| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Depts of Clinical Radiology (J.O.K., Y.L., R.L.V., P.L.K.P., P.A.V., S.S., H.J.A.), Neurology (J.N., R.R.), and Clinical Physiology and Nuclear Medicine (E.J.V., J.T.K.), Kuopio University Hospital, Puijonlaaksontie 2, FIN-70210 Kuopio, Finland; Dept of Neuroradiology, Aarhus University Hospitals, Denmark (L.Ø.); Haage Neurological Research Center, Helsinki, Finland (R.R.); Dept of Forensic Psychiatry, Niuvanniemi Hospital, Kuopio (J.T.K.); and Dept of Radiology, Helsinki University Central Hospital (H.J.A.). From the 1999 RSNA scientific assembly. Received Dec 7, 1999; revision requested Jan 21, 2000; revision received Apr 7; accepted Apr 20. Supported in part by the Kuopio University Hospital (EVO funding 307/97 and 21/98; J.O.K., Y.L., J.N., H.J.A.), Radiological Society of Finland (J.O.K., Y.L.), Instrumentarium Science Foundation (J.O.K., Y.L.), Aarne Koskelo Foundation (J.O.K.), Paavo Nurmi Foundation (H.J.A.), Sigrid Juselius Foundation (J.O.K., Y.L., H.J.A.), and Academy of Finland (H.J.A.). Address correspondence to J.O.K. (e-mail: jari.karonen@kuh.fi).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Combined diffusion-weighted (DW) and perfusion-weighted (PW) MR imaging was performed in 49 patients with acute (<24 hours) stroke, on the 1st and 2nd days and 1 week after stroke. Volumes of hypoperfused tissue on maps of relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and mean transit time (MTT) were compared with the volume of infarcted tissue at DW imaging.
RESULTS: The mean infarct volume increased from 41 to 65 cm3 between the 1st and 2nd days (P < .001; n = 49). On the 1st day, all perfusion maps on average showed hypoperfusion lesions larger than the infarct at DW imaging (P < .001; n = 49). MTT maps showed significantly (P < .001) larger hypoperfusion lesions than did rCBF maps, which showed significantly (P < .001) larger hypoperfusion lesions than did rCBV maps. The sizes of the initial perfusion-diffusion mismatches correlated significantly with the extent of infarct growth (0.479 < r < 0.657; P 
.001). The hypoperfusion volume on the initial rCBV maps correlated best with the final infarct size at 1 week (r = 0.891; P < .001).
CONCLUSION: Combined DW and PW imaging is a powerful tool in evaluating the hemodynamics of acute ischemic stroke.
Index terms: Blood vessels, MR, 17.12142, 17.12144 • Brain, diffusion, 13.12144 • Brain, infarction, 13.78 • Brain, MR, 13.12144 • Brain, perfusion, 13.12144 • Magnetic resonance (MR), diffusion study, 13.12144 • Magnetic resonance (MR), perfusion study, 13.12144 • Magnetic resonance (MR), vascular studies, 13.12144
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Clinical trials on treatment of stroke have used computed tomography (CT) to select or exclude patients for thrombolysis. In some patients, CT can be used to detect vasogenic edema in infarcted brain tissue within hours of the onset of symptoms (2), in addition to excluding hemorrhage. However in many patients, positive CT findings do not appear during the time window in which thrombolysis can be useful.
Any magnetic resonance (MR) imaging protocol for acute ischemic stroke must be able to depict early ischemic changes in a sensitive manner. This can be achieved with diffusion-weighted (DW) and perfusion-weighted (PW) MR imaging. DW MR imaging depicts infarcted tissue within 5 minutes after the occlusion of the feeding vessel (3). PW imaging is able to depict hypoperfused brain tissue around the infarcted core (4). The mismatch between lesion sizes at PW and DW imaging (perfusion-diffusion mismatch) in the acute phase has been considered as an estimate of the ischemic penumbra (5,6) and is helpful in selecting patients for different treatment groups (7). The perfusion-diffusion mismatch can be determined by using different perfusion parameters, such as relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and mean transit time (MTT) (8,9). However, it is by no means clear which of these perfusion parameters is most useful during the acute phase of stroke in predicting the infarct growth and final infarct size.
In a series of 23 patients who underwent imaging within 12 hours after the onset of stroke, rCBV maps were shown to predict final infarct size more accurately than did rCBF maps (10). To our knowledge, that is the only published study in which rCBV, rCBF, and MTT in clinical stroke were compared.
The purposes of the present study were as follows:
1. Volumetric measurements: We compared the lesion volumes on maps of rCBV, rCBF, and MTT during the 1st week after the onset of ischemic stroke. Their potential in the prediction of infarct growth and final infarct size was evaluated.
2. Dichotomized PW and DW imaging findings: Simulating a clinical situation with time for only visual review of images without actual volumetric measurements, we also tested whether a rough estimate of a clear perfusion-diffusion mismatch could be used to predict whether the patient is likely to have clinically meaningful infarct growth subsequently.
3. Reproducibility measurements: In addition, we studied reproducibility of manual measurement of lesion size on perfusion maps and on DW images.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The patients underwent the first MR imaging examination within 24 hours after the onset of symptoms or after the time they were last seen healthy. Eight patients were excluded from the analysis because PW imaging was not performed successfully at the initial MR imaging examination. The final study population thus consisted of 49 patients (28 men, 21 women; mean age, 70 years; age range, 48–87 years). The second MR imaging examination was performed on the 2nd day, and the third MR imaging examination was performed 1 week after the onset of symptoms. Written informed consent was obtained from the patient or a relative of the patient. The study design was approved by the ethics committee of our institution.
MR Imaging Protocol
MR imaging was performed with a 1.5-T whole-body imager (Magnetom Vision; Siemens Medical Systems, Erlangen, Germany) capable of echo-planar imaging by using a head coil and standard restraints. Each MR imaging examination included DW, PW, and conventional imaging. The total imaging time was approximately 20 minutes. Conventional MR images were obtained for clinical use and are not reported here.
After localizer images were obtained, DW imaging was performed with an echo-planar spin-echo sequence. Nineteen transverse sections (thickness, 5 mm; gap, 1.5 mm) tilted along the orbitomeatal line and covering the whole supratentorial brain were imaged. Other imaging parameters were as follows: 4,000–6,000/103 (repetition time msec/echo time msec); field of view, 260 mm; matrix size, 96 x 128 interpolated to 256 x 256; and acquisition time, 20 seconds. Three DW images with orthogonally applied diffusion gradients (b value, 1,000 sec/mm2) and a T2-weighted reference image (b value, 0 sec/mm2) per section were obtained.
PW imaging was performed with a spin-echo instead of a gradient-echo sequence to achieve sensitivity to perfusion at the capillary level. In the echo-planar spin-echo sequence, the repetition time was 1,500 msec, and the echo time was 78 msec. After the review of DW images, seven transverse sections (thickness, 5 mm; gap, 1.5 mm; field of view, 260 mm; matrix size, 116 x 256) were obtained at the positions that contained the largest diffusion defects at DW imaging. The set of seven sections was imaged 40 times every 1.5 seconds, which resulted in an imaging time of 1 minute. Before the injection of contrast medium, four sets of baseline images were collected. More baseline images resulted from the time delay between the intravenous injection of the contrast medium and its arrival at the cerebral arterial circulation.
A bolus of gadopentetate dimeglumine (0.2 mmol per kilogram of body weight; Magnevist, Schering, Berlin, Germany) was injected into an antecubital vein by using an 18-gauge cannula. The injection was performed at a speed of 5 mL/sec by using an MR-compatible power injector (Spectris; Medrad, Pittsburgh, Pa). The bolus of contrast medium was followed by a 15-mL bolus of saline solution at the same injection rate. During imaging, the intravenous cannula was kept open by flushing it with 0.25 mL of saline solution per minute.
Postprocessing of the Imaging Data
Effects of diffusion anisotropy were minimized by generating trace images (trace of the diffusion tensor) as the mean of the three DW images (11). Maps of the apparent diffusion coefficient were also generated to control the effects of T2 shine-through in the interpretation of DW imaging findings. The apparent diffusion coefficient was calculated on a pixel-by-pixel basis as the negative slope of the line fitting the two points for b versus ln(SI), where SI is the signal intensity.
Raw PW images were transferred to a Unix workstation (SPARC Ultra; Sun Microsystems, Palo Alto, Calif) for postprocessing. Maps of rCBV, rCBF, and MTT were generated. Tissue and arterial levels of contrast medium were determined by assuming a linear relationship between transverse relaxivity and intravascular concentration of gadopentetate dimeglumine (12–14). The rCBV was determined by using numeric integration of the first-pass concentration-time curve. The shape of the arterial input function was determined from pixels located at a large branch of the middle cerebral artery, which showed large losses of signal intensity during the bolus passage. In each imaging pixel, the tissue concentration-time curve was deconvolved with the arterial input function to determine the tissue impulse response. The rCBF was subsequently determined as the height of the deconvolved tissue impulse response (8,9). The tissue MTT was determined by using the central volume theorem (15) as the ratio of rCBV to rCBF.
Volumetric Measurements
Volumes of decreased diffusion were measured by drawing regions of interest around the lesions on trace images and by multiplying the lesion areas by the section and gap thicknesses. However, the lesion area in the lowermost section was multiplied by only the section thickness. Two interpreters (J.O.K., Y.L.) drew the regions of interest and followed the same rules in including the tissue in the regions of interest. They were blinded to the patients’ clinical data.
The volume of decreased perfusion was determined by one interpreter (J.O.K.) by drawing a region of interest around the pathologic area section by section. A software package (CHESHIRE; Hayden Image Processing Group, Boulder, Colo) was used for image analysis. After the regions of interest were multiplied by the section and gap thicknesses, they were summed to give the total volume of hypoperfused tissue in the seven imaged sections. Again, the lesion area in the lowermost section containing the lesion was multiplied by only the section thickness. Because acute hypoperfusion could not be distinguished from probable old perfusion defects, the interpreter was not blinded to the DW imaging findings, but he was blinded to the patients’ clinical data.
In summary, the following volumes were measured: (a) the total volume of tissue with decreased diffusion at DW imaging, (b) the volume of hypoperfused tissue on the three types of perfusion maps (rCBV, rCBF, and MTT) in the seven sections included at PW imaging, (c) the volume of tissue with decreased diffusion at DW imaging limited to the seven sections comparable with PW imaging performed on the same day, and (d) the volumes of decreased diffusion at DW imaging on the 2nd day and at 1 week from the same section positions in which PW imaging had been performed on the 1st day to study the growth of the DW imaging lesion volumes.
These lesion volumes were used in calculating perfusion-diffusion mismatches and extents of infarct growth. The perfusion-diffusion mismatch was calculated by subtracting the infarcted tissue volume, defined as high-intensity regions at DW imaging, separately from the hypoperfusion volume on rCBV, rCBF, and MTT maps. This resulted in three perfusion-diffusion mismatches: mismatch between hypoperfusion volume on rCBV maps and infarct volume at DW imaging (rCBV–DW imaging mismatch), mismatch between hypoperfusion volume on rCBF maps and infarct volume at DW imaging (rCBF–DW imaging mismatch), and mismatch between hypoperfusion volume on MTT maps and infarct volume at DW imaging (MTT–DW imaging mismatch). The infarct growth was defined as the increase in infarcted tissue volumes measured on the seven DW imaging sections matching the positions of the sections at the initial PW imaging examination. The infarct growth was calculated for the change between the 1st and 2nd days and for the change between the 1st day and 1 week.
Analysis of Dichotomized PW and DW Imaging Findings
The patients were placed into one of four groups: (a) the patients with a large initial perfusion-diffusion mismatch and large infarct growth, (b) the patients with a large initial perfusion-diffusion mismatch but small or no infarct growth, (c) the patients with a small initial perfusion-diffusion mismatch or no perfusion-diffusion mismatch and small or no infarct growth, and (d) the patients with a small initial perfusion-diffusion mismatch or no perfusion-diffusion mismatch but large infarct growth. A lesion at PW imaging at least 50% larger than the lesion at DW imaging was defined as a large perfusion-diffusion mismatch. Infarct growth of at least 50% was defined as large infarct growth.
The cutoff point of 50% for perfusion-diffusion mismatch was arbitrarily selected to represent a clear mismatch that is visible on the images without volumetric measurements. The cutoff point of 50% for infarct growth was arbitrarily selected to represent a clinically meaningful infarct growth. The infarct growth for this analysis was measured between the 1st day and 1 week in all but those five patients who died during the week (n = 3) or refused to undergo the third imaging examination (n = 2). For these patients, the infarct growth was measured between the 1st and the 2nd days.
Reproducibility Measurements
A subgroup of 15 patients with various infarct sizes was randomly selected for the analysis of interobserver and intraobserver reproducibility. DW and PW imaging findings on the 1st day were included in the analyses. Determination of variability was performed for the lesion area (section by section) and by lesion volume (patient by patient).
The lesion measurements at DW imaging included 285 sections in 15 patients. To determine intraobserver reproducibility, one observer (Y.L.) drew regions of interest on trace images twice, the second time after an interval of at least 3 months. To determine interobserver reproducibility, the first set of measurements of this observer was compared with corresponding measurements obtained by a second observer (J.O.K.).
The reproducibility for measuring lesion size at PW imaging was studied by using the same approach for 105 sections in 15 patients. Lesion sizes on rCBV, rCBF, and MTT maps were analyzed separately. One observer (J.O.K.) measured the lesions twice, with an interval of at least 3 months between measurements, to determine the intraobserver reproducibility. To determine interobserver reproducibility, the first set of measurements was compared with lesion sizes measured by the second observer (Y.L.).
Statistical Analysis
Since the distribution of the lesion volumes clearly deviated from the normal distribution, nonparametric tests were used. A P value lower than .05 was considered statistically significant. Statistical analysis was performed with a statistical software package (SPSS for Windows 9.0; SPSS, Chicago, Ill).
The significance of the changes in mean infarct volumes and the differences between the infarct volume and the hypoperfusion volumes were tested by using the Wilcoxon matched pairs signed rank test. The same test was used to test the significance of the differences between the hypoperfusion volumes measured from the rCBV, rCBF, and MTT maps on each day. In analyzing the association of the dichotomized perfusion-diffusion mismatch (<50% or 
50%) with the dichotomized extent of infarct growth (<50% or 50%), the 
2 test was used to study the association between the size of perfusion-diffusion mismatch and the extent of infarct growth.
The Spearman correlation coefficient was used to calculate (a) the correlation between the total infarct volume and the infarct volume in the seven sections matching with PW imaging; (b) the correlation between the initial perfusion-diffusion mismatch and the infarct growth; (c) the correlation between the initial hypoperfusion volume on PW imaging maps and the infarct size at 1 week; and (d) in reproducibility analyses, the correlations between the two measurements of lesion sizes obtained at DW and PW imaging.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
DW imaging was successfully completed at every MR imaging session. As stated earlier, eight patients were excluded from the study because PW imaging could not be performed successfully on the 1st day owing to patient motion (n = 5), imager malfunction (n = 1), or an operator-dependent error (n = 2). In the 49 included patients, the hypoperfusion volume at PW imaging on the 2nd day was not available in eight patients because of patient motion (n = 4), serious imaging artifact (n = 3), or an operator-dependent error (n = 1). The hypoperfusion volume was not available at 1 week in seven patients because of patient motion (n = 1) or imager malfunction (n = 1) or because the patient had died (n = 3) or refused to undergo imaging (n = 2).
Lesion Volumes
All patients had acute ischemic lesions at DW imaging on the 1st day. The mean total volume of infarcted tissue measured from the trace DW images covering the whole supratentorial brain increased significantly from 41 cm3 ± 65 on the 1st day to 65 cm3 ± 93 on the 2nd day (n = 49; P < .001). In 46 (94%) of 49 patients, the infarct grew between the 1st and 2nd days. The mean relative growth was 142% (range, -21% to 1,417%). When measured at 1 week, the mean total infarct volume had increased slightly to 66 cm3 ± 75 (n = 44; P = .002). The mean relative growth from the 2nd day to 1 week was 55% (range, -66% to 1,167%).
The lesion volumes at DW imaging in the seven sections matching with the PW imaging sections correlated with the total lesion volumes at DW imaging on the 1st day (r = 0.995; P < .001; n = 49), on the 2nd day (r = 0.990; P < .001; n = 41), and 1 week after stroke (r = 0.951; P < .001; n = 42). On the 1st day, the total infarct volume was larger than that measured from the seven sections in 31 (63%) of 49 patients; the mean infarct volume outside the seven sections was 14.3 cm3 ± 25.3. In the rest of the patients (n = 18), the infarcts were limited to the seven sections.
On the 1st day, rCBV maps showed hypoperfusion in 40 patients; rCBF maps, in 43 patients; and MTT maps, in 43 patients. In 39 patients on the 1st day, hypoperfusion was detected on every type of perfusion map. The mean total infarct volume at DW imaging on the 1st day in these 39 patients was 51 cm3 ± 70 (range, 1–385 cm3). In 10 patients who did not have hypoperfusion on every type of perfusion map on the 1st day, the mean total infarct volume was significantly smaller: 2 cm3 ± 2 (range, 0.1–6 cm3; P < .001).
Of the 39 patients whose initial rCBV, rCBF, and MTT maps showed hypoperfusion, 34 underwent MR imaging at 1 week. In 14 (41%) of 34 patients, the final infarct volume in the seven sections was smaller than the initial rCBV lesion volume (Fig 1). In 10 (29%) of 34 patients, the final infarct volume in the seven sections was greater than the initial rCBV lesion volume but smaller than the initial rCBF lesion volume. The final infarct volume in the seven sections was greater than the initial rCBF lesion volume in 10 (29%) of 34 patients, and in six of them the final infarct volume in the seven sections was even greater than the initial MTT lesion volume (Fig 2). In the six patients whose infarct grew larger than the initial MTT lesion volume, the volume difference between the final infarct and the MTT lesion volume was 0.5–34.5 cm3 (9.8 cm3 ± 11.7).
|
|
|
At 1 week, the mean hypoperfusion volumes had decreased on all types of perfusion maps, and on rCBV and rCBF maps they were smaller than the mean infarct volume at DW imaging. This is probably due to reperfusion of some of the infarcted tissue, and in 20 patients findings indicative of hyperperfusion were detected on rCBF maps. The differences in hypoperfusion volumes were significant when rCBV and rCBF maps (P < .001) and rCBV and MTT maps (P = .014) were compared but not significant between rCBF and MTT maps (P = .124).
The rCBV–DW imaging, rCBF–DW imaging, and MTT–DW imaging perfusion-diffusion mismatch volumes on the 1st day correlated significantly with the infarct growth between the 1st and 2nd days and between the 1st day and 1 week (Table 2). Patients with a substantial perfusion-diffusion mismatch on the 1st day were more prone to have substantial infarct growth (Fig 3) than were patients in whom the perfusion-diffusion mismatch was smaller or not detectable (Fig 4).
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In a number of studies, ischemic penumbra has been estimated by determining perfusion-diffusion mismatch with combined DW and PW imaging (4–6, 10,25) or DW imaging and single photon emission CT (26). According to the results of these studies, patients with a clear perfusion-diffusion mismatch are more likely to have infarct growth than are patients without perfusion-diffusion mismatch. Thus far, to our knowledge, there are only a few published studies in which the effect of thrombolysis in humans with clinical stroke is evaluated by means of DW and PW imaging. In a study by Marks et al (27), five of six patients who underwent thrombolysis had a hypoperfused lesion on time-to-peak maps that was smaller than the DW imaging lesion after thrombolysis, whereas only one of five patients not receiving thrombolysis had a time-to-peak lesion smaller than the DW imaging lesion. This finding suggests that intravenous thrombolysis is associated with early reperfusion. However, large multicenter trials are needed to allow conclusions based on a sufficient number of patients.
We detected infarct growth between the 1st and 2nd days at DW imaging in a majority (94%) of patients. This finding supports the view that there is salvageable tissue left in the area of ischemic tissue after the first 3–6 hours. Especially in larger infarcts, space-occupying edema may cause expansion of the infarcted tissue, which inevitably results in some overestimation of infarct growth. Edema may also cause further hypoperfusion by compressing the surrounding vessels. However, the volumes of hypoperfused tissue on the 2nd day, when edema had developed, were not much greater than those on the 1st day. With the exception of the smallest infarcts not visible at PW imaging because of poorer spatial resolution, the initial hypoperfusion lesion tends to be larger than the initial lesion at DW imaging, which suggests the potential presence of penumbral tissue. This perfusion-diffusion mismatch was largest when MTT was compared with DW imaging and smallest when rCBV was compared with DW imaging. The sizes of rCBV–DW imaging, rCBF–DW imaging, and MTT–DW imaging perfusion-diffusion mismatches correlated significantly with infarct growth.
On average, the initial rCBV lesion was closest to the final infarct size, but this was not always true on a patient-by-patient basis. In 20 patients, the infarct grew larger than the initial rCBV lesion, to the area of the rCBF and MTT lesions and rarely even to the tissue that had appeared normal on all initial perfusion maps. However, in patients with infarct growth greater than the initial hypoperfusion lesions, secondary ischemic events could not be excluded. These can be caused by clot progression; clot fragmentation with migration to distal branches; and new emboli from the carotid arteries, thoracic aorta, or heart. Our method does not allow one to distinguish whether the tissue proceeding to infarct is due to the secondary thromboembolic events or primary failure of the collateral circulation. On the other hand, spontaneous thrombolysis in some of the patients cannot be ruled out, which may cause some bias in this study.
Although all perfusion-diffusion mismatches seem to serve as estimates of the penumbra, all tissue appearing with decreased perfusion is not necessarily at risk of infarction. MTT correlates well with local cerebral perfusion pressure (28) and seems to be a sensitive measure of disturbed cerebral perfusion. Calculation of MTT by means of the central volume theorem requires determination of the arterial input function for rCBF, and this procedure increases the postprocessing time. Neumann-Haefelin et al (29) used time-to-peak maps and DW imaging for detecting perfusion-diffusion mismatch. They showed that a delay of at least 6 seconds in time to peak could be used to predict infarction, whereas tissue with a shorter delay survived, even though it might have contributed to the symptoms. In another study (7) with 41 patients, time-to-peak maps in combination with DW imaging were useful in triage of patients with hyperacute stroke. When favoring MTT and time to peak over rCBV and rCBF because they allow visualization of perfusion-diffusion mismatch in a greater number of patients, one must keep in mind that these maps can lead to serious overestimation of the final infarct size in some patients.
DW imaging alone cannot be considered sufficient in acute stroke, because incomplete ischemia with adequate collateral circulation may cause symptoms, but that tissue does not necessarily undergo irreversible damage. On the other hand, PW imaging alone is not satisfactory because the smallest infarcts, and possible hypoperfused zones around them, are beyond the spatial resolution of PW imaging. These facts enhance the importance of combining DW and PW imaging in acute stroke.
In the present study, the mean volume of hypoperfused tissue measured from rCBV maps was closer to, although smaller than, the final infarct size than that measured from rCBF and MTT maps and also showed the best correlation (Table 3). When the lesion volumes were compared separately in 34 patients, the final infarct size was smaller than the initial rCBV hypoperfusion volume in 14 patients and greater than the initial rCBV lesion and smaller than the initial rCBF lesion in 10. In 10 patients, the final infarct size was larger than the initial rCBF lesion. Despite the strong correlations and mean volumes presented in Table 3, it is difficult on a patient-by-patient basis to predict the final size of the infarct on the basis of the perfusion-diffusion mismatches. In a study by Sorensen et al (10), the final infarct size was almost always underestimated by using the initial rCBV maps and almost always overestimated by using rCBF and MTT maps. The reason for this is difficult to address, but one explanation could be the different time windows used by Sorensen et al (12 hours) and us (24 hours). We speculate that the hemodynamic situation in the ischemic or hypoperfused brain can change during the 1st day and the time delay from the ictus to imaging may have a strong influence on the PW imaging findings.
Our method was based on visual interpretation of images and manual drawing of regions of interest, but we found strong correlations between the lesion size measurements. We agree with Sorensen et al (10) about the difficulty of using automated segmentation tools in defining lesions on perfusion maps, one reason for this being the difficulties in distinguishing between gray matter and white matter. Our manual method of defining the region of interest is comparable with their semiautomated method.
It is difficult to name one perfusion map type that could, even when combined with DW imaging, be used to predict infarct growth with satisfactory accuracy. Although the initial lesion volume on rCBV maps correlates best with the final infarct size and on average is closest to it, the initial rCBV map in many patients leads to underestimation of the final infarct size. Estimating the final infarct size with the initial rCBF finding seems to be more accurate. However, some infarcts grow to the area where the only initially detected imaging abnormality is prolonged MTT.
A visually evident large (50%) perfusion-diffusion mismatch is associated with, in most patients, clinically relevant (50%), infarct growth. The overall accuracy of a mismatch of 50% or more between DW imaging and rCBF to predict infarct growth of 50% or more was highest (71.4%). Future software programs for image analysis are likely to provide accurate volumetric results robustly and almost in real time. Until they are in everyday use, visual analysis of perfusion-diffusion mismatch may be helpful in situations in which urgent treatment decisions are needed.
Intraobserver reproducibility for rCBF lesion size measurements was poorer than it was for rCBV and MTT lesion size measurements, but interobserver reproducibility for rCBF lesion size measurements was better than it was for rCBV lesion size measurements. Thus far, inclusion of rCBV, rCBF, and MTT perfusion maps appears to be necessary in the evaluation of acute ischemic stroke, because each perfusion parameter provides information that cannot be extracted visually from the others.
In conclusion, both DW and PW imaging are important in acute stroke imaging. DW imaging can depict small infarcts beyond the spatial resolution of PW imaging, and PW imaging can depict hypoperfused, still viable tissue that is not visible at DW imaging. Our results demonstrate that in almost all patients the infarct grows between the 1st and 2nd days. The sizes of the initial perfusion-diffusion mismatches are significantly associated with the extent of infarct growth. However, it is difficult to predict exactly on a patient-by-patient basis the infarct development and the final infarct size on the basis of the different initial perfusion maps and DW imaging, although, on average, the final infarct size is closest to that of the initial rCBV lesion. Combined DW and PW imaging is a powerful tool in evaluating the hemodynamics of acute ischemic stroke.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, J.O.K., H.J.A.; study concepts, J.O.K., H.J.A., R.L.V., P.L.K.P., E.J.V., J.T.K., J.N., R.R., S.S.; study design, J.O.K., H.J.A., R.L.V., E.J.V., J.T.K., P.L.K.P.; definition of intellectual content, H.J.A.; literature research, J.O.K., R.L.V., P.A.V., Y.L.; clinical studies, J.O.K., R.L.V., P.L.K.P., P.A.V.; data acquisition, J.O.K., J.N., R.R.; data analysis, J.O.K., Y.L., L.Ø., P.A.V.; statistical analysis, J.O.K., R.L.V., Y.L., H.J.A.; manuscript preparation, J.O.K., R.L.V., Y.L., L.Ø., H.J.A.; manuscript editing, J.O.K., H.J.A.; manuscript review, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P.G. Kranz and J.D. Eastwood Does Diffusion-Weighted Imaging Represent the Ischemic Core? An Evidence-Based Systematic Review AJNR Am. J. Neuroradiol., June 1, 2009; 30(6): 1206 - 1212. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B.D. Murphy, A.J. Fox, D.H. Lee, D.J. Sahlas, S.E. Black, M.J. Hogan, S.B. Coutts, A.M. Demchuk, M. Goyal, R.I. Aviv, et al. Identification of Penumbra and Infarct in Acute Ischemic Stroke Using Computed Tomography Perfusion-Derived Blood Flow and Blood Volume Measurements Stroke, July 1, 2006; 37(7): 1771 - 1777. [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]
|
|
|
|
|
|
J. Perkio, L. Soinne, L. Ostergaard, J. Helenius, A. Kangasmaki, S. Martinkauppi, O. Salonen, S. Savolainen, M. Kaste, T. Tatlisumak, et al. Abnormal Intravoxel Cerebral Blood Flow Heterogeneity in Human Ischemic Stroke Determined by Dynamic Susceptibility Contrast Magnetic Resonance Imaging Stroke, January 1, 2005; 36(1): 44 - 49. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Liu, J. O. Karonen, R. L. Vanninen, J. Nuutinen, A. Koskela, S. Soimakallio, and H. J. Aronen Acute Ischemic Stroke: Predictive Value of 2D Phase-Contrast MR Angiography--Serial Study with Combined Diffusion and Perfusion MR Imaging Radiology, May 1, 2004; 231(2): 517 - 527. [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]
|
|
|
|
|
|
C. A. Doege, C. M. Kerskens, B. I. Romero, P. Brunecker, J. Junge-Hulsing, W. von Pannwitz, B. Muller, and A. Villringer Assessment of Diffusion and Perfusion Deficits in Patients with Small Subcortical Ischemia AJNR Am. J. Neuroradiol., August 1, 2003; 24(7): 1355 - 1363. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Kudo, S. Terae, C. Katoh, M. Oka, T. Shiga, N. Tamaki, and K. Miyasaka Quantitative Cerebral Blood Flow Measurement with Dynamic Perfusion CT Using the Vascular-Pixel Elimination Method: Comparison with H215O Positron Emission Tomography AJNR Am. J. Neuroradiol., March 1, 2003; 24(3): 419 - 426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Nighoghossian, M. Hermier, P. Adeleine, L. Derex, J.F. Dugor, F. Philippeau, H. Ylmaz, J. Honnorat, P. Dardel, Y. Berthezene, et al. Baseline Magnetic Resonance Imaging Parameters and Stroke Outcome in Patients Treated by Intravenous Tissue Plasminogen Activator Stroke, February 1, 2003; 34(2): 458 - 463. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. Eastwood, M. H. Lev, T. Azhari, T.-Y. Lee, D. P. Barboriak, D. M. Delong, C. Fitzek, M. Herzau, M. Wintermark, R. Meuli, et al. CT Perfusion Scanning with Deconvolution Analysis: Pilot Study in Patients with Acute Middle Cerebral Artery Stroke Radiology, January 1, 2002; 222(1): 227 - 236. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. B. Grandin, T. P. Duprez, A. M. Smith, C. Oppenheim, A. Peeters, A. R. Robert, and G. Cosnard Which MR-derived Perfusion Parameters are the Best Predictors of Infarct Growth in Hyperacute Stroke? Comparative Study between Relative and Quantitative Measurements Radiology, May 1, 2002; 223(2): 361 - 370. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
