Dual energy CT: a step ahead in brain and spine imaging

Abstract

Objective:

The purpose of this pictorial essay is to illustrate the utility of dual energy CT as an adjunct or alternative to routine single energy CT (SECT) scan of the brain and spine in emergency neuroradiology practice.

Conclusion:

Dual energy CT can be used as a problem-solving tool in brain and spine imaging. It enables one to make a confident and accurate diagnosis for a variety of clinical conditions thereby impacting patient management.

Introduction

In recent years, there has been an increasing number of studies on the technology, diagnostic capabilities and limitations of dual energy CT (DECT). Exploiting the fact that materials with higher atomic numbers (such as iodine and calcium) attenuate X-rays differently when exposed to low and high peak kilovoltage, DECT uses specific post-processing algorithms to break down voxels into their material components. Blended images, basis material decomposition maps, and virtual monochromatic images can be generated to answer specific clinical questions.¹ Oftentimes, DECT is acquired with comparable or less patient radiation exposure than with conventional single energy CT (SECT).²

DECT can quickly and accurately diagnose specific conditions that are especially important in the trauma setting.^1,3 DECT may also be useful for patients for whom MRI is contraindicated, are in a location where MRI is unavailable, or is associated with lengthy wait-times. The use of DECT of the brain and spine has now become widespread, and new indications are constantly emerging. The purpose of this article is to present an illustrated review of current applications of DECT in the brain and spine.

Principles of dual energy CT

DECT is based on the attenuation difference of various materials when simultaneously exposed to two X-ray energy levels by setting the X-ray tube voltage at high and low peak voltage (kVp) levels.^1,4,5 Typical energies used are 80 kVp for lower energy spectrum and 140–150 kVp for the high-energy spectrum. At medical imaging energy levels, the attenuation of materials found in human body is primarily due to the photoelectric absorption and compton scattering. The photoelectric absorption depends on the X-ray beam energy, the atomic number (Z) of the material under consideration, and the K-edge effect which results in sudden increase in attenuation for photon energies greater than the K-shell binding energy of the material. Compton scatter is dependent on X-ray energy and tissue electron density, mainly at lower X-ray energy levels with relatively constant atomic number. Since each material has a unique electron density and atomic number, DECT uses this inherent variability of different substances in order to infer material composition at different energy levels. Therefore, DECT allows to separate different materials as image acquisition is at two different energy levels.

DECT data analysis and post-processing

DECT enables the generation of additional reconstructions and quantitative analysis, not possible with conventional SECT acquisition.

The most commonly used DECT image reconstructions include:

1. Weighted average (WA) or blended images.

2. Basis material decomposition (BMD) Maps [e.g. Virtual non-contrast (VNC), iodine overlay and bone/calcium subtraction images].

3. Virtual monochromatic images (VMI’s)

Weighted average or blended images

The default reconstructions produced by most dual-source for clinical analysis are a linear mix of low- and high-energy acquisition information blended in distinct proportions to generate images.

A combination of 30% low-energy and 70% high-energy data, are typically considered equivalent to the standard 120 kVp SECT acquisition on all available DECT scanners.^4–6

Basis material decomposition (BMD) maps

The data set from different energy levels are used to identify materials based on their elemental configuration, to evaluate the tissue distribution and estimate the concentration of a given material or element. Three-material decomposition techniques can be reconstructed to create Iodine overlay, VNC, calcium maps and virtual non-calcium (VNCa) images.^1,5,6

Virtual monochromatic images (VMI)

VMI is the most common type of reconstruction obtained by combining data from the low and high energy DECT acquisitions. The process is carried out using a sophisticated algorithm that accounts for the energy dependent behavior of photoelectric effect and compton scattering. Typically, reconstructed VMI energies range between 40 and 140 keV. A 65–70 keV is considered equivalent to the standard 120 kVp SECT.^1,5,6

Dual-energy unenhanced head CT reconstructed at 65 keV optimizes gray-white matter differentiation of the supratentorial brain parenchyma while the posterior fossa is better evaluated at 75 keV (Figure 1). At lower monochromatic energy levels, attenuation increases for high-atomic-number materials such as iodine and calcium and decreases for low-atomic-number materials such as fat. Similarly, there is greater image noise at lower VMI energies, representing a trade-off between iodine attenuation and noise.

Figure 1.

Dual-energy unenhanced head CT through the posterior fossa in a 35-year-old female. Virtual monochromatic energy levels of 40, 75 and 140 Key's were reconstructed. At lower keV, there is increased image noise and streak artifact. At higher keV, there is less artifact but reduced soft-tissue contrast. Optimal evaluation of posterior fossa is demonstrated at 75 keV.

DECT clinical applications of the brain

Post-processing of CT angiography

Various techniques including manual segmentation of bones and vessels, Threshold-based semi-automatic segmentation and digital subtraction of non-contrast CT from the CT angiography (CTA) images have been employed to remove bone and facilitate better visualization of the cerebral vasculature. Each of these individual techniques has drawbacks that include operator dependence, susceptibility to subtraction and motion artifacts, and increased radiation dose.

DECT-based bone removal technique is a single-click physics-based removal of bones from the image. It is effective in elucidating cerebral vascular anatomy while overcoming limitations of previously used conventional methods^1,7 (Figure 2).

Figure 2.

Comparison of bone removal algorithm for dual-energy CT (a) and conventional bone removal CT angiography (b) demonstrating superiority of dual-energy CT in visualization of cerebral vasculature.

Evaluation of atherosclerotic arterial stenosis

Calcified atheromatous plaque appears isodense to the adjacent contrast-enhanced lumen and often impedes assessment of intra- and extracranial vasculature stenosis on Carotid CT angiography (Figure 3). DECTA technique using a hard-plaque algorithm based on calcium material decomposition, highlights the plaque and allows better visualization of the residual lumen^1,3,8 (Figure 4).

Figure 3.

Dual-energy CTA of head and neck (a, b) in a 73-year-old male acquired at 80 and 140 kVp suggests focal dissection/atherosclerotic ulcer (arrow) of the proximal brachiocephalic artery. Post-processing of the imaging data using a hard plaque application algorithm (c, d) and virtual non-contrast image (e, f) by iodine subtraction confirmed calcified atherosclerotic plaque (arrow).

Figure 4.

Dual-energy CTA of the head (a) in an 82-year-old male acquired at 80 and 140 kVp shows peripherally calcified plaques of the cavernous segments of bilateral internal carotid arteries. Post-processing of the imaging data provides plaque characterization obtained using a hard plaque application algorithm (b) differentiates iodine (blue) from calcium (red), bone removal algorithm (c) and MIP images (d) provides better visualisation of the intracranial vasculature.

Differentiation of hemorrhage and calcification

Intracranial hemorrhage and calcification, both of which appear hyperdense on conventional CT, have overlapping attenuation coefficients depending on their concentrations. Calcium exhibits strong differential X-ray absorption at low and high kilovoltage settings. Using dual-energy calcium and water decomposition, calcium would be mapped on the calcium overlay image (Figures 5) and not on the virtual non-calcium image. On the other hand, materials like water would be mapped on the virtual non-calcium image and not on the calcium overlay image.^1,9

Figure 5.

85-year-old female with DM and HTN presented with acute right-sided weakness and sensory loss. Conventional NCCT of the head (a–c) demonstrates curvilinear hyperdensity (arrow) measuring approximately 50 HU in the left posterior pons, without significant surrounding edema, may represent calcification or hemorrhage. Foci of susceptibility blooming (arrowhead) extending from the left midbrain to the left pons (f, g) with minimal adjacent T2/FLAIR (d, e) abnormality may represent calcification with superimposed age indeterminate hemorrhage. Follow-up examination with dual-energy CT (h) demonstrates no significant interval change of the hyperdensity (arrow) of the posterior left midbrain and pons. Dual-energy CT with post-processing of the imaging data showed no hyperdensity on calcium overlay images (i), but was preserved (arrowhead) on bone removal images (j), suggesting that it does not have substantial calcium composition and therefore likely represent blood product associated with acute hemorrhage. DM, diabetes mellitus; HTN, hypertension; NCCT, non-contrast CT.

Differentiation of iodine and hemorrhage

On SECT, it is difficult to distinguish acute hemorrhage and lower concentrations of iodine when they co-exist. Iodine–water material decomposition helps to differentiate intracranial hemorrhage from iodinated contrast material. Iodine would be mapped on the iodine overlay image and not on VNC image. On the other hand, blood would be mapped on the VNC image but not on the iodine overlay image.^1,10

DECT may be used in the following applications:

1. intracerebral hemorrhage vs iodinated contrast material staining (Figure 6),

2. confirmation of spot sign—a quantitative marker for intracranial hematoma expansion¹¹ (Figure 7),

3. iodine vs hemorrhage in a suspected tumor^3,12 (Figure 8).

Figure 6.

68-year-old male, post-endovascular therapy for late window acute stroke. Dual-energy CT of the head (a) at the level of the superior lateral ventricle exhibits left MCA territory infarct with focal areas of hyperattenuation (arrow) of the left posterior parietal region. The hyperdensity could be secondary to contrast staining following angiography or hemorrhagic transformation of the stroke. Post-processing of the imaging data with an iodine overlay image (b) demonstrates that the area of hyperattenuation corresponds to an area of diffuse contrast material staining (arrow). No areas of hyperattenuation identified on the corresponding VNC image (c). Subtle patchy hyperdensities (arrowhead) along the cortical sulci of the infarcted brain parenchyma on images (a, c) with no contrast staining on the Iodine overlay map suggest interspersed subarachnoid hemorrhage. In the same patient, dual-energy CT images (d) and VNC images (f) at the level of the basal ganglia demonstrates hyperattenuation (arrow) along the left sylvian fissure. Iodine overlay image (e) demonstrates no corresponding region of contrast material staining, confirming intracerebral hemorrhage rather than iodinated contrast material staining. MCA, middle cerebral artery; VNC, virtual non-contrast.

Figure 7.

77-year-old female with right sided weakness. NCCT of the head (a) shows intraparenchymal hematoma of the left frontoparietal parasagittal region. Dual-energy CTA of the head (b) acquired at 80 and 140 kVp and an iodine overlay image (c) demonstrates a “spot sign” (arrow) within the hematoma compatible with active hemorrhage. Virtual non-contrast image (d) displays the acute intraparenchymal hematoma. NCCT, non-contrast CT.

Figure 8.

Dual-energy post-contrast CT scan of the head (a) demonstrates irregular peripherally enhancing mass of the right frontotemporal lobe (arrow) with contrast enhancement/tumoral hemorrhage (arrowhead), lateral to the mass lesion. (b) Iodine overlay image accurately maps iodinated contrast enhancement along the periphery of the lesion (arrow) but no contrast staining of the suspected hemorrhage identified, lateral to the mass lesion. Tumoral hemorrhage (arrowhead) is well-mapped out on the virtual unenhanced image (c).

Artifact reduction

Aneurysmal coils/clips

Beam hardening artifact is reduced at a higher keV (100 keV and above); however, at higher keV the enhancement within the vessel is reduced. The reconstructed VMI energy range between 100 and 120 keV helps achieve a balance by decreasing the metal artifact and improving the visualization of the vessels adjacent to the aneurysmal clip/coil (Figure 9).

Figure 9.

DECTA images in a 55-year-old female post-clipping of a right middle cerebral artery (MCA) aneurysm. Monoenergetic images were reformatted at 40, 60, 80, 100, 120, 140, 160 and 180 keV. Beam hardening artifact is reduced at higher keV (100 kev and above). However, at higher KeV, the enhancement of the vessel is reduced. DECTA, dual-energy CT angiogram; MCA, middle cerebral artery.

Miscellaneous

Extracranial artery dissection

A tear of the intimal layer of an arterial wall causes blood to enter the media resulting in the creation of a false lumen. Arterial dissection suspected on CTA can be confirmed with a T₁ fat-saturated MRI sequence, where intramural blood demonstrates T₁ hyperintense signal. The intramural hematoma can alternatively be identified on VNC images from DECT (Figure 10). However, MRI is superior in demonstrating intramural hematoma of dissection, compared to DECT.

Figure 10.

CT angiography of the carotids (a) displays irregular contour and reduced caliber of the extracranial left distal cervical ICA (arrow) suspicious for dissection. Post-processed DECT, VNC images (b, c) demonstrates curvilinear hyperdensity surrounding the ICA segment (arrow heads) suspected to represent intramural hematoma. This was confirmed by MRI on the axial T₁ fat saturated sequence (d). DECT, dual energy CT; ICA, internal carotid artery; VNC, virtual non-contrast.

Better delineation of small extra-axial hematoma

Bone subtracted images allows for better identification of subtle extra axial hemorrhage that may be obscured by the calvarium on conventional CT¹ (Figure 11).

Figure 11.

Patients (a, b) with traumatic brain injury. Bone subtracted images are helpful to demonstrate extent of extra-axial hematoma (arrow) adjacent to the calvarium and to detect subtle extra-axial hemorrhage (arrow), not well appreciated on the routine post-processed images.

Differentiating chronic subdural hematoma vs subdural hygromas

On VMIs, there is increased contrast between water and blood products at lower keV. This allows for confident differentiation between chronic subdural hematomas and hygromas (Figure 12). Clinically, this differentiation is important when determining the need for radiological follow-up.

Figure 12.

76-year-old male, post-evacuation of right convexity subdural hematoma. Conventional CT (a) at 120 kVp and Low energy virtual monoenergetic spectral CT at 50 keV (b) of the residual right parietal subdural hematoma were obtained. Low monoenergetic keV images (b) allows for improved visualization of residual hyperdensity (arrow) along the right parietal convexity helping to distinguish chronic subdural hematoma from hygromas. Subacute subdural hematoma is also demonstrated along the left cerebral convexity.

DECT clinical applications in spinal imaging

Assessment of intervertebral disks

The collagen content of the intervertebral disks permits improved visualization by using a collagen post-processing algorithm¹³ (Figure 13) to highlight collagen. It can be effectively employed as an alternative modality to MR imaging when MR imaging is not easily available or contraindicated.

Figure 13.

48-year-old female with a suspected L5-S1 disc protrusion. (a) Conventional CT, (b) dual energy CT with collagen post-processing algorithm and (C) dual energy CT with bone removal algorithm and color coding allows better visualization of the disc protrusion.

Vertebral compression fractures

Vertebral compression fractures can be difficult to diagnose on conventional CT, especially in patients with osteoporosis. DECT is a promising clinical tool for the diagnosis of vertebral compression fractures owing to its superior material differentiation using material decomposition algorithms.¹⁴ Edema detected with dual-energy CT can be illustrated as color coding (Figure 14). It can be used as an alternative modality to MR imaging when MR imaging is contraindicated or not easily accessible.

Figure 14.

47-year-old male with traumatic lumbar pain and acute compression fracture of L2 vertebral body. Conventional CT (a) and sagittal reconstructed DECT color-coded images b, c [magnified image] demonstrates marrow edema (arrow) of the L2 vertebral body.

Spine hardware artifacts

Diagnostic capabilities of CT are reduced in the presence of metallic hardware due to artifacts from beam hardening, photon starvation, and scattered radiation. Despite advances in detector technology and optimization of image reconstruction, artifacts from metal implants continue to be unresolved. DECT VMIs can be used to circumvent this problem without increased radiation exposure to the patient. Images at higher energies are better suited for limiting beam hardening artifact from spinal metallic hardware¹⁵ (Figure 15).

Figure 15.

Monoenergetic reconstructions from DECT of a 58-year-old female with remote L5-S1 fusion. Reconstructions from 40 to 180 keV and color-coded images demonstrate that increasing keV allows better visualization of the screws and improved delineation of the spinal canal anatomy due to decreased beam-hardening artifacts. DECT, dual-energy CT.

Conclusion

DECT is a problem-solving tool that offers a new dimension for approaching commonly encountered diagnostic challenges in head and spine imaging. This technique needs to be further explored in various applications to recognize its full potential.