| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Neuroradiology Division (M.E.M., P.W.S., A.G.S., E.F.H., J.H., R.G.G.) and Stroke Service (H.A., W.J.K.), Massachusetts General Hospital, Harvard Medical School, 55 Fruit St, GRB285, Boston, MA 02114. From the 1999 RSNA scientific assembly. Received May 1, 2001; revision requested June 4; revision received October 12; accepted December 4. R.G.G. supported by National Institutes of Health grants RR13213 and NS34626 and Department of Defense DAMD 17-99-2-9001. A.G.S. and W.J.K. supported by NIH grant NS38477. Address correspondence to R.G.G. (e-mail: rggonzalez@partners.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Clinical data and images obtained in 691 consecutive patients with suspected acute stroke were examined. Results of first and second head CT and brain diffusion-weighted (DW) and conventional MR imaging were compared with each other and with the final neurologic discharge diagnosis.
RESULTS: Five hundred seventy-three patients underwent CT at presentation, with 42% sensitivity (95% CI: 37%, 46%) and 91% specificity (95% CI: 82%, 96%). A total of 173 patients underwent a second CT examination, with 77% sensitivity (95% CI: 70%, 84%) and 79% specificity (95% CI: 49%, 95%). Of 498 MR images, 411 were DW, with 94% sensitivity (95% CI: 1%, 96%) and 97% specificity (95% CI: 88%, 100%), and 87 were conventional, with 70% sensitivity (95% CI: 58%, 81%) and 94% specificity (95% CI: 70%, 100%). By using DW MR imaging in the early period (<6 hours after presentation to emergency department), a 97% sensitivity (95% CI: 92%, 100%) and a 100% specificity (95% CI: 69%, 100%) were achieved, compared with 58% (29%–84%) and 100% (16%–100%), respectively, with conventional MR imaging, and 40% (35%–45%) and 92% (84%–97%), respectively, with CT. Negative predictive value was higher with DW MR imaging (73%) than with conventional (42%) MR imaging or CT (24%). In studies conducted within 12 hours, DW MR imaging achieved substantially superior accuracy than did CT. After 12 hours, accuracy was equivalent.
CONCLUSION: In the diagnosis of stroke in the early period (<12 hours after presentation), DW MR imaging is superior to conventional MR imaging and CT.
© RSNA, 2002
Index terms: Brain, CT, 10.12111 • Brain, infarction, 10.78 • Brain, ischemia, 10.781 • Brain, MR, 10.121411, 10.121412, 10.121413, 10.121416, 10.12143, 10.12144
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The purpose of our study was to compare findings obtained with DW and conventional MR imaging and CT in the assessment of acute stroke in a routine clinical setting that involved a large number of patients.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
CT and MR Imaging
A typical head CT examination (GE Medical Systems, Waukesha, Wis) was performed without intravenous contrast material, with 5-mm contiguous transverse sections, 140 kVp, 340 mAs, and 1-second scanning time. Interpretations of scans were made after review of hard-copy images at standard window and level settings.
MR imaging was performed with a 1.5-T system (Signa; GE Medical Systems) with echo-planar capability. A typical MR imaging examination consisted of sagittal T1-weighted imaging (650/16 [repetition time msec/echo time msec]; field of view, 20 x 20 cm; acquisition matrix, 256 x 192 pixels; 5-mm section thickness with a 1-mm gap; and one signal acquired). Fast spin-echo intermediate-weighted MR images were obtained with 2,500/18, field of view of 20 x 20 cm, acquisition matrix of 256 x 256 pixels, 5-mm section thickness with 1-mm gap, and one signal acquired. Transverse fluid-attenuated inversion recovery MR images were obtained with 10,002/141/2,200 (inversion time msec), field of view of 24 cm, acquisition matrix of 256 x 192 pixels, 5-mm section thickness with 1-mm gap, and one signal acquired. Transverse fast spin-echo T2-weighted MR images were obtained with 4,200/102, field of view of 20 x 20 cm, acquisition matrix of 256 x 256 pixels, 5-mm section thickness with 1-mm gap, and one signal acquired. DW MR images were obtained with single-shot echo-planar imaging with 6,000/118, field of view of 40 x 20 cm, matrix of 256 x 128 pixels, 6-mm section thickness with 1-mm gap, and 20 transverse sections.
The effective gradient strength was 14 mT/m, and b values were 1,221 and 4 sec/mm2 with six gradient directions and three signals acquired, with an image acquisition time of 126 seconds. Trace DW MR images and apparent diffusion coefficient (ADC) maps were computed by taking the geometric mean of the six gradient-direction DW MR images. The ADC maps were computed in standard fashion (voxel by voxel with the standard formula of Stejskal and Tanner [24] by using one high-b-value trace DW image and one low-b-value image). DW MR images and ADC maps were available for review. Originally, DW images were available online, and ADC maps were available off-line. This combination constituted approximately 50% of the cohort. The delay was 30 minutes to 24 hours until ADC data were available. In the reminder of the cohort, both DW MR images and ADC maps were available online and were reviewed simultaneously. Software for these images was developed at our hospital.
Data Review
After institutional review board approval, hospital records were reviewed to determine sex, age, triage date and time, and discharge diagnosis. The following discharge diagnoses were considered positive for acute infarction: stroke, hemorrhagic stroke, lacunar stroke, and cerebral infarction with transient symptoms. Any other discharge diagnoses were considered negative for acute infarction.
Radiology reports were reviewed for interpretations and to determine the timing of first head CT, second head CT, first brain MR imaging, and first DW MR imaging. Terms considered positive for acute infarction included infarct, intraparenchymal hemorrhagic stroke, and lacunar stroke. Terms considered negative for acute infarction included possible, rule-out, cannot exclude, no evidence, or old (prior or remote) hemorrhage or hemorrhage not associated with stroke, infarct, or ischemia.
In cases in which the MR imaging report disagreed with the discharge diagnosis, three neuroradiologists (M.E.M., J.H., and P.W.S.) reviewed the images together and reached a consensus interpretation. In cases with agreement between DW MR imaging results and the clinical diagnosis of stroke, review of ADC maps confirmed that hyperintensity on DW MR images represented restricted diffusion rather than T2 shine through. In cases with discordance, we re-reviewed DW MR images and ADC maps. For 11 patients with an initial assessment of a false-negative finding at DW MR imaging, a consensus was reached that infarct was present. It was illustrated in the report, but the finding was interpreted as equivocal in eight instances. In the remaining three patients, the finding was clearly present but not noted in the report. The results in these patients were modified with consensus to true-positive, as it was deemed more accurate.
Temporal Assignments
In this patient cohort, we reviewed a sample of the patient charts for the time of stroke symptom onset and were able to obtain this information in fewer than 25% of patients. Therefore, since a precise time of symptom onset could not be confirmed in the majority of patients, we chose the time at presentation to the emergency department as zero and calculated the interval from this time until diagnostic imaging. Patients with acute stroke compose the entire patient cohort. The intervals were divided into early (
6 hours), with some delay (6–24 hours), and delayed (>24 hours). These intervals were used to estimate sensitivity, specificity, and positive predictive and negative predictive values. The intervals were then further subdivided into 2- and 4-hour time blocks to devise a graph that demonstrated the accuracy of CT and DW MR imaging. The time of imaging was not available for 31 patients who underwent first CT, seven patients who underwent second CT, five patients who underwent MR imaging, and 25 patients who underwent DW MR imaging. These patients were considered in the overall tabulations but were excluded from individual time assignments.
Statistics
Primary neurologic discharge diagnoses, obtained from medical records, were used as the standard against which the diagnostic results were compared. If the results of first CT, second CT, MR imaging, or DW MR imaging agreed with the discharge diagnosis, a designation of true-positive or true-negative was made. If the results conflicted with the standard, a designation of false-positive or false-negative was made. Standard statistical evaluations of sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were performed. The 95% CIs for sensitivity and specificity were calculated according to standard methods.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The findings in this study support the fact that DW MR imaging is superior to MR imaging, first CT, and second CT in the examination of patients with acute stroke within 24 hours of presentation. This superiority is most clearly demonstrated when imaging is performed in less than or equal to 6 hours following presentation. This superiority is maintained even after negative head CT findings. For DW MR imaging, sensitivity (94%), specificity (97%), positive predictive value (99%), and negative predictive value (73%) are similar to those previously reported in studies (8,20,21) with smaller groups of patients. We believe the large patient number and the emergency department environment reported herein strongly confirm the superiority of DW MR imaging in the evaluation of patients with acute stroke.
Lansberg et al (27), in a study of 19 patients with acute middle cerebral artery stroke involving more than 33% of the middle cerebral artery territory, reported an improved sensitivity of DW MR imaging (57%–86%) compared with that of CT (14%–43%). Gonzalez et al (8), in a blinded review of 22 patients with acute strokelike symptoms who underwent imaging within 6 hours, also reported an improved sensitivity of DW MR imaging (100% sensitivity; 95% CI), compared with that of MR imaging (18%) and CT (45%). To our knowledge, these are the only published studies in which CT is compared with DW MR imaging.
DW MR imaging is more accurate than CT early in the course of an ischemic event because it is sensitive to changes in diffusion of water molecules. These molecules become restricted shortly after the onset of ischemia (28–30). As cytotoxic edema develops, the measured ADC of water decreases (31). Abnormal signal intensity on DW MR images may appear within 30 minutes of stroke onset (32); DW MR imaging of ischemia offers high contrast resolution (8). In comparison, detection of acute stroke at CT and T2-weighted MR imaging depends on a substantial increase in tissue water and requires more time to develop.
The findings in this study suggest that three types of false-negative findings exist with DW MR imaging: (a) absent findings that correspond to the clinical scenario (radiographically silent); (b) technically challenging cases; and (c) missed cases, seen in retrospect, which accounted for 80% of false-negative examinations. The most common sites of missed and subtle findings were the brainstem and deep gray nuclei, which are similar to findings reported previously (33). False-positive findings with DW MR imaging are very unusual; they occurred once in this study. Restricted diffusion is observed in other neurologic diseases, such as brain tumor and abscess, but additional imaging findings and clinical presentation permit differentiation from acute stroke (18).
Our study has limitations. Inclusion solely of patients who have been admitted with a clinical suspicion of acute stroke may introduce selection bias and may result in a higher accuracy. The intent of the study was to compare MR imaging with CT and, hopefully, bias would be shared between the groups and thus be of minimal effect. To address these and other possible bias issues, a prospective study is underway. The absence of reliable documentation concerning the time of symptom onset may be considered a limitation of this study. However, this lack of knowledge applied to all of the patients examined, regardless of the imaging group. Thus, it does not weaken our major finding of the comparative ability of CT, MR imaging, and DW MR imaging. Further, as it is common for a patient with stroke at presentation to an emergency department to have an indeterminate history of onset, the use of the time of symptom onset would have resulted in both the exclusion of a large number of patients and the introduction of a substantial bias to the study.
Another potential limitation is that in the majority of DW MR imaging cases, CT had been previously performed, which may have enhanced interpretation of the MR study. However, we observed identical accuracy in the interpretation of DW MR images obtained after negative CT scans and those obtained after positive CT scans. This finding suggests that this potential bias was not a major factor. The retrospective nature of this study is a limitation, but in light of recent work, a minor one.
In a convincing empirical examination of study designs used in the evaluation of diagnostic tests, Lijmer et al (34) demonstrated that retrospective analyses, when compared with prospective data collection, do not result in different results when other sources of bias are minimized. The exclusion of patients with transient ischemic attack may be considered a potential bias to some readers, but our understanding of transient ischemic attack physiology and imaging is that it would be inaccurate to include patients with transient ischemic attack as having either acute stroke or no apparent acute stroke. Thus, to remove this potential bias, we excluded these patients.
The use of the primary neurologic discharge diagnosis as the standard may be considered a limitation, as it is subjective and may depend, in part, on the imaging results. Moreover, this potential incorporation bias may be more heavily weighted toward MR imaging than CT in difficult or conflicting cases. MR imaging was, on average, performed later than first CT and thus may be potentially more likely to depict findings of ischemic injury. In part to address this potential bias, we compared CT and MR imaging results within individual patients during the initial 6 hours after presentation and found that the results paralleled those of the overall cohort (Table 4). A total of 83% (411 of 498) of the patients underwent follow-up MR imaging (data not shown), and none had reversible diffusion abnormalities reported. In our experience and according to the literature, reversibility of diffusion abnormalities in the setting of acute arterial stroke in the absence of thrombolysis is exceedingly rare (35). We conclude that limitations of this study do not substantially mitigate the robustness of our primary finding of the superiority of DW MR imaging in the diagnosis of acute ischemic stroke in patients.
The optimal method for stroke detection should provide a rapid and cost-effective diagnosis that can lead to effective intervention. Such a technique should be readily available, take no more than a few minutes to perform, and be accurate and precise, especially in the acute period (36–41). When combined with prior published data, our results indicate that DW MR imaging fulfills these criteria. Its superior performance during an early period may be of clinical importance, as therapeutic intervention is possible only during a narrow window of opportunity after ischemia, if the onset time of stroke ictus is also known (3,42,43).
One of the primary classic indications for screening head CT in a patient suspected of having a stroke is to exclude intracranial hemorrhage. The evaluation of 21 patients in this cohort with CT and MR imaging is the topic of an ongoing study, as MR imaging may be less sensitive in subtle cases. Of practical importance, DW MR imaging is now widely available. The potential increased cost of DW MR imaging compared with the CT cost might be offset with the elimination of screening head CT examinations and multiple CT examinations when initial scans are negative, thus avoiding unnecessary stroke therapy and improving patient outcome with enhanced diagnostic accuracy.
Our findings support the use of DW MR imaging for examination of patients suspected of having an acute stroke during the first 12 hours after presentation to the emergency department. Thereafter, DW MR imaging is not demonstrably superior to CT for the evaluation of these patients. Accurate emergency diagnosis of stroke continues to evolve. CT and MR imaging of the brain may allow accurate identification of ischemia.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, M.E.M., P.W.S., R.G.G.; study concepts and design, M.E.M., P.W.S., H.A., W.J.K., R.G.G.; literature research, M.E.M., P.W.S., W.J.K., R.G.G.; clinical studies, all authors; data acquisition, M.E.M., J.H., P.W.S., H.A., W.J.K.; data analysis/interpretation, all authors; statistical analysis, M.E.M., E.F.H., R.G.G.; manuscript preparation, definition of intellectual content, editing, revision/review, and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Ay, E. M. Arsava, S. C. Johnston, M. Vangel, L. H. Schwamm, K. L. Furie, W. J. Koroshetz, and A. G. Sorensen Clinical- and Imaging-Based Prediction of Stroke Risk After Transient Ischemic Attack: The CIP Model Stroke, January 1, 2009; 40(1): 181 - 186. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. DeLaPaz and for the Expert Panel on Neurologic Imaging Cerebrovascular Disease AJNR Am. J. Neuroradiol., June 1, 2007; 28(6): 1197 - 1199. [Full Text] [PDF]
|
|
|
|
 |
|
 H. P. Adams Jr, G. del Zoppo, M. J. Alberts, D. L. Bhatt, L. Brass, A. Furlan, R. L. Grubb, R. T. Higashida, E. C. Jauch, C. Kidwell, et al. Guidelines for the Early Management of Adults With Ischemic Stroke: A Guideline From the American Heart Association/American Stroke Association Stroke Council, Clinical Cardiology Council, Cardiovascular Radiology and Intervention Council, and the Atherosclerotic Peripheral Vascular Disease and Quality of Care Outcomes in Research Interdisciplinary Working Groups: The American Academy of Neurology affirms the value of this guideline as an educational tool for neurologists. Circulation, May 22, 2007; 115(20): e478 - e534. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. P. Adams Jr, G. del Zoppo, M. J. Alberts, D. L. Bhatt, L. Brass, A. Furlan, R. L. Grubb, R. T. Higashida, E. C. Jauch, C. Kidwell, et al. Guidelines for the Early Management of Adults With Ischemic Stroke: A Guideline From the American Heart Association/ American Stroke Association Stroke Council, Clinical Cardiology Council, Cardiovascular Radiology and Intervention Council, and the Atherosclerotic Peripheral Vascular Disease and Quality of Care Outcomes in Research Interdisciplinary Working Groups: The American Academy of Neurology affirms the value of this guideline as an educational tool for neurologists Stroke, May 1, 2007; 38(5): 1655 - 1711. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Cook, J. Huston III, M. R. Trenerry, R. D. Brown Jr, K. J. Zehr, and T. M. Sundt III Postcardiac Surgical Cognitive Impairment in the Aged Using Diffusion-Weighted Magnetic Resonance Imaging Ann. Thorac. Surg., April 1, 2007; 83(4): 1389 - 1395. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S P Chung, H S Chung, S Rhu, S W Kim, I S Yoo, J Kim, and C J Song Emergency department experience of primary diffusion weighted magnetic resonance imaging for the patient with lacunar syndrome. Emerg. Med. J., September 1, 2006; 23(9): 675 - 678. [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]
|
|
|
|
 |
|
 P L Tan, D King, C J Durkin, T M Meagher, and D Briley Diffusion weighted magnetic resonance imaging for acute stroke: practical and popular. Postgrad. Med. J., April 1, 2006; 82(966): 289 - 292. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. D. Zimmerman Stroke Wars: Episode IV CT Strikes Back AJNR Am. J. Neuroradiol., September 1, 2004; 25(8): 1304 - 1309. [Full Text] [PDF]
|
|
|
|
 |
|
 M Symms, H R Jager, K Schmierer, and T A Yousry A review of structural magnetic resonance neuroimaging J. Neurol. Neurosurg. Psychiatry, September 1, 2004; 75(9): 1235 - 1244. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. C. Ionita, A. R. Xavier, J. Farkas, P. Pullicino, F. Pico, and P. Amarenco Intracranial arterial dolichoectasia and its relation with atherosclerosis and stroke subtype Neurology, August 10, 2004; 63(3): 596 - 596. [Full Text] [PDF]
|
|
|
|
|
|
S. Warach and C. S. Kidwell The redefinition of TIA: The uses and limitations of DWI in acute ischemic cerebrovascular syndromes Neurology, February 10, 2004; 62(3): 359 - 360. [Full Text] [PDF]
|
|
|
|
|
|
P. A. Hein, C. J. Eskey, J. F. Dunn, and E. B. Hug Diffusion-Weighted Imaging in the Follow-up of Treated High-Grade Gliomas: Tumor Recurrence versus Radiation Injury AJNR Am. J. Neuroradiol., February 1, 2004; 25(2): 201 - 209. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Saur, T. Kucinski, U. Grzyska, B. Eckert, C. Eggers, W. Niesen, V. Schoder, H. Zeumer, C. Weiller, and J. Rother Sensitivity and Interrater Agreement of CT and Diffusion-Weighted MR Imaging in Hyperacute Stroke AJNR Am. J. Neuroradiol., May 1, 2003; 24(5): 878 - 885. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Warach Stroke Neuroimaging Stroke, February 1, 2003; 34(2): 345 - 347. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
) had higher accuracy than first CT (CT1) (
) in the diagnosis of stroke with respect to the time following patient presentation to the emergency department. Error bars indicate the standard error of the mean.
