- and intraobserver reproducibility were measured. A statistically significant correlation was found between two independent readers for the extent of regional CBV abnormality, CBF abnormality, and MTT abnormality. The coefficients were 0.73, 0.83, and 0.92 for CBV, CBF, and MTT abnormalities, respectively. The mean intraobserver variability values for extent of CBV, CBF, and MTT abnormalities were 10.1%, 8.9%, and 7.6%, respectively. The perfusion abnormality extents for the two readers are shown in Figure 4.
|
Correlation of the extents of regional perfusion abnormalities with final infarct extent was determined. In seven patients with acute stroke, follow-up CT scans or MR images were available for review. The mean time to follow-up imaging was 98.5 hours (range, 24–142 hours). The mean area of final infarction within the plane of section on the original CT perfusion scan was 13.3 cm2 ± 14.5. The mean final infarct volume was 66.7 cm3 ± 91.8. The correlation data used to compare the extents of visually apparent regional perfusion abnormalities with final infarct extents are summarized in Table 2.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Because of the advantages (speed, availability, and the ability to depict intracerebral hemorrhage) that CT, compared with other imaging modalities, offers for the examination of patients with stroke, development of a comprehensive CT-based imaging protocol for stroke would represent an important advancement. A successful CT-based protocol that includes unenhanced CT, CT angiography, and CT perfusion scanning could provide all information necessary prior to thrombolytic therapy, which would obviate a second study such as MR imaging and, thus, save time.
A number of methods for performing perfusion imaging studies with CT are presently available. However, compared with other CT-based methods, dynamic CT perfusion scanning offers advantages. First, compared with xenon-enhanced CT, dynamic CT perfusion scanning can be performed with technology and equipment that are available in most radiology departments. Second, findings of dynamic CT perfusion scanning can provide information about multiple hemodynamic parameters (eg, CBF, CBV, time to peak enhancement, or MTT) from one examination (3,7,8). Findings in previous studies (1,21,22) have shown that comparison of multiple perfusion parameters can be used for determining the risk of infarction in viable ischemic tissue. Compared with dynamic CT perfusion scanning, other CT perfusion methods may provide information about a single perfusion parameter. For example, xenon-enhanced CT provides information about only CBF. CT perfusion scanning that is based on subtraction of unenhanced CT scans from contrast material–enhanced CT scans (23) provides information about only CBV.
Until recently, use of dynamic CT perfusion scanning has been limited by use of analytic methods that depend on the use of rapid infusion rates. For example, the method used in the studies of Klotz and König (4) and Koenig et al (7) depends on the accurate measurement of the rate of upslope of the tissue time-attenuation curve to compute CBF. Use of slower infusion rates (and thus, use of longer infusion times) would violate an important underlying assumption of this method (ie, no venous outflow).
Deconvolution-based analysis of dynamic CT perfusion examinations is a method that can be used to rapidly provide maps of CBF, CBV, and MTT from data acquired by using infusion rates substantially slower (eg, 4 mL/sec) than those used in other dynamic CT perfusion scanning methods. In our experience, infusion rates of 4 mL/sec may be safely accomplished with either 18- or 20-gauge antecubital venous catheters.
Although the deconvolution method of analysis was originally described for use with CT by Axel (11) in 1980, the method was then limited in its usefulness because of limitations in hardware and lack of a deconvolution algorithm that was stable with regard to noise. Recent developments in hardware and numeric algorithms of analysis (ie, software) have substantially improved this method for use in the clinical environment. The advent of slip-ring CT scanner technology has enabled more rapid scanning times, which has increased temporal resolution and, thus, permitted more accurate MTT and CBF measurements. Two important recent advances in software technology have also made the method more practical for routine use. First, Cenic et al (13) improved the stability of the method with regard to noise. Second, a deconvolution algorithm (CT PERFUSION; GE Medical Systems), based on the method of Cenic et al but substantially faster, is now available. Use of this algorithm reduces analysis time to as little as 5 minutes, which makes the method potentially relevant for use in clinical decision making.
By validating our first hypothesis that decreased CBF would be found in territories affected by ischemia, findings in this study showed that the deconvolution-based method of CT perfusion scanning used in this investigation is helpful in the detection of diminished blood flow within affected MCA territories. The mean absolute CBF value of 13.1 mL/100 g/min ± 8.4 found in this study within low-attenuation MCA territories is similar to previously published (24) mean CBF values of 15 mL/100 g/min ± 8.1 in patients with MCA stroke by using xenon-enhanced CT. In this study, the mean CBF value measured in the basal ganglia gray matter in control patients (59.6 mL/100 g/min) was greater than the mean CBF value of 42.4 mL/100 g/min reported for the basal ganglia location by Leenders et al (25) in a study in which positron emission tomography (PET) was used. It was also greater than the mean CBF value for cortical gray matter of 48.5 mL/100 g/min reported by Hagen et al (26) in a study in which xenon-enhanced CT was used. Future studies in which CT perfusion scanning is compared with other methods may help to determine if a scaling factor is needed to interpret CBF values obtained with dynamic CT perfusion scanning.
Our finding that measured CBV values were decreased on the affected sides in most but not all patients with acute stroke was an expected finding. Prior studies (21,22) have shown that CBV may be either increased or decreased in acute stroke. This likely depends on the degree and sufficiency of autoregulatory vasodilation.
By validating our second hypothesis that decreased CBF values would be found in low-attenuation regions of the brain, this study provided evidence that CT perfusion scanning with deconvolution analysis could depict regional differences in blood flow between tissues of differing degrees of ischemia. Although we are not aware of prior studies that are directly comparable with ours, Grond et al (27) have shown that low-attenuation regions on initial CT scans correlate with severe regional hypoperfusion on PET scans.
The dynamic CT perfusion scanning described in this investigation provided greater sensitivity than did unenhanced CT. This result validated our third hypothesis. The perfusion studies in all patients with acute stroke showed visually detected MTT abnormalities that were seen by two independent readers. By means of comparison, three of five patients with acute stroke had no acute abnormality or only a minimal abnormality (eg, insular ribbon obscuration), which was seen on unenhanced CT scans; this was consistent with acute large-vessel stroke.
For the comparison of the sizes of the visually apparent abnormalities of MTT, CBF, and CBV, the results of this study (with MTT being largest and CBV the smallest) agree with those of prior MR imaging studies (1,27) in which the relative sizes of CBF, CBV, and MTT abnormalities in patients with acute stroke were measured.
The finding of regional MTT abnormalities in two control patients in this study is notable. One control patient without infarction in whom a regional MTT abnormality was seen had internal carotid artery occlusion on the same side as the MTT abnormality. The observed prolongation of MTT was thought to be most likely caused by collateral flow. The other control patient with a regional MTT abnormality experienced reversible focal symptoms referable to the hemisphere in which the MTT abnormality was seen. Because no arterial occlusion was seen at CT angiography, the presence of a small embolus (that may have subsequently undergone lysis) is likely. No CBV or CBF regional abnormalities were seen in either of these two patients. These findings suggest that the specificity of regional decreases in CBF and CBV at CT perfusion scanning is higher than that of increases in MTT. Further studies are indicated to determine whether changes in CBV and CBF are more specific for determination of stroke than are changes in MTT.
Assessment of the reproducibility of measurements of two independent readers sought to simulate a clinical situation in which CT perfusion scans were presented to readers with approximately equal probability (ie, about 50%) that a given scan would be that of a patient with acute stroke or that of a control patient. In this study, increased interobserver agreement for the presence and extent of perfusion abnormalities was found. Along with the decreased intraobserver variability in this study, this finding suggests that CT perfusion scanning provides results that are reproducible. Increased interreader and intrareader reproducibility is a necessary requirement for a clinically useful method.
Correlation with final infarct extent was possible in slightly greater than half of the patients with acute stroke. Regional perfusion abnormalities on the single-section CT perfusion scans correlated well with final infarct size. Future studies of CT perfusion scanning may help to address the issue of which perfusion parameters are the best predictors of (a) tissue and (b) clinical outcomes.
Although the results of this study are encouraging, the study is limited in a number of ways that will need to be addressed in future trials with use of this method. First, because the present study was retrospective in nature, the scope of this study was necessarily limited. Second, it is important that future clinical studies of CT perfusion scanning in patients with stroke provide correlation with clinical stroke severity assessment (eg, National Institutes of Health Stroke Scale [28]) at presentation and at follow-up (eg, 90 days). Third, future studies should include scheduled follow-up CT or MR imaging for correlation of findings with final infarct extent. Finally, formal comparison of this method with another established method of cerebral perfusion imaging (eg, MR with diffusion and perfusion imaging, xenon-enhanced CT, or PET) should be undertaken to validate the method in humans.
One current technical disadvantage of dynamic CT perfusion scanning is limitation of coverage to either a single 10-mm section (single-section CT scanners) or two adjacent 10-mm sections (multisection CT scanners), with the associated possibility that ischemic regions located outside the chosen level could be missed. This limitation could be addressed, in part, by routinely examining two different levels during each examination.
In conclusion, dynamic CT perfusion scanning with deconvolution analysis is a method that has important practical advantages over previously available methods, including use of slower infusion rates. Validation of the hypotheses and measurement of reproducibility in this study indicate that CT perfusion scanning with deconvolution analysis is a promising tool for the examination of patients with stroke. Our findings provide support for further study of this method with prospective trials.
|
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Deconvolution Analysis
Deconvolution analysis, the method used in this study, represents an alternative means of measuring CBF in which the outflow of the tracer from the brain is explicitly considered in the calculation. If one considers a network of capillaries in a certain mass of brain tissue, then CBF is determined in units of F(mL · min-1 · 100 g-1), and contrast media concentration, in units of Ca(t) (mmol · mL-1). The tissue concentration function, or the tissue residue function, Q(t) (mmol · mL-1), can be measured by using dynamic CT. In the special case when F · Ca is a delta function such that a unit mass of contrast media is deposited in the tissue instantaneously at time 0, then the tissue residue function becomes the impulse residue function, or R(t) (32). For a tracer that remains intravascular, such as contrast media in the brain, the impulse residue function is of the general form shown in Figure A1. The length of the initial plateau at unity height corresponds to the minimum time required for the blood to traverse the network from the arterial inlet to the venous outlet, or the minimum transit time.
When contrast medium is injected intravenously at a peripheral vein, the rate of delivery of the tracer to the capillary network is calculated with the following formula: F · Ca(t). If the mass of contrast medium in the network is linear with respect to the arterial (input) concentration and F is constant in time, then with linear superimposition, the following equation can be calculated:
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: CBF = cerebral blood flow, CBV = cerebral blood volume, MCA = middle cerebral artery, MTT = mean transit time, ROI = region of interest, TIA = transient ischemic attack
Author contributions: Guarantor of integrity of entire study, J.D.E.; study concepts and design, all authors; literature research, J.D.E., T.Y.L., M.W., R.M.; clinical studies, J.D.E., M.H.L., T.A., D.P.B., C.F., M.H., M.W., R.M., D.B., J.M.P.; data acquisition, all authors; data analysis/interpretation, J.D.E., J.M.P., D.M.D.; statistical analysis, J.D.E., D.M.D.; manuscript preparation, J.D.E.; manuscript editing, J.M.P., J.D.E.; manuscript definition of intellectual content, revision/review, and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Wintermark, H. A. Rowley, and M. H. Lev Acute Stroke Triage to Intravenous Thrombolysis and Other Therapies with Advanced CT or MR Imaging: Pro CT Radiology, June 1, 2009; 251(3): 619 - 626. [Full Text] [PDF]
|
|
|
|
 |
|
 A.A. Konstas, G.V. Goldmakher, T.-Y. Lee, and M.H. Lev Theoretic Basis and Technical Implementations of CT Perfusion in Acute Ischemic Stroke, Part 2: Technical Implementations AJNR Am. J. Neuroradiol., May 1, 2009; 30(5): 885 - 892. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Saver Proposal for a Universal Definition of Cerebral Infarction Stroke, November 1, 2008; 39(11): 3110 - 3115. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.M. Provenzale, K. Shah, U. Patel, and D.C. McCrory Systematic Review of CT and MR Perfusion Imaging for Assessment of Acute Cerebrovascular Disease AJNR Am. J. Neuroradiol., September 1, 2008; 29(8): 1476 - 1482. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Lin, K.S. Kazmi, M. Law, J. Babb, N. Peccerelli, and B.K. Pramanik Measuring Elevated Microvascular Permeability and Predicting Hemorrhagic Transformation in Acute Ischemic Stroke Using First-Pass Dynamic Perfusion CT Imaging AJNR Am. J. Neuroradiol., August 1, 2007; 28(7): 1292 - 1298. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G TAN and T GODDARD Neuroimaging applications of multislice CT perfusion Imaging, June 1, 2007; 19(2): 142 - 152. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. C. Sanelli, G. Nicola, A. J. Tsiouris, I. Ougorets, C. Knight, B. Frommer, S. Veronelli, and R. D. Zimmerman Reproducibility of Postprocessing of Quantitative CT Perfusion Maps Am. J. Roentgenol., January 1, 2007; 188(1): 213 - 218. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Chen, M.-H. Shyr, T.-Y. Chen, H.-Y. Lai, C.-C. Lin, and P.-S. Yen Dynamic CT Perfusion Imaging with Acetazolamide Challenge for Evaluation of Patients with Unilateral Cerebrovascular Steno-Occlusive Disease. AJNR Am. J. Neuroradiol., October 1, 2006; 27(9): 1876 - 1881. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Srinivasan, M. Goyal, F. A. Azri, and C. Lum State-of-the-Art Imaging of Acute Stroke RadioGraphics, October 1, 2006; 26(suppl_1): S75 - S95. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Kielstein, F. Donnerstag, S. Gasper, J. Menne, A. Kielstein, J. Martens-Lobenhoffer, F. Scalera, J. P. Cooke, D. Fliser, and S. M. Bode-Boger ADMA Increases Arterial Stiffness and Decreases Cerebral Blood Flow in Humans Stroke, August 1, 2006; 37(8): 2024 - 2029. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R.G. Gonzalez Imaging-guided acute ischemic stroke therapy: From "time is brain" to "physiology is brain". AJNR Am. J. Neuroradiol., April 1, 2006; 27(4): 728 - 735. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wintermark, A. E. Flanders, B. Velthuis, R. Meuli, M. van Leeuwen, D. Goldsher, C. Pineda, J. Serena, I. v. d. Schaaf, A. Waaijer, et al. Perfusion-CT Assessment of Infarct Core and Penumbra: Receiver Operating Characteristic Curve Analysis in 130 Patients Suspected of Acute Hemispheric Stroke Stroke, April 1, 2006; 37(4): 979 - 985. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K W Muir, H M Halbert, T A Baird, M McCormick, and E Teasdale Visual evaluation of perfusion computed tomography in acute stroke accurately estimates infarct volume and tissue viability J. Neurol. Neurosurg. Psychiatry, March 1, 2006; 77(3): 334 - 339. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P.W. Schaefer, L. Roccatagliata, C. Ledezma, B. Hoh, L.H. Schwamm, W. Koroshetz, R.G. Gonzalez, and M.H. Lev First-Pass Quantitative CT Perfusion Identifies Thresholds for Salvageable Penumbra in Acute Stroke Patients Treated with Intra-arterial Therapy AJNR Am. J. Neuroradiol., January 1, 2006; 27(1): 20 - 25. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Kan, S. Kobayashi, S. Phongkitkarun, and C. Charnsangavej Functional CT Quantification of Tumor Perfusion after Transhepatic Arterial Embolization in a Rat Model Radiology, October 1, 2005; 237(1): 144 - 150. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wintermark, M. Sesay, E. Barbier, K. Borbely, W. P. Dillon, J. D. Eastwood, T. C. Glenn, C. B. Grandin, S. Pedraza, J.-F. Soustiel, et al. Comparative Overview of Brain Perfusion Imaging Techniques Stroke, September 1, 2005; 36(9): e83 - e99. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Rumboldt, R. Al-Okaili, and J. P. Deveikis Perfusion CT for Head and Neck Tumors: Pilot Study AJNR Am. J. Neuroradiol., May 1, 2005; 26(5): 1178 - 1185. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Feng, C. L. Dumoulin, S. Dashnaw, R. D. Darrow, R. L. DeLaPaz, P. L. Bishop, and J. Pile-Spellman Feasibility of Stent Placement in Carotid Arteries with Real-time MR Imaging Guidance in Pigs Radiology, February 1, 2005; 234(2): 558 - 562. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S F S Halpin Brain imaging using multislice CT: a personal perspective Br. J. Radiol., December 1, 2004; 77(suppl_1): S20 - S26. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. H. Dennis Cheong, C. C. Tchoyoson Lim, and T. S. Koh Dynamic Contrast-enhanced CT of Intracranial Meningioma: Comparison of Distributed and Compartmental Tracer Kinetic Models--Initial Results Radiology, September 1, 2004; 232(3): 921 - 930. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wintermark, G. van Melle, P. Schnyder, J.-P. Revelly, F. Porchet, L. Regli, R. Meuli, P. Maeder, and R. Chiolero Admission Perfusion CT: Prognostic Value in Patients with Severe Head Trauma Radiology, July 1, 2004; 232(1): 211 - 220. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. G. Hoeffner, I. Case, R. Jain, S. K. Gujar, G. V. Shah, J. P. Deveikis, R. C. Carlos, B. G. Thompson, M. R. Harrigan, and S. K. Mukherji Cerebral Perfusion CT: Technique and Clinical Applications Radiology, June 1, 2004; 231(3): 632 - 644. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Kealey, V. A. Loving, D. M. Delong, and J. D. Eastwood User-defined Vascular Input Function Curves: Influence on Mean Perfusion Parameter Values and Signal-to-Noise Ratio Radiology, May 1, 2004; 231(2): 587 - 593. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Burdette Is CT Perfusion Ready for Prime Time? AJNR Am. J. Neuroradiol., January 1, 2004; 25(1): 3 - 4. [Full Text] [PDF]
|
|
|
|
|
|
D. Fiorella, J. Heiserman, E. Prenger, and S. Partovi Assessment of the Reproducibility of Postprocessing Dynamic CT Perfusion Data AJNR Am. J. Neuroradiol., January 1, 2004; 25(1): 97 - 107. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Provenzale, R. Jahan, T. P. Naidich, and A. J. Fox Assessment of the Patient with Hyperacute Stroke: Imaging and Therapy Radiology, November 1, 2003; 229(2): 347 - 359. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. Eastwood, M. H. Lev, M. Wintermark, C. Fitzek, D. P. Barboriak, D. M. Delong, T.-Y. Lee, T. Azhari, M. Herzau, V. R. Chilukuri, et al. Correlation of Early Dynamic CT Perfusion Imaging with Whole-Brain MR Diffusion and Perfusion Imaging in Acute Hemispheric Stroke AJNR Am. J. Neuroradiol., October 1, 2003; 24(9): 1869 - 1875. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. F. Tomandl, E. Klotz, R. Handschu, B. Stemper, F. Reinhardt, W. J. Huk, K.E. Eberhardt, and S. Fateh-Moghadam Comprehensive Imaging of Ischemic Stroke with Multisection CT RadioGraphics, May 1, 2003; 23(3): 565 - 592. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Warach Stroke Neuroimaging Stroke, February 1, 2003; 34(2): 345 - 347. [Full Text] [PDF]
|
|
|
|
|
|
J. D. Eastwood, M. H. Lev, and J. M. Provenzale Perfusion CT with Iodinated Contrast Material Am. J. Roentgenol., January 1, 2003; 180(1): 3 - 12. [Full Text] [PDF]
|
|
|
|
|
|
D. G. Nabavi, S. P. Kloska, E.-M. Nam, M. Freund, C. G. Gaus, E. Klotz, W. Heindel, and E. B. Ringelstein MOSAIC: Multimodal Stroke Assessment Using Computed Tomography: Novel Diagnostic Approach for the Prediction of Infarction Size and Clinical Outcome Stroke, December 1, 2002; 33(12): 2819 - 2826. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wintermark, M. Reichhart, O. Cuisenaire, P. Maeder, J.-P. Thiran, P. Schnyder, J. Bogousslavsky, and R. Meuli Comparison of Admission Perfusion Computed Tomography and Qualitative Diffusion- and Perfusion-Weighted Magnetic Resonance Imaging in Acute Stroke Patients Stroke, August 1, 2002; 33(8): 2025 - 2031. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
= CBV abnormality, 
= CBF abnormality, and 
= MTT abnormality.