Background
We evaluated whether X-map, a novel imaging technique, can visualize ischemic lesions within 20 hours after the onset in patients with acute ischemic stroke, using noncontrast dual-energy computed tomography (DECT).
Materials and Methods
Six patients with acute ischemic stroke were included in this study. Noncontrast head DECT scans were acquired with 2 X-ray tubes operated at 80 kV and Sn150 kV between 32 minutes and 20 hours after the onset. Using these DECT scans, the X-map was reconstructed based on 3-material decomposition and compared with a simulated standard (120 kV) computed tomography (CT) and diffusion-weighted imaging (DWI).
Results
The X-map showed more sensitivity to identify the lesions as an area of lower attenuation value than a simulated standard CT in all 6 patients. The lesions on the X-map correlated well with those on DWI. In 3 of 6 patients, the X-map detected a transient decrease in the attenuation value in the peri-infarct area within 1 day after the onset.
Conclusions
The X-map is a powerful tool to supplement a simulated standard CT and characterize acute ischemic lesions. However, the X-map cannot replace a simulated standard CT to diagnose acute cerebral infarction.
Key Words
Introduction
There is increasing evidence that thrombolysis and/or mechanical thrombectomy can significantly improve the outcomes in patients with ischemic stroke, when they are treated within an appropriate therapeutic window. Computed tomography (CT) and magnetic resonance imaging (MRI) are useful modalities in determining therapeutic strategy because these techniques can identify tissue that has been irreversibly damaged by cerebral ischemia. A noncontrast CT scan is most often used for this purpose because of lower cost, greater availability, and faster imaging. An early ischemic sign on CT scan is known to be useful to identify irreversibly damaged tissue. CT perfusion may be more helpful to define the ischemic core and penumbra, but contrast materials are always required. Diffusion-weighted imaging (DWI) and perfusion-weighted imaging permit a more sensitive estimation ofthe infarct core and the extent of penumbral tissue. However, MRI is not available in many institutions and cannot always be used in a timely fashion. Therefore, it is important to develop an imaging technique that is noninvasive and identifies promptly the irreversibly damaged tissue in the super-acute (<4.5 hours) stage of ischemic stroke, thus avoiding hemorrhagic transformation after aggressive treatments.
1
Several dual-energy computed tomography (DECT) applications are currently in clinical use.
2
, 3
, 4
, 5
, 6
, 7
, 8
, 9
The major advantage of DECT is its material differentiation ability, which is based on high- and low-peak voltage acquisition.2
, 3
Three-material decomposition (3MD) is used to create a virtual unenhanced scan by subtracting iodine from contrast agent-enhanced DECT.4
, 5
, 6
, 7
, 8
The 3MD is also used to create a virtual noncalcium image from an unenhanced image, which makes bone marrow accessible for CT diagnosis.9
Using these techniques, the authors have recently developed X-map, a novel imaging technique by noncontrast DECT based on 3MD that shows water content, which clearly reflects cerebral edema. Here, we report the preliminary results of applying the X-map imaging algorithm to identify acute ischemic stroke within 24 hours after the onset.
Materials and Methods
Institutional review board approval was obtained with waived informed consent for retrospective analysis.
There were 6 patients enrolled into this study. These patients were admitted to our hospital between April 2015 and November 2015 because of acute ischemic stroke. There were 4 males and 2 females. Their mean age was 70.1 years, ranging from 58 to 76 years. In all 6 patients, initial nonenhanced head DECT was performed between 30 minutes and 20 hours after the onset. In 2 of 6 patients, DECT was repeated within 4.5 hours after the onset. A total of 8 DECT scans were acquired (Table 1).
Table 1Summary of clinical and radiological data in 6 patients included in this study
| Case | Age/Gender | Onset to CT/MRI time | Ischemic lesion on CT | Ischemic lesion on MRI (DWI) | Decreased attenuation value in peri-infarct area on X-map | |
|---|---|---|---|---|---|---|
| Standard CT | X-map | |||||
| 1 | 68/Male | 6 h | Right P | Right P | Yes | |
| 7 h | Right P | |||||
| 1 d | Right P | Right P | Yes | |||
| 28 d | Right P | Right P | No | |||
| 2 | 71/Female | 30 min | None | Left insula and F, T, P | Yes | |
| 70 min | Left insula and F, T, P | |||||
| 4 h | Left insula and T | Left insula and T | No | |||
| 1 d | Left insula and T | Left insula and T | No | |||
| 3 | 58/Male | 50 min | None | Left insula and F, T | No | |
| 1.5 h | Left insula and F, T | |||||
| 3.5 h | Left insula and F, T | Left insula and F, T | No | |||
| 4 | 73/Female | 3 h | Left F | Left F | No | |
| 3.5 h | Left F | |||||
| 5 | 76/Male | 8 h | Right F | Right F | No | |
| 8.5 h | Right F | |||||
| 6 | 75/Male | 20 h | Left centrum semiovale | Left centrum semiovale and F, P | Yes | |
| 20.5 h | Left centrum semiovale | |||||
| 1 d | Left centrum semiovale | Left centrum semiovale | No | |||
Abbreviations: CT, computed tomography; DWI, diffusion-weighted imaging; F, frontal lobe; ICA, internal carotid artery; MCA, middle cerebral artery; MRI, magnetic resonance imaging; P, parietal lobe; T, temporal lobe.
All CT examinations were performed using a dual-source DECT scanner (SOMATOM Force, Siemens Healthcare, Forchheim, Germany). Noncontrast head CT imaging was performed in a dual-energy (DE) acquisition with 2 X-ray tubes operated at 80 kV and Sn150 kV, where “Sn” denotes the use of an additional tin filter that increases the mean photon energy of the respective spectrum. Scan parameters were as follows: collimation width, 192 × .6 mm; rotation time, 1.0 s; pitch, .7. The effective mAs values were set to 800 mAs at 80 kV and 533 mAs at Sn150 kV. The mean value of the volume CT dose index (CTDIvol) was 80.08 mGy. Original DE datasets (80 kV and Sn150 kV) were reconstructed with an increment of 1 mm and a slice thickness of 1 mm and 6 mm using an iterative reconstruction algorithm ADMIRE (Siemens Healthcare, Forchheim, Germany) strength levels 2 and 5. Simulated standard CT was reconstructed by the linear combination of 80 kV image and Sn150 kV image with a weighted factor of .5 to simulate an equivalent of 120 kV image.
The difference in the Hounsfield number between white and gray matter is about 5.5 Hounsfield units (HU) at standard CT energy. This difference arises because gray matter contains 8% more oxygen and 8% less carbon, as a result of its higher water and lower lipid content, and this increases the photoelectric absorption.
10
The X-map is a virtual gray matter and water content map subtracting lipid content from white matter, based on the assumption that each voxel within the brain parenchyma consists mainly of those 3 materials and the fraction of those materials contributes to the total X-ray attenuation of the voxel (Fig 1). The DE datasets were transferred to a postprocessing workstation (syngo.via VA30, Siemens Healthcare, Forchheim, Germany) to generate the X-Map using the DE bone-marrow application with modified parameters for 3MD into gray matter, 42/33 HU (80 kV/Sn150 kV); white matter, 34/29 HU; baseline (pure water–gray matter connecting line); and characteristic lipid slope, 2.0 (gray–white matter ratio, 2:1). The color lookup table for the color overlay was used in all patients for evaluating the X-map images. Lower attenuation values, which indicate large water content, are displayed in blue, and higher attenuation values are displayed in red.
Figure 1The principle of X-map. Brain parenchyma mainly consists of 3 materials: gray matter (42/33 HU low/high kV), white matter (34/29 HU), and water. The attenuation difference between gray matter and white matter assumes a difference in the lipid content. If all voxels are projected to the baseline (pure water–gray matter connecting line) (blue line) by using the characteristic lipid slope, 2.0 (gray–white matter ratio, 2:1) (red line) for removal of the lipid content, the difference in the attenuation level among each voxel on the baseline will reflect mainly the water content. Abbreviation: HU, Hounsfield unit. (Color version of figure is available online.)
In this study, DWI, T2*-weighted imaging, fluid-attenuated inversion recovery, T2-weighted imaging, T1-weighted imaging, and 3D time-of-flight magnetic resonance angiography were performed in all 6 patients between 1.5 and 20.5 hours after the onset. MRI was performed using a clinical 1.5-T MRI unit (Magnetom Avanto, Siemens AG, Erlangen, Germany) with a standard 12-channel head coil. The intervals between nonenhanced head DECT and DWI ranged from 20 to 60 minutes. Two investigators (KN and SK) independently compared the findings on DWI with those on simulated standard CT, low kV (80 kV), high kV (Sn150 kV), and X-map. We also performed a preliminary study to evaluate the uniformity of the normal hemisphere on the X-map in another 15 control subjects (6 males and 9 females; mean age was 61.2 years, age range 23-78 years). The control subjects had undergone DECT for evaluation for headache, trauma, and vertigo; the CT findings were normal.
Results
On the initial X-map, acute ischemic lesions were identified as the area of lower attenuation value in all 6 patients. Radiological data are summarized in Table 1.
When noncontrast head DECT was performed within 1 hour after the onset (cases 2 and 3), no ischemic lesions were detected on simulated standard CT, low kV, and high kV CT, but the X-map clearly identified them as the area with a decreased attenuation value, which correlated well with those on DWI (Fig 2). On the other hand, when noncontrast head DECT was performed between 3 and 20 hours after the onset (cases 1, 4, 5, and 6), the ischemic lesions on the X-map correlated well with those on simulated standard CT, high kV CT, and DWI (Fig 3). The findings suggest that the X-map is valuable for identifying ischemic lesions with sensitivity similar to DWI in the acute stage of ischemic stroke.
Figure 2Serial radiological findings in case 2, a patient with acute ischemic stroke. (A) The first simulated standard CT obtained 32 minutes after the onset shows no definite early sign. (B) The first X-map obtained 32 minutes after the onset shows an abnormality in the left middle cerebral artery territory (arrows). (C) Diffusion-weighted MRI obtained 1 hour after the onset also shows an abnormality in the left middle cerebral artery territory (arrows). (D) and (E) The second and third X-maps obtained 4 hours and 1 day after the onset, respectively, show an abnormality in the left insula and temporal lobe lesions (arrows) and normalized left frontoparietal lobe lesions. (F) Follow-up fluid-attenuated inversion recovery-magnetic resonance images obtained 11 days after the onset show a final cerebral infarction in the left insula and temporal lobe (arrows) and no lesions in the left frontoparietal lobe. Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging.
Figure 3Serial radiological findings in case 4, a patient with acute ischemic stroke. (A), (B), and (C) The simulated standard, low kV, and high kV images obtained 3 hours and 2 minutes after onset show an early sign in the left frontal lobe (arrows). The simulated standard is equal to the high kV, and the low kV is inferior to the simulated standard and the high kV in detecting the acute ischemic lesion. (D) The first X-map obtained 3 hours and 2 minutes after the onset shows a definite abnormality in the left frontal lobe (arrows) more clearly than simulated standard CT. (E) Diffusion-weighted MRI obtained 3.5 hours after the onset shows an abnormality in the left frontal lobe (arrows). Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging.
The initial X-map revealed that the area with a lower attenuation value was observed more widely than the final infarction in 3 of 6 patients (cases 1, 2, and 6). In these 3 patients, the initial X-map was performed between 3 and 20 hours. Thus, there was a significant mismatch between final infarction and the area with lower attenuation value on the initial X-map. However, such a mismatch was observed for only 4-24 hours and spontaneously disappeared thereafter. In case 2, early DWI hyperintensity lesions representing reversible ischemia could be visualized on the first X-map (Fig 2).
The simulated standard CT was equal to the high kV CT in detecting acute ischemic lesions in all 6 patients (Figure 3, Figure 4). The low kV CT was inferior to the simulated standard CT and the high kV CT in detecting acute ischemic lesions in all 6 patients (Figure 3, Figure 4).
Figure 4Serial radiological findings in case 1, a patient with acute ischemic stroke. (A), (B), and (C) The simulated standard, low kV, and high kV images obtained 6 hours after onset show an early sign in the right parietal lobe (arrows). The simulated standard is equal to the high kV, and the low kV was inferior to the simulated standard and the high kV in detecting the acute ischemic lesion. (D) The first X-map obtained 6 hours after the onset shows a definite abnormality in the right parietal lobe (arrows) more clearly than simulated standard CT. This image also suggests another suspected abnormality in the right frontal lobe (arrowheads). (E) The second X-map obtained 1 day after the onset shows a definite abnormality in the right parietal lobe (arrows). This image also indicates a mild suspected abnormality at the right frontal lobe (arrowheads). (F) The third X-map obtained 28 days after the onset shows a definite abnormality in the right parietal lobe (arrows) and shows normalization of the suspected right frontal lobe lesion. Abbreviation: CT, computed tomography.
The X-map showed good uniformity in the cerebral hemisphere at the slice level of the basal ganglia, lateral ventricle, and centrum semiovale in all control subjects (Fig 5). The X-map showed laterality at the lower slice of the frontal and temporal lobe base in 3 of 15 (20%) control subjects.
Figure 5Normal control subject. (A), (B), and (C) The simulated standard CT shows no definite abnormal findings. (D), (E), and (F) The X-map shows good uniformity of cerebral hemisphere at the slice level of the basal ganglia, lateral ventricle, and centrum semiovale.
Discussion
In this preliminary study, the X-map clearly detected ischemic lesions, and it did so more sensitively than simulated standard CT, low kV, and high kV CT in all 6 patients with acute (<20 hours) stage of ischemic stroke. The X-map sensitively identified ischemic lesions even within 1 hour after the onset (30 minutes and 50 minutes in cases 2 and 3, respectively), whereas simultaneously simulated standard CT, low kV, and high kV CT could not. The X-map sensitivity was comparable to that of DWI, which is a promising result. To the best of our knowledge, this is the first study to suggest that the X-map is a possibly valuable diagnostic tool for patients with acute ischemic stroke.
We hypothesize that the brain parenchyma consists mainly of 3 materials: the gray matter, white matter, and water. An attenuation difference between the gray matter and white matter arises because gray matter contains 8% more oxygen and 8% less carbon, as a result of its higher water and lower lipid content, and this increases the photoelectric absorption.
10
The attenuation difference, except the water content, may result from the difference in their lipid content. If all voxels within the white matter are projected to the baseline using the characteristic lipid slope, the difference in attenuation level among each voxel on the baseline will reflect mainly the water content (Fig 1). In other words, the X-map can, thus, be a virtual gray matter map with water content using the 3MD technique.In our study, a low kV CT was inferior to simulated standard CT and high kV CT, and simulated standard CT was equal to high kV CT to detect acute ischemic lesions in all 6 patients. The most important early ischemic sign on CT is a loss of the gray–white matter interface. Although low kV CT may be superior to high kV in the contrast of the gray–white matter interface, greater contrast by lower X-ray energy may be offset by the noise increase at low energy. The noise decrease as well as excellent contrast of the gray–white matter interface may also be an important factor to detect an early ischemic sign on CT. However, the X-map is a new DECT technique to emphasize attenuation level changes by removing the normal difference in the attenuation level between gray matter and white matter.
In most acute ischemic stroke, the attenuation changes in the x-axis direction (high kV) largely determine whether the X-map shows the abnormality or not (Fig 6). The degree of change of the x-axis direction (high kV) is extended in what is projected on a diagonal baseline. Therefore, we named this imaging technique as the X-map.
Figure 6The mechanism of X-map. If normal brain tissue (green point) shifts to infarction 1 (red point in the red color area) because of edema (edema related to high kV dominant changes), infarction 1 (red point in the red color area) is projected to the red point on the baseline using the lipid slope, and the difference between normal and infarction 1 on the X-map is much larger than those on high kV or low kV. In addition, if normal tissue (green point) minimally shifts to minimal changes (blue points in the blue color area) because of minimal edema (minimal changes 1 or 2) or hyperemia (minimal changes 3), minimal changes (blue points in the blue color area) are projected to the blue point on the baseline using the lipid slope, and the difference between normal and minimal changes on the X-map is larger than those on high kV or low kV. However, if normal tissue (green point) shifts to infarction 2 (black point on the lipid slope line) as a result of edema with low kV dominant changes, infarction 2 (black point) is projected to the normal green point on the baseline using the lipid slope, and there is no difference between normal and infarction 2 on the X-map. If cerebral infarction-induced attenuation changes (due to low kV dominant edema) from the normal state is in a more acute angle direction more than the lipid slope angle, the X-map cannot show acute ischemic stroke. The infarction in the gray color area has the potential to be both for the difference between normal tissue and the infarction to become larger or smaller on the X-map. (Color version of figure is available online.)
Recent studies have shown that earlier thrombolytic therapy could result in better clinical outcome, but this treatment is potentially hazardous and causes fatal hemorrhagic transformation in patients with dense cerebral ischemia and/or extended tissue damage.
11
Using standard CT scan, however, it is sometimes difficult to identify early ischemic signs in a certain subgroup of patients. MRI is widely known to be more sensitive to detect irreversibly damaged lesions in the super-acute stage of ischemic stroke, but it requires a longer time for diagnosis than standard CT scans.12
Therefore, because it is more easily accessible than MRI, the X-map may possibly contribute to improving the prognosis of patients with acute ischemic stroke by shortening the time for radiological screening before thrombolytic therapy.In this study, in 3 of 6 patients (cases 1, 2, and 6), the area with the lower attenuation value was more widely observed on the X-map than final cerebral infarct. This mismatch between the X-map and follow-up CT/MRI was observed between .5 and 24 hours after the onset and disappeared thereafter (Table 1). In case 2, early DWI hyperintensity lesions representing reversible ischemia could be visualized on the first X-map (Fig 2).
Although pathophysiological mechanism is still obscure, some explanations may be possible. That is, the minimal attenuation changes around the cerebral infarction may be related to reversible edema or hyperemia. Theoretically, the decreased attenuation on the X-map most likely reflects tissue edema in the x-axis direction (high kV) only or predominantly (Fig 6; minimal changes 1 or 2). Cerebral blood flow is moderately reduced, but the tissue is still vital in the peri-infarct area, called the “penumbra.” A recent multimodality study has shown that oxygen metabolism is also suppressed up to 60% of the contralateral value even in the peri-infarct area (normal diffusion initially and on day 3).
13
Therefore, an acute metabolic disturbance may lead to the oscillation of the water content in the gray matter of the hypoperfused, but still vital tissue around cerebral infarct, which is expressed as transient, minimal attenuation changes on the X-map. Alternatively, the decreased attenuation on the X-map may also reflect the minimal increase in the attenuation level because of hyperemia in the y-axis direction (low kV) only or predominantly (Fig 6; minimal changes 3).In case 1, a slight suspected abnormality in the left frontal lobe, in addition to an abnormality in the right frontal lobe, is suggested on the first X-map (Fig 4D). The third X-map obtained 28 days after the onset shows normalization of the suspected bilateral frontal lobe suspected lesions (Fig 4F). These changes were not consistent with the vessel territory. It is still unknown if these findings reflect true conditions or if they were false lesions caused by artifacts. Further examinations are essential.
There are several limitations to this study. First, it is quite easy to visualize the damaging lesions on the X-map in the acute (<20 hours) stage of ischemic stroke when the lesions have a significant volume. However, in theory, the X-map cannot show acute ischemic lesion because of low kV (y-axis) dominant edema. In addition, smaller ischemic lesions may be difficult to identify because of their lower spatial resolution and partial volume effect on the X-map. Therefore, the X-map cannot replace simulated standard CT for the diagnosis of acute ischemic stroke at this time. Future optimization of X-map algorithm parameters, including the baseline, slope, and their combination, may improve the sensitivity to detect y-axis dominant edema lesions and small lesions on the X-map. Second, it is essential to precisely compare the findings on the X-map with those on cerebral blood flow imaging such as SPECT and PET, which would explain the underlying pathophysiology of the transiently lowered attenuation in the peri-infarct area on the X-map. Finally, the number of patients was limited and no statistical analysis was performed. Further studies with a larger cohort are warranted to confirm the findings in this study.
Conclusion
This preliminary study indicated that the X-map can identify acute ischemic lesions more clearly than simulated standard CT. The X-map may also potentially detect minimal attenuation changes that have not ever been noticed on other modalities. Using noncontrast DECT, the X-map may be a valuable tool to identify ischemic lesions and to determine therapeutic strategy as early as possible in patients with acute ischemic lesions because of its easy accessibility. The X-map cannot replace the simulated standard CT in diagnosing acute ischemic stroke at this time. However, the X-map is a powerful tool to supplement simulated standard CT for characterizing acute ischemic lesions. Further examination and refinement of the algorithm is required.
References
- Data science of stroke imaging and enlightenment of the penumbra.Front Neurol. 2015; 6: 8
- General principles.AJR Am J Roentgenol. 2012; 199: S3-S8
- Dual-and multi-energy CT: principles, technical approaches, and clinical applications.Radiology. 2015; 276: 637-653
- Applications of dual-energy CT in emergency radiology.AJR Am J Roentgenol. 2014; 202: W314-W324
- Dual energy CT: preliminary observations and potential clinical applications in the abdomen.Eur Radiol. 2009; 19: 13-23
- Dual-energy CT: clinical applications in various pulmonary diseases.Radiographics. 2010; 30: 685-698
- Dual energy CT: what the neuroradiologist should know.Curr Radiol Rep. 2015; 3: 16
- Imaging of contrast medium extravasation in anticoagulation-associated intracerebral hemorrhage with dual-energy computed tomography.Stroke. 2013; 44: 2883-2890
- Dual-energy CT virtual noncalcium technique: detecting posttraumatic bone marrow lesions—feasibility study.Radiology. 2010; 256: 617-624
- Explanation of cerebral white–gray contrast in computed tomography.J Comput Assist Tomogr. 1980; 4: 489-491
- Time to treatment with intravenous alteplase and outcome in stroke: an updated pooled analysis of ECASS, ATLANTIS, NINDS, and EPITHET trials.Lancet. 2010; 375: 1695-1703
- Advanced imaging to extend the therapeutic time window of acute ischemic stroke.Ann Neurol. 2013; 73: 4-9
- Metabolic penumbra of acute brain infarction: a correlation with infarct growth.Ann Neurol. 2005; 57: 495-504
Article Info
Publication History
Published online: September 16, 2016
Accepted:
August 17,
2016
Received in revised form:
July 22,
2016
Received:
June 7,
2016
Identification
DOI: https://doi.org/10.1016/j.jstrokecerebrovasdis.2016.08.025
Copyright
© 2017 The Authors. Published by Elsevier Inc. on behalf of National Stroke Association.
User License
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0) | How you can reuse 
Elsevier's open access license policy
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0)
Permitted
For non-commercial purposes:
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article (private use only, not for distribution)
- Reuse portions or extracts from the article in other works
Not Permitted
- Sell or re-use for commercial purposes
- Distribute translations or adaptations of the article
Elsevier's open access license policy
30297-X&id=yjscd2783-fig-0001.jpg){kind=link}
30297-X&id=yjscd2783-fig-0002.jpg){kind=link}
30297-X&id=yjscd2783-fig-0003.jpg){kind=link}
30297-X&id=yjscd2783-fig-0004.jpg){kind=link}
30297-X&id=yjscd2783-fig-0005.jpg){kind=link}
30297-X&id=yjscd2783-fig-0006.jpg){kind=link}