FYI. TLDR. So go ask your doctor how this is going to get you to 100% recovery. And keep asking until they finally acknowledge that they know nothing on how to get there!
Clinical Applications of Dual‐Energy Computed Tomography for Acute Ischemic Stroke
Abstract
Acute ischemic stroke is a leading cause for neurological disability worldwide, and treatment strategies are rapidly evolving. Patient selection for recanalization therapy and postintervention management relies heavily on diagnostic imaging. In this narrative review, we searched the existing literature for clinical applications of dual‐energy computed tomography for acute ischemic stroke. We summarized the current clinical evidence on the use of dual‐energy computed tomography for identifying early cerebral ischemia, detecting and predicting hemorrhagic transformations, and characterizing clots and stenotic plaques. We also highlight future opportunities for dual‐energy computed tomography to be used to address important diagnostic challenges during acute stroke triage and postintervention management. Dual‐energy computed tomography is a powerful tool that can be used to improve the diagnostic accuracy of ischemia, hemorrhage, and vascular lesions in the context of acute ischemic stroke.
Acute ischemic stroke (AIS) is a leading cause of neurological disability worldwide, and its incidence is on the rise with the global population aging.1 Clinical care for AIS has evolved over the past 3 decades, and, recently, endovascular thrombectomy (EVT) has emerged as a powerful treatment for AIS due to large‐vessel occlusions and has become standard of care for select patients.2, 3 Currently, patient selection for acute stroke therapy relies heavily on diagnostic imaging to promptly rule out hemorrhage and identify infarcted or at‐risk tissue. Neuroimaging also plays a critical role after acute AIS treatment, particularly for prevention and management of hemorrhagic transformations. While much work has been done to leverage imaging techniques to better characterize ischemic tissue during acute stroke triage and manage hemorrhage following treatment, many patients still do poorly,4 signaling a need for better diagnostic and predictive tools to further optimize patient triage and management.
Dual‐energy computed tomography (DECT) is a unique tool that allows for material decomposition on the basis of their differences in the change of x‐ray attenuation from one energy level to another.5 In addition, DECT images can also be reconstructed into virtual monoenergetic images (VMIs), which may improve iodine attenuation and suppress artifacts.6 Clinical availability of DECT is rapidly expanding, with a survey of chest radiologists showing that DECT is available in up to 75% of academic institutions worldwide.7 In this review, we summarize the current clinical data on the use of DECT to identify and predict hemorrhage, assess blood–brain barrier (BBB) and endothelial damage, identify strokes, and characterize stroke clots and carotid plaques. We also highlight current knowledge gaps in stroke medicine that could be addressed by DECT in future research.
PRINCIPLES OF DECT
The principle of computed tomography (CT) is based on the interactions of x‐ray photons with matter. When a photon strikes an atom, some energy can be lost in the process, resulting in energy attenuation. The amount of attenuation is measured by Hounsfield units, and it is driven by 2 factors: (1) the mass density of the material and (2) the attenuation coefficient. The attenuation coefficient is specific to each material and the energy of the incident x‐ray photon. As such, it is possible for different materials with different mass densities to have similar Hounsfield units on conventional CT.
Within the X‐ray energy range (≈40–140 kilo‐electron volts [keV]), the attenuation coefficient of each material is composed of 2 competing mechanisms of light‐matter interactions: Compton scattering and photoelectric effect.8
Compton scattering is a phenomenon where the incident x‐ray photon ejects or ionizes a loosely bound electron, leading to a partial reduction of the incident photon energy and an altered photon trajectory.9 The amount of energy attenuation and the degree of trajectory change due to Compton scattering are not associated with the size of the atom. If the trajectory change is small, the attenuated photon would be received by the CT detector at an unexpected location, resulting in noise; if the trajectory change is large, the photon would not strike the CT detector at all and appear completely attenuated. Compton scattering is the predominant mechanism of x‐ray attenuation at high photon energies.
Separately, the photoelectric effect is a phenomenon where the incident photon ejects or ionizes a tightly bound electron (typically in the innermost K‐shell of a large atom), leading to the absorption and complete attenuation of the photon's energy. The probability of the occurrence of the photoelectric effect is higher for matter with high atomic number (Z), and it is also highly dependent on the energy of the photon. If the incident photon provides sufficient energy to overcome the binding energy of the K‐shell electron (also termed the K‐edge), the probability of ejecting the K‐shell electron is inversely proportional to the cube of the incident photon's energy. Thus, the photoelectric effect is most likely to occur when the incident photon carries an energy at or slightly above the K‐edge energy.10
CLINICAL PERSPECTIVE
Dual‐energy computed tomography is a powerful tool that allows for material decomposition and generation of virtual monoenergetic images.
Processed dual‐energy computed tomography images can improve identification of early stroke, detection and prediction of hemorrhage, and characterization of stenotic vascular lesions.
High‐atomic‐number materials have higher K‐edge values. For example, iodine has a K‐edge of 33.2 keV. Thus, in the x‐ray energy range, there is a sizeable probability of energy attenuation due to the photoelectric effect, especially at low x‐ray energies.11, 12 However, since the probability of photoelectric attenuation exponentially decreases with higher x‐ray energies, low‐energy photons are significantly more likely than high‐energy photons to be attenuated by the photoelectric effect of large atoms, resulting in significant differences in attenuation coefficients across the different x‐ray energies. On the other hand, common elements in the human body (hydrogen, carbon, nitrogen, and oxygen) all have low atomic numbers, in which the K‐edge energies are low. For these materials, the probability of photoelectric interactions is low at all levels of x‐ray energies,11, 12 resulting in only small differences in attenuation coefficients between low and high photon energies.
Conventional CT uses an x‐ray tube with a peak energy of 120‐kilovolt peak, which generates a single spectrum of x‐ray photons that carries energies up to 120 keV. With a single energy spectrum, it is not possible to differentiate 2 materials purely on the basis of their net attenuation since neither the attenuation coefficient nor the mass densities are known. With DECT, however, 2 spectra of x‐ray energies (typically 80 and 140 kilovolt peak) are used, and the changes in attenuation coefficients between the 2 energy spectra (which are primarily driven by the photoelectric effect) can be used to identify materials with high atomic numbers.13 Thus, DECT can be a powerful tool to characterize material composition, and it has the ability to create maps of a specific material's contribution to overall attenuation (eg, iodine map), and also subtract a specific material's contribution to the overall attenuation (eg, virtual noncontrast or VNC images).
Obtaining CT images using 2 spectra of x‐ray energies also allows for image reconstructions to simulate what would result from acquisitions with monoenergetic x‐ray beams at hypothetical energy levels.14 These images are termed VMIs. At a high theoretical energy level far from the K‐edges of tissue and contrast, VMIs can reduce beam hardening artifacts and improve overall image quality 15; at a low hypothetical energy level close to the K‐edge of iodine, VMIs can accentuate photoelectric attenuation and improve iodine contrast‐to‐noise ratio.16
Further details regarding the principles and physics of DECT are outlined in prior reviews.11, 12, 13, 14 In real‐world practice, the multitude of sequences (eg, material decomposition or VMI) that can be reconstructed from imaging data acquired from commercially available DECT scanners are rapidly processed automatically without any human intervention. Thus, DECT can be easily incorporated into clinical use, even during time‐sensitive aspects of patient care such as during acute stroke triage.
IDENTIFICATION OF HEMORRHAGE
As EVT procedures proliferate worldwide, reliable identification and management of downstream complications such as hemorrhagic transformations are needed for optimizing patient care. Rates of hemorrhagic transformation after EVT may also increase with expanding EVT indications; for example, in the ANGEL‐ASPECT (Endovascular Therapy in Acute Anterior Circulation Large Vessel Occlusive Patients With a Large Infarct Core) trial of EVT for large ischemic cores, nearly half of the EVT‐treated patients developed subsequent hemorrhage,17 which is markedly higher than earlier EVT trials. Early identification of hemorrhagic transformation is particularly important for determining the timing and dosage of antithrombotic medications, obtaining further follow‐up imaging to determine stability as well as potential tighter regulation of blood pressure. Prior randomized trials have demonstrated the importance of prompt initiation of antithrombotic agents in preventing early stroke recurrence,18 and mischaracterizing postintervention contrast staining as hemorrhagic transformation can lead to treatment delays and potentially impact patient outcomes. However, the persistence of iodinated contrast used for EVT procedures can significantly confound the ability of conventional CT to distinguish hemorrhage from contrast staining, potentially leading to misdiagnosis and suboptimal care.19
One of the most widely used features of DECT in stroke is material decomposition for distinguishing iodinated contrast from hemorrhage. Numerous studies have shown that DECT can reliably distinguish contrast staining versus hemorrhage after spontaneous intracranial hemorrhage (ICH)20, 21 and hemorrhagic transformations of ischemic stroke.22, 23, 24, 25 In addition to characterizing parenchymal lesions, DECT can also be used to for hyperdense lesions in other brain compartments such as the subdural and subarachnoid spaces.26 Using DECT, VNC images can be generated by removing contrast‐related hyperdensities from iodine uptake, and this can be used to identify acute hemorrhage. Representative images are presented in Figure 1.
Figure 1. Representative images of a patient with a left middle cerebral artery stroke and contrast staining (blue arrows in A, B, and C) following endovascular thrombectomy, and another patient with a left middle cerebral artery stroke with hemorrhagic transformation (yellow arrows in D, E, and F) following endovascular thrombectomy. As can be appreciated, conventional CTs show increased attenuation in the infarct territory, while VNC can identify areas of hemorrhage and the iodine map can visualize areas with contrast staining. CT indicates computed tomography; and VNC, virtual noncontrast imaging.
In 2020, a systematic review and meta‐analysis of 9 studies demonstrated that DECT can be a sensitive (96%) and specific (98%) tool for differentiating acute ICH from contrast or calcifications.27 Interestingly, the study also found that DECT may be more specific (98.7% versus 92.6%) for identifying hemorrhagic transformation after AIS than other forms of hemorrhages, though it may be less sensitive (43.5% versus 94.2%).27 A subsequent systematic review and meta‐analysis of 7 retrospective studies in 2022 showed that for patients with CT hyperdense lesions after EVT, DECT was able to identify hemorrhage with 77% sensitivity, 100% specificity, and 99% accuracy.28 DECT can also help distinguish contrast and hemorrhage after intravenous thrombolysis.29 The reasons underlying disparate sensitivity measures for DECT are likely multifactorial. Timing of DECT may be an important factor, as imaging in the early period may be less accurate than late imaging.30 Furthermore, the timing of the confirmatory scan (either via a later magnetic resonance imaging or CT) to identify hemorrhage may reflect delayed hemorrhagic transformation, which could be preceded by pure contrast staining and thus lead to the misinterpretation of a negative early DECT scan as false negative. Putting variability of sensitivity aside, these results demonstrate that DECT can be a powerful and specific tool in distinguishing hemorrhagic transformation versus contrast staining following AIS and recanalization therapy.
Accurate and early identification of hemorrhagic transformations with DECT overall leads to lower rates of misdiagnoses of ICH. In a retrospective study of 372 patients undergoing EVT, Almqvist et al showed that the use of DECT within 36 hours after EVT changed the diagnosis of ICH to contrast staining in 34% of patients, and DECT had higher interreader agreement compared with conventional CT.19 While the sensitivity of DECT may be lower immediately after EVT procedures, a retrospective study by Liu et al of 106 patients showed that when used immediately after EVT, DECT reduced the diagnosis of hemorrhage from 74.5% to 10.4%.31 More recently, flat‐panel DECTs in the angiography suite have also been shown to be feasible, which could allow for even earlier assessments of post‐EVT hemorrhage32; however, this technology is primarily limited to research applications at this time, and it is not yet widely available for clinical use. Overall, DECT can be a powerful tool to increase the diagnostic confidence of hemorrhagic transformations following EVT procedures, especially during the early hours after EVT.
BLOOD–BRAIN BARRIER INTEGRITY AND PREDICTION OF HEMORRHAGIC TRANSFORMATIONS
While DECT can be used to remove iodinated contrast signals to generate VNC images to assess for hemorrhage, the iodine maps themselves (Figure 1) can also be clinically useful in quantifying the integrity of the BBB33 and predicting future hemorrhage.34, 35, 36, 37, 38
EVT procedures have been associated with endothelial damage and BBB disruption,39, 40 and the number of stent retriever passes has been associated with contrast extravasation on DECT iodine maps.41 Overall, contrast extravasation on DECT likely indicates BBB disruption, and in a prospective cohort of 132 patients undergoing EVT, Renú et al demonstrated that contrast staining is associated with higher odds of poor patient outcomes.42 In a separate study of 402 patients undergoing EVT, Pinckaers et al found that the presence of contrast extravasation was common and associated with worse outcomes and stroke progression.43 Other studies have also reported similar findings, both for the anterior37 and posterior circulation EVTs.44 Finally, a meta‐analysis of 11 studies including 1123 patients undergoing EVT showed that contrast staining on DECT after EVT may be associated with higher risk of hemorrhage and poor outcomes.45 Overall, it remains unclear if BBB disruption represents clinically significant iatrogenic injury or simply reflects more established infarcts, and the pathophysiological link between BBB disruption and stroke outcomes awaits further investigation.
Presence and extent of contrast extravasation can predict ICH. In a study of 85 patients who received DECT immediately after EVT, Bonatti et al showed that all patients with ICH had contrast staining (representing perfect sensitivity and negative predictive value), and that iodine concentration had an area under the receiver operating curve of 0.89 for identifying patients with hemorrhagic transformation.35 In another study of 71 patients, Byrne et al demonstrated that 55% of patients had iodine staining following successful EVT, and that parenchymal iodine concentration as a relative percentage of iodine concentration in the superior sagittal sinus enabled 95% sensitivity of identifying patients who later developed ICH.38 Other studies further corroborated these results.34, 37 DECT parameters can also be combined with clinical markers to predict hemorrhagic transformation following EVT.46 Overall, these studies indicate that DECT may be a powerful tool to predict hemorrhagic transformation following EVT.
ICH risk stratification may be particularly important for establishing safe blood pressure goals following EVT. Recently, the ENCHANTED2/MT (Second Enhanced Control of Hypertension and Thrombectomy Stroke Study)47 and OPTIMAL‐BP (Outcome in Patients Treated with Intraarterial thrombectomy ‐ optiMAL Blood Pressure control)48 trials suggested that intensive blood pressure control following EVT may lead to worse outcomes. However, while the general goal of lower blood pressure targets is to prevent hemorrhagic transformation, these trials did not employ strategies to identify patients at higher risk of ICH. In a separate study using a magnetic resonance imaging–based biomarker for BBB integrity, Upadhyaya et al suggested that for patients with evidence of BBB disruption, lower blood pressures are associated with favorable outcomes.49 Future studies are needed to assess the utility of post‐EVT DECT iodine maps and iodine quantification for ICH risk stratification and optimizing blood pressure goals.
DECT can also assess subarachnoid contrast extravasation after EVT, and this too has prognostic value. In a retrospective study of 424 patients who underwent EVT, Renú et al suggested that diffuse subarachnoid hyperdensities after EVT are associated with worse outcomes, regardless of whether the subarachnoid contents were contrast or blood.50 In a separate study of 781 patients undergoing EVT, Benalia et al demonstrated that pure subarachnoid hemorrhage did not impact long‐term outcomes.51 These studies may suggest that the extent of post‐EVT endothelial damage demonstrated by contrast extravasation into the subarachnoid space may impact patient outcomes, and future studies are needed to better characterize this phenomenon.
Antiplatelet dosing and timing can be particularly challenging for patients with tandem carotid occlusions concurrently treated with EVT and carotid stents. For these patients, providers must balance the risk of stent occlusion with the risk of hemorrhagic transformation. In a pilot study of 50 patients with stroke with tandem occlusions treated with intracranial EVT and carotid stenting, Murias et al demonstrated that a protocol using the presence of hemorrhage and contrast staining to guide the initiation of antiplatelet agents led to concurrently low rates of hemorrhagic and thrombotic complications.21 Future studies are needed to further investigate the utility of DECT for optimizing the timing and dosage of antithrombotic agents after acute treatment of tandem occlusions.
EARLY ISCHEMIC CHANGES AND INFARCT BURDEN
Despite its limited ability to characterize ischemic tissue compared with magnetic resonance imaging, CT remains the primary imaging modality during acute stroke triage at most institutions, and most trials that established the efficacy of EVTs have employed CT‐based inclusion and exclusion criteria.52, 53, 54, 55, 56, 57, 58, 59, 60 As such, current clinical practice and guidelines rely heavily on CT modalities for patient selection.
While early ischemic changes following acute stroke can be visualized on conventional noncontrast CT studies,61 the findings may be subtle, and interrater reliability has been historically low.62 Despite the imperfection of conventional CTs for identifying early ischemia, recent trials have shown that EVT may be beneficial for patients with large areas of early ischemic changes.17, 63 However, it remains unclear if the involved areas seen on conventional CT are fully infarcted.64 Thus, better characterization and identification of infarction may be needed to optimize patient selection for EVT.
One CT imaging biomarker of early ischemia is loss of gray–white matter differentiation.61 VMIs are reconstructions from DECT that can reduce image artifacts and improve tissue contrast, thus enhancing the visualization of gray–white matter differentiation. Numerous studies have demonstrated the feasibility of using VMIs to identify early ischemic changes.65, 66, 67 In a retrospective study of 125 patients, 80 and 90 keV VMIs were significantly superior to traditional noncontrast CT for detection and localization of infarcts.68 Other studies also demonstrated that VMIs may be more accurate in identifying early ischemia,69 and diagnostic performances of VMIs are not statistically different across energy levels.67 At our institution, we primarily use 190 keV VMIs as they have an added benefit of improving diagnostic accuracy of intracranial hemorrhages.70, 71, 72 Furthermore, while not optimized to detect early stroke, VNC images have also been suggested to be superior to conventional CT images.73, 74 Representative images of DECT when used to characterize AIS are presented in Figure 2.
Figure 2. Representative conventional CT (A) and 190‐keV virtual monoenergetic image (panel B) of a patient with acute ischemic stroke with confirmatory MRI showing cerebral infarction. As can be seen, high‐energy VMIs accentuates the visualization of changes in material density, which allows for better identification and localization of infarcted tissue. CT indicates computed tomography; DWI, diffusion‐weighted imaging; keV, kilo‐electron volts; MRI, magnetic resonance imaging; and VMI, virtual monoenergetic image.
VMI's ability to suppress artifacts and enhance gray–white matter differentiation may be particularly helpful for visualizing acute ischemia in the posterior fossa. While methods exist for quantifying early ischemia in the posterior circulation using conventional noncontrast CT,75 these images are prone to artifact, particularly in the brainstem.76 Furthermore, unlike for the anterior circulation,57 CT perfusion metrics have not been validated for the posterior circulation. Thus, providers are often left with subpar images when making treatment decisions. Importantly, the success of the ATTENTION (Endovascular Treatment for Acute Basilar Artery Occlusion)59 and BAOCHE (Basilar Artery Occlusion Chinese Endovascular Trial)60 trials in demonstrating the efficacy of EVT for acute basilar strokes can be attributed to the exclusion of patients with extensive signs of early ischemic changes on conventional CT,77 which highlights the importance of reliably quantifying early ischemia in patients with acute posterior circulation strokes. Furthermore, posterior circulation strokes can present with nonspecific neurological symptoms (eg, dizziness, vertigo, headache), and in the absence of visible vascular occlusions, early identification of acute ischemia can also aid thrombolytic decision making. In a study of 30 patients with symptoms of posterior fossa stroke, Hixson et al showed that VMIs (80 or 100 keV reconstructions) significantly reduced artifacts and improved image quality, resulting in overall improved stroke detection.78 Future studies are needed to further explore the utility of DECT in quantifying early ischemia in posterior circulation strokes, and whether improved imaging quality offered by DECT may assist treatment decisions.
The underlying pathophysiology of AIS involves failure of cell surface ionic pumps, which leads to cytotoxic edema.79 Thus, prompt identification of tissue edema may be an alternative, and perhaps more direct, method of identifying early strokes. Instead of enhancing gray–white matter differentiation with VMIs, material decomposition with DECT can be used to suppress gray–white matter differentiation to generate edema maps. Various methods have been described. In 2017, Noguchi et al80 described the “X‐map” (analogous to a DECT bone marrow edema map81) where fat signals are subtracted to suppress native gray–white matter differentiation, thereby enhancing the appearance of tissue edema due to early ischemic changes. Subsequently, Mohammed et al82 and Grams et al74 described similar techniques. Overall, using edema maps to identify early strokes increased the sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating curve than conventional noncontrast CT.74, 82 In additional, edema maps can also be used to quantify and predict ischemic core size.83, 84, 85 While edema maps may provide valuable information, our empiric institutional experience is that high‐energy VMIs provide higher‐quality images for AIS; future head‐to‐head studies are needed to compare the comparative performance of edema maps with other DECT modalities during acute stroke triage.
CLOT COMPOSITION
Intracranial clots can be grossly categorized into red blood cell (RBC)‐rich and fibrin‐rich clots, and prior studies have suggested that RBC clots may be more amenable to intervention.86 Thus, detailed characterization of clot composition may present future opportunities to develop tailored devices and personalized care for each patient.
Using material decomposition, DECT can provide valuable insight into clot composition. Past studies have demonstrated that DECT can differentiate RBC‐ and fibrin‐rich clots in both in vitro87, 88 and in vivo animal models.89 In a clinical study of 88 patients who underwent dual‐energy CT angiography (DE‐CTA), Jiang et al demonstrated that higher VNC values and lower iodine concentrations are associated with RBC‐dominant clots.90 VMIs reconstructed at 80 keV can also be used to distinguish RBC‐rich and RBC‐poor clots.91 Furthermore, DE‐CTA can also help assess thrombus perviousness; in a study of 86 patients who underwent DE‐CTA, iodine concentrations within thrombi were independently associated with favorable outcomes, higher likelihood of recanalization, and smaller final infarct volumes.92 Thus, DECT and DE‐CTA may emerge as useful tools to characterize clot composition and perviousness, and future research is needed to translate this information into actionable items for clinical care.
CAROTID PLAQUE CHARACTERIZATION
Carotid stenosis is a risk factor for ischemic stroke,93 and accurate prediction of stroke risk is critical for treatment decisions, particularly for asymptomatic lesions.94
One key determinant of stroke risk is degree of stenosis95; however, as carotid plaques are often calcified, degree of stenosis can be exaggerated on conventional CT angiogram, potentially leading to unnecessary treatment.96 With material decomposition, DE‐CTA can be used to suppress calcium signals to better visualize luminal narrowing (Figure 3). In a retrospective study of 36 patients with 46 stenotic lesions, Korn et al demonstrated that DE‐CTA allowed the grading of stenosis in lesions that were otherwise impossible with conventional noninvasive imaging.97 Mannil et al showed similar results, confirming that conventional CTA resulted in overcalling stenotic lesions, while DE‐CTA was able to accurately assess grade of stenosis compared with digital subtraction angiography.98
Figure 3. Representative conventional CTA (A and B) and DE‐CTA with calcium subtraction (C, D, and E) of a patient with calcified plaques in the proximal internal carotid arteries. As can be seen, the carotid lumen is better visualized after calcium subtraction. CTA indicates computed tomography angiography; DE‐CTA, dual‐energy computed tomography angiography; and MIPs, maximum intensity projections.
Carotid plaque composition can also influence subsequent stroke risk.99 To this end, DE‐CTA can also be used to characterize plaque composition. In a clinical study of 30 patients with histological correlates, Das et al showed that DECT may be a sensitive and accurate tool to identify calcified, mixed, and fatty plaques.100 In a separate study of 12 patients with histological correlates, Saito et al demonstrated that DECT can be used to identify cholesterol crystals in carotid plaques.101 Iodine penetration into carotid plaques can also be assessed by DE‐CTA, which, interestingly, has been associated with higher burdens of cerebral white matter disease.102 Information regarding plaque composition may also help predict stroke risk or treatment response; as such, machine learning techniques can incorporate DECT radiographic biomarkers with clinical features to predict symptomatic carotid plaques.103 Future research is needed to translate DECT's ability of characterizing carotid plaque composition into improved patient care.
CONCLUSIONS
DECT is a powerful technology that allows for more accurate characterization of hemorrhage and edema. It also has the capability of assessing the integrity of the BBB and the composition of stroke clots and carotid plaques (Table 1). Future studies are needed to further explore the utility of DECT for acute stroke triage and management in routine clinical practice. We have also included our institutional protocol for DECT for reference in the Supplementary Materials.
More at link.
No comments:
Post a Comment