Dual-energy computed tomography in children: Technique and clinical applications

Abstract

Dual-energy CT (DECT) is an advancement in CT technology that allows for the acquisition of images at two different energy levels. Two main post-processing tools, which form the backbone of DECT, include material decomposition and virtual monoenergetic images. Material decomposition helps in the generation of virtual nonenhanced, iodine, pulmonary lung blood volume, lung vessel, automated bone removal, and renal stone characterization images. DECT offers a broad spectrum of clinical applications in pediatric imaging, including vascular, neurological, thoracic, abdominal, skeletal, and oncologic assessments. Additionally, it contributes to shorter scan times and enables the use of lower contrast media volumes while maintaining diagnostic image quality. This technique provides unique qualitative and quantitative information about the composition of the tissue, allowing differentiation of materials, including iodinated contrast agents. The radiation dose of DECT is equivalent to or lower than that of a single-energy CT, adding to the advantages of DECT, especially in children who are more sensitive to the harmful effects of radiation. In this minireview we outlined the basic principles of the DECT technique and its post-processing techniques with emphasis on clinical applications in pediatric imaging.

Key Words: Children; Computed tomography; Dual-energy; Imaging; Pediatric

Core Tip: Dual-energy CT has a wide range of applications in vascular, brain, chest, abdomen, skeletal, and oncological imaging in children. Concerns have always been raised about the potential risks of ionizing radiation in the pediatric population from CT scans. Amongst the recent advancements in CT technology, dual-energy CT stands out because of its ability to provide enhanced diagnostic information, reduce radiation doses, and facilitate faster scans, making it a highly promising imaging investigation in children.

INTRODUCTION

Dual-energy CT (DECT) allows for the acquisition of images at two different energy levels. It is based on the fact that materials behave differently with respect to their attenuation at high-energy and low-energy spectra due to the photoelectric effect. Apart from generating two image datasets at two different energy levels, a blended contrast image dataset is also generated for interpretation. Studies have shown that the radiation dose of DECT is not significantly higher than that of single-energy CT (SECT)[1-5]. This is especially important in children, who are more sensitive to the hazardous effects of radiation. By using the least amount of radiation to achieve the principles of DECT, the radiation dose can be minimized to a great extent[6,7]. The mean CT dose index in children in the dual-energy technique approximates to 3.7 mGy while for SECT to about 4.4 mGy[2]. Further, an inherent limitation of conventionally used SECT is that structures having different tissue compositions, such as calcification, blood products, and iodine, can sometimes be indistinguishable because of the overlap in their respective attenuation.

Magnetic resonance imaging (MRI) is a radiation-free alternative to CT that can be considered especially in the evaluation of the head and neck, spine, pelvis, hepatobiliary system, and chest. Recently, unenhanced MRI has also been found in a study to be useful for the initial evaluation of abdominal emergencies in children. However, in head and neck emergencies, CT takes precedence over MRI. MRI is also compromised due to increased sensitivity to motion-related artifacts, poorer spatial resolution compared with that of CT, and limited compatibility with equipment used in the intensive care unit[8]. In addition the anatomy as well as pathologies of paranasal sinuses and temporal bones are better assessed on CT than MRI. Therefore, considering the potential benefits of DECT in children, we discussed its applications in the pediatric population in this article.

TECHNIQUES

Different vendors use different technologies for the generation of dual-energy image datasets[9,10]. These include the following.

Sequential acquisition

There is an acquisition of two different helical datasets, one followed by another in the same scanner at different energy spectra or subsequent rotations at alternating tube voltages. Its disadvantage is the delay between the acquisitions, making misregistration due to motion unavoidable.

Rapid kilo-volt switching

There is a single X-ray tube, which rapidly alternates between high-energy and low-energy kilo-voltage to generate two different sets of images. Its main disadvantages are the long acquisition time and the high dose[9].

Dual-layer detector

There are two layers of detectors, one is sensitive to low-energy photons and the other to high-energy photons. The top layer is yttrium-based for absorption of low-energy data, and the bottom layer is gadolinium-oxysulfide-based for high-energy data. The advantages are a lack of misregistration and cross-scatter, whereas decreased spectral separation is a drawback[10,11].

Twin beam dual-energy

This technique uses two filters, tin and gold, to split the 120 tube kilovoltage peak (kVp) X-ray beam into two different energy levels. Decreased spectral separation is a limitation here also[10,11].

Dual-source dual-energy

There are two different X-ray tubes in one gantry to generate X-ray photons at two different energy levels (mostly 80 kV and 140 kV) simultaneously (Figure 1). The advantage of this technique is the ability to choose voltage, current, and filters for each of the tubes independent of one another. The disadvantage is the need for correction of cross-scatter radiation[9].

Figure 1 Acquisition of raw data at high and low kV by two different tubes in the scanner. Reconstruction of images at high and low kV follows. Various applications of dual-energy CT, including the blended contrast-enhanced image, iodine-specific image, and virtual non-contrast (plain) image.

POST-PROCESSING TOOLS

Material decomposition

The material decomposition process is able to identify and differentiate materials based on their elemental composition and attenuation properties, allowing for the assessment of their concentration and distribution in body tissues. With the help of photoelectric and Compton scattering effects, it is able to generate material-specific maps. The two main types of material decomposition approaches that are available include two-material and three-material decomposition. The material chosen for decomposition in dual source systems depends on clinical application. For example in abdominal scans, three material decomposition is conducted using iodine, fat, and soft tissue. For chest scans decomposition is achieved using soft tissue, air, and iodine. For single-source systems like rapid kV switching and dual-layer scanner, a two-material decomposition algorithm is used. It typically uses iodine and water and generates iodine density and water density images[11-13].

Based on this decomposition process, mathematical algorithms are able to remove the iodine from scanned voxels, thereby generating iodine-less or virtual unenhanced images. This would help in reducing the overall radiation dose and study duration[11]. Similarly, iodine-only images will help in quantitative as well as qualitative assessment of iodine contrast distribution in body tissue. Iodine levels can be shown using either Hounsfield units (HU) or concentration (Figures 1 and 2)[13].

Figure 2 Different reconstructions in a dual-energy CT study. Axial sections of the upper abdomen at the level of the liver showed normal liver. A: Blended contrast-enhanced image; B: Iodine-specific image; C: Iodine overlay image with superimposed iodine map on gray–scale virtual non-contrast image; D: Virtual non-contrast image.

Virtual monoenergetic images

These images are virtually generated by extrapolating datasets acquired at low-energy and high-energy values to simulate images at a particular kiloelectron volt, which generally ranges between 40 keV to 200 keV. Images at low kiloelectron volt have improved contrast and high noise, whereas images at high kiloelectron volt have reduced noise and low contrast resolution (Figure 3)[11,12].

Figure 3 An 11-year-old female with generalized tonic-clonic seizure and post-ictal drowsiness. A: Axial section of contrast-enhanced CT head image at 70 kV showed better contrast enhancement of the wall in the ring-enhancing lesion with perilesional edema; B: There was only subtle contrast enhancement in the image obtained at 150 kV.

Virtual monoenergetic images (VMIs) at low energy values demonstrate higher contrast in the vessels even at lower doses of contrast (Figure 4) due to their energy level being closer to the K-edge value of iodine[9]. In children with suboptimal contrast enhancement due to technical factors, these images may help in visualizing the vessels without compromising the diagnostic quality. High-energy VMIs lack the low-energy photos that are responsible for the generation of artifacts in the form of streaks around metal implants. These images will be particularly helpful in pediatric patients who have spinal fusion implants or osteosynthetic material[13].

Figure 4 A 4-year-old male with trauma. Axial virtual monoenergetic images of CT angiography of the bilateral lower limb vessels showed popliteal arteries. A: 40 kV; B: 80 kV; C: 120 kV; D: 160 kV. There was increased conspicuity of the vessels at low kV as compared with high kV. This application allows for the generation of acceptable image quality even with a suboptimal dose of contrast.

PEDIATRIC DECT PROTOCOL OPTIMIZATION

The unique body composition, size, and physiological aspects of children warrant the need for optimization of DECT protocols. These differences can be utilized to tailor the protocols as follows.

Lower tube voltage

Children have relatively higher organ size, lower body fat, and higher tissue water content compared with adults. As a result of this, a lower kVp (70 kVp-80 kVp) can be used to reduce the radiation dose and boost the iodine contrast. Phantom studies have shown that lowering the tube voltage does not lead to a significant increase in noise to hamper the overall quality of images[14]. This trade-off between noise and iodine contrast can be tackled with the use of iterative techniques or artificial intelligence (AI) to increase overall image quality[15].

Automatic exposure control and filtering: Modern scanners automatically adjust tube current to patient size (CARE dose or equivalent), ensuring only as many photons as needed. Tin filtration on the high-energy source (as on Siemens dual-source scanners) “cuts out unnecessary low energy photons” and increases dose efficiency by sharpening the energy spectra. For example Tabari et al[4] reported 30%-40% lower size-specific dose estimation for chest and abdomen-pelvic using dual energy against matched single energy scans.

Virtual non-contrast imaging

The ability of DECT to derive virtual non-contrast (VNC) images from post-contrast scans allows the generation of true unenhanced phases. This can halve the number of scans needed. In oncologic studies using VNC images helped reduce total dose length product by a factor of 5-6 compared with earlier multiscan protocols[3].

Contrast volume reduction

Better detection of iodine uptake at lower energy voltage has the potential to lower the requirement of iodinated contrast agent for scanning. It is important particularly in imaging newborns and infants as they can tolerate only a lower volume of contrast due to their faster metabolic rate and developing renal excretory system. It is helpful for patients in whom it is difficult to establish the peripheral access and who are at risk of contrast-induced nephropathy. One of the recent studies conducted on pediatric and adult phantom models comparing SECT and DECT demonstrated a reduction in dose of iodinated contrast material ranging from 22%-52% at various VMIs while maintaining the contrast-to-noise ratio. A higher reduction in contrast dose, as high as 74%, was achieved using dual layer CT[16].

Weight-based contrast administration

Weight-based contrast administration protocols are to be followed. For newborns, neonates, and patients weighing < 10 kg, a smaller-gauge catheter, such as 24 G, is preferred with an injection rate of 0.5-1.0 mL/second. As with the increase in age and weight, a larger gauge canula with a higher injection rate is to be used (e.g., in patients > 30-40 kg, a 20-gauge canula can be used with 3.0-5.0 mL/s injection rate)[17].

In summary protocol adjustments for pediatric DECT include using low kVp, tin filters, automatic dose modulation, fewer phases (with VNC/iodine maps), iterative and AI reconstruction, and judicious contrast dosing. These measures collectively maintain or improve image quality while substantially lowering radiation exposure.

At our institution the Siemens SOMATOM Force scanner with dual-source dual-energy technique is used to acquire image datasets. The protocols used vary depending on the area of interest to be scanned as well as the specific indication. Patients are adequately centered as the field of view (FOV) of the DECT is limited (35 cm approximately). The contrast dose is generally kept at 1 mL per kilogram of the child’s weight. In most cases only a single post-contrast phase acquired in dual-energy mode suffices. In cases where there is a need for a second phase (in suspected contrast leak or oncology), the same is added. A low-energy (70 kVp or 80 kVp) and high-energy (140 kVp or 150 kVp) image and a blended image that combines the high-energy and low-energy data (50% from each data set usually) are generated at the CT console. Post-processing, like the creation of VNC, iodine-specific, iodine overlay, or VMIs, is done afterwards on the reporting workstation, depending on the clinical indication[11,12].

APPLICATIONS IN CHILDREN

Cardiovascular

Dual energy is preferred for assessment of congenital cardiac (tetralogy of Fallot, transposition of great arteries) and vascular anomalies (coarctation, arteritis, etc.), producing high-quality three-dimensional images aiding in surgical planning[18]. VMIs derived using dual energy have the potential of reducing the overall radiation dose and contrast volume[19].

A comparative study between single-energy and dual-energy cardiac CT in pediatric patients showed promising results for DECT. It achieved a 40% reduction in contrast volume while maintaining comparable image quality across both groups. The 45 keV setting emerged as the most representative VMI for dual-energy imaging. Notably, DECT also led to a 64% reduction in radiation dose (mean: 0.4 ± 0.2 mSv). Diagnostic accuracy improved from 74% with SECT to 81% with dual energy, highlighting its superior diagnostic performance[19].

The application of bone removal aids in better detection of abnormalities in the vasculature (Figure 5). Although SECT also offers a threshold-based bone removal tool, this is sometimes not accurate. This is because these tools rely on particular attenuation values to differentiate bone from soft tissue, and errors can occur that result in incomplete bone removal and false removal of vessel parts[18,20].

Figure 5 A 4-year-old male with trauma. Coronal images of CT angiography of bilateral lower limb vessels to show the automated bone removal application of dual energy CT. A: Coronal maximum-intensity projection image showed bilateral superficial femoral arteries overlapped by bones, along with soft tissue swelling along the right thigh; B: Application of the bone removal tool of dual-energy CT enabled better delineation of the arteries and confirmed the absence of any pseudoaneurysm.

Brain

In children presenting with altered sensorium with a history of both fever and prior trauma, intracranial hemorrhage and infectious etiology are among the main possibilities. However, the former indication warrants a non-contrast CT and the latter, contrast-enhanced CT. This leads to higher radiation exposure, which can be avoided by the VNC images generated by DECT (Figure 6). Iodine overlay maps can aid in the detection of subtly enhancing tumors, which may be the cause of hemorrhage[21,22]. Low-energy reconstructions can help in the detection of subtle lesions, edema, and ischemia due to increased soft tissue contrast as well as increasing contrast-to-noise ratios (CNR) on angiographic imaging.

Figure 6 A 7-month-old female with fever and abnormal body movements, suspected of having either a bleed or meningoencephalitis. A: Axial blended contrast enhanced image showed abnormal leptomeningeal enhancement; B: Image at 70 kV showed marked increased contrast with increased conspicuity of the abnormal enhancement; C: Image at 150 kV showed decreased noise but at the cost of poor contrast; D: Virtual non-contrast image ruled out any abnormal bleed; E: Iodine-specific image showed iodine uptake in the region of hyperdensity, adding to the confidence in the diagnosis of meningitis.

Compared with conventional images, low-energy VMIs offer superior image quality. In unenhanced scans performed with dual-layer CT, VMIs in the 45–75 keV range demonstrated significantly higher CNR and signal-to-noise ratio than standard images. Among children under 6 years of age, the grey-white matter CNR was markedly improved at 50 keV VMIs (2.20 ± 0.84) compared with conventional images (1.11 ± 0.70). Based on these findings, 50 keV VMIs are optimal for evaluating brain parenchyma[23]. Higher energy reconstructions can reduce image artifacts from aneurysm clips and coils, spinal hardware, and dental amalgam[24]. A bone removal tool makes the lesions more conspicuous, more so in the case of lesions located in close proximity to the bony surfaces. Detection of small epidural/subarachnoid hemorrhages is also improved by the bone removal application (Figure 7)[21,22].

Figure 7 A 4-year-old male with hemophilia A who presented with acute headache and neck rigidity. A: Axial mixed-enhanced CT head images showed subtle extra-axial high-density lesions (indicated by arrows) in the left frontal and parietal lobes, resulting in a slight midline shift to the opposite side; B: The virtual non-contrast image confirmed that the hyperdensity (arrows) represented an acute bleed and not abnormal enhancement; C: Bone removal tool of dual energy demonstrated the bleed more clearly (arrows); D: Iodine overlay image did not show any iodine uptake in the region of the bleed.

Chest

DECT has been found to have high sensitivity, specificity, and diagnostic accuracy in the detection of acute pulmonary thromboembolism[25,26]. In pulmonary thromboembolism the pulmonary blood volume (PBV) images allow for the detection of perfusion defects in the lung parenchyma, and the lung vessel images are used for picking up thrombi/emboli. Areas of infarcts show no or low iodine content compared with the surrounding normal lung on PBV images (images that are generated for visual assessment and quantification of iodine uptake on lung images), thereby enhancing the level of confidence in making a diagnosis. On lung vessel images, pulmonary arteries are color-coded blue, whereas thrombi appear as areas of red within the blue pulmonary arteries (Figure 8). It also enhances the visualization of arteriovenous malformation. Combined images can also be generated to show both the thrombi as well as perfusion defects[27].

Figure 8 Axial sections of CT pulmonary angiography in a 2-month-old female with infective endocarditis. A: Complete occlusion of the left pulmonary artery by a thrombus on blended contrast image; B: Complete occlusion of the left pulmonary artery by a thrombus on low kV (80 kV) image, showing increased contrast in the vasculature; C: Lung vessel image showed blue color-coded right pulmonary artery (white arrow) with thrombus being color-coded red in the left pulmonary artery (black arrow), suggesting diminished iodine content; D: Lung blood volume image showed diffusely reduced perfusion in the left lung as compared with the right lung.

By virtue of the generation of PBV images, it also has a potential role in congenital heart diseases with hypoplastic or absent pulmonary arteries[11]. The PBV images can help differentiate various parenchymal abnormalities, such as atelectasis, infarcts, pneumonias, and airway abnormalities. Pulmonary infarctions show homogeneous low iodine distribution on iodine or PBV images, whereas pneumonia demonstrates heterogeneously decreased or increased iodine distribution. Homogeneously increased iodine distribution occurs in atelectasis[28]. PBV can be used as a surrogate for pulmonary perfusion, especially in neonates and infants, in diagnosing conditions such as congenital lobar overinflation, hypoplastic lung, or bronchopulmonary dysplasia. In a recent study perfusion imaging conducted with DECT has been advocated as an alternative to single-photon emission computed V/Q scan in infants[29].

An important pitfall is the false perfusion defect due to the selection of the wrong maximum HU value for PBV images in younger children. It should be kept between -300 HU to -200 HU (Figure 9)[11].

Figure 9 Pitfall related to the incorrect CT threshold used for the assessment of lung blood volume. A: Iodine defects were seen when the maximum threshold was set at -600 Hounsfield units; B: Lung parenchyma showed normal perfusion when the threshold was changed to -300 Hounsfield units. A higher threshold was needed in children to compensate for the higher lung density.

The uses of virtual unenhanced images in the chest include the differentiation of calcifications, talc, and enhanced thoracic structures. DECT can also help in distinguishing benign lung lesions from malignant lesions by using iodine-specific images. In the pediatric population this may be useful in the confirmation of diagnosis of congenital malformations, like pulmonary airway malformation (Figure 10) or duplication cysts with proteinaceous debris, which may resemble a complex solid cystic mass at times and pose a diagnostic challenge. Demonstration of a non-iodine-containing lesion favors a complex cyst or benign solid mass while demonstration of iodine within a mediastinal mass raises the suspicion for malignancy[30]. In a study by Lee et al[31], significant differences in iodine concentrations between benign and malignant tumors were found, while there was no significant difference in attenuation values. The best cutoff iodine value for differentiating benign from malignant tumors was found to be 1.40 mg/mL for the early phase DECT and 1.58 mg/mL for the delayed phase[31]. In addition to this DECT is also useful for treatment planning and assessment of response to treatment in chest oncology[30].

Figure 10  A case of congenital pulmonary airway malformation in a 5-day-old male. A: Axial lung window image showed multiple variable-sized cystic lesions interspersed in the right lung; B: Perfused blood volume image showed perfusion deficit in the affected areas; C: Lung vessel image superimposed on lung volume image showed no abnormal vasculature within or surrounding the lesion.

Abdomen

Liver: Small hepatic lesions in children with primary malignancy elsewhere can be benign granulomas or cysts and may be confused with hypovascular metastases on a SECT. An iodine map can show non-enhancement in the benign lesions, thus resolving the dilemma[17,32].

Bowel: Diseases like Crohn’s disease have an active phase and a fibrotic phase that need to be differentiated to decide the treatment[11]. By measuring the iodine density and keeping a cutoff value of 2.5 mg/mL, DECT is advantageous over conventional CT by pinpointing regions of active inflammation within the affected bowel. DECT with an iodine overlay imaging can confirm or exclude the presence of gangrenous change that is visualized as an area of markedly decreased enhancement or reduced iodine uptake (Figure 11)[33,34]. In a small cohort of 21 infants, DECT demonstrated 100% sensitivity and specificity in detecting bowel ischemia associated with necrotizing enterocolitis with surgical findings serving as the reference standard[35].

Figure 11  A 10-year-old female with acute pain in the right iliac fossa. A: Blended dual-energy contrast-enhanced coronal image showed circumferential homogenously enhancing mural thickening involving the wall of the appendix (arrow), which was dilated and fluid-filled, suggestive of acute appendicitis; B: There was also the presence of an oval-shaped hyperdensity (arrow) within its lumen on virtual non-contrast image, suggestive of an appendicolith; C: The iodine overlay map showed high iodine uptake in its wall (arrow), representing the hyperemia associated with active inflammation.

Genitourinary system

Due to its material decomposition ability, DECT is able to differentiate uric acid from non-uric acid stones (calcium oxalate, calcium phosphate, struvite, cysteine, and hydroxyapatite) (Figure 12). Low-energy and high-energy attenuation ratios are < 1.1 for uric acid, 1.1–1.24 for cystine, and > 1.24 for calcified stones[36]. This has implications in the management as the uric acid calculi are managed medically using alkalizing agents, whereas non-uric acid stones are treated either by lithotripsy or surgery. DECT parameters such as attenuation value at different spectral energies and effective atomic number are correlated with outcomes of lithotripsy[37].

Figure 12  Dual energy scan in nephrolithiasis. A: Non-contrast CT section of the abdomen showed right renal calculus in the pelvis; B: The calculus was color coded as blue (non-uric acid); C: Attenuation values of calculi on the right side at 70 kV, 150 kV, and blended images; D: The corresponding positions of the attenuation values on the graph indicated non-uric acid composition of the calculi.

Assessment of congenital anomalies of the urinary tract, tumors, and blunt trauma to the abdomen with suspected renal injury often requires multiphasic examination, including acquisition of non-contrast and delayed excretory phase. Addition of these phases for complete evaluation leads to increased radiation dose and overall increase in contrast volume administration. DECT, by virtue of generating virtual unenhanced images, can omit the need for a non-contrast phase. In a randomized controlled trial, Chen et al[38] demonstrated that DECT urography with a split-bolus technique significantly reduced radiation dose by 48%-56% in pediatric patients. This study also showed superior corticomedullary phase image quality based on improved CNR and signal-to-noise ratio values[38].

Bones

Corrective surgeries for scoliosis or developmental dysplasia of the hip joint in children involve the insertion of metallic implants, which can lead to a lot of streak artifacts. These streak artifacts hinder the visualization of surrounding soft tissues. VMIs at high kiloelectron volt (110 keV-130 keV), in addition to the iterative metal artifact reduction algorithms, can reduce these streak artifacts (Figure 13)[39].

Figure 13  A 2-year-old female who underwent CT of the bilateral hip joints after salter osteotomy for developmental dysplasia of the hip to look for the status of the femur head and acetabulum. A: Axial bone window image at 70 kV showed marked streak artifacts due to the K-wire inserted along the left iliac bone; B: Axial bone window image at 150 kV showed almost no streak artifacts.

Oncology

Hyperdense masses in organs like the liver, kidney, or brain can often pose a challenge given their resemblance to a hematoma. VNC images can easily demonstrate the baseline attenuation of masses (Figure 14), including the presence of any underlying hemorrhage or calcification, which can then be compared with attenuation on contrast-enhanced scans. Another application is the use of iodine-specific images for the assessment of treatment response. The viable tumors show higher iodine uptake compared with post-treatment non-viable or fibrotic lesions[40,41]. DECT iodine overlay images can be a key imaging parameter to assess the response to chemotherapy as newer drugs reduce the blood supply to the tumor. These drugs do not significantly change the size; however, a change in attenuation or iodine uptake is noticeable. These changes are readily picked up by iodine overlay images. Literature in this direction is still developing, especially in pediatric patients[30].

Figure 14  An 11-month-old male with hepatoblastoma. A: Blended contrast-enhanced axial image showed a large isodense mass in the liver with poor demarcation of the mass with the adjacent liver due to minimal enhancement; B: Virtual non-contrast image showed no evidence of calcification/fat/hemorrhage within the lesion; C and D: Iodine overlay images showed iodine uptake in the periphery of the mass with attenuation on the virtual non-contrast image being 475 Hounsfield units and on the blended image being 903 Hounsfield units, suggesting significant enhancement that was not well appreciated on the blended image; E: Iodine-specific image showed the iodine content in the periphery of the mass to a better advantage.

PITFALLS

Clinically relevant artifacts that interfere with the interpretation of dual-energy images are beam hardening and photon starvation. Beam hardening, caused by selective attenuation of low-energy photons, produces streaking and cupping near metal implants while photon starvation results in complete attenuation of all photons by the foreign body, causing zero transmission projections[13,10]. Beam hardening and photon starvation occur more at low kilo-electron volt (e.g., 80 keV) energies, which degrades the image quality[42]. On color-coded maps these streak artifacts may render the images non-diagnostic.

Appropriate window/Level settings are a must for the demonstration of correct color-coded iodine maps[42,43]. In iodine quantification, there are no well-validated thresholds of iodine concentration to define true enhancement[42,43]. Sometimes, VNC images may show some residual iodine in vessels or renal collecting systems that could not be subtracted. This may mimic pathology and confound interpretation[42,43]. Calcium remains bright on VNC images as well as color-coded iodine maps. Therefore, color-coded maps must be viewed in conjunction with VNC images so as not to confuse with iodine[42,43].

Different dual-energy systems available on the market have their specific limitations that are clinically relevant. A dual-source system has an inherent difference in the FOV of the two X-ray tubes with a smaller FOV available with higher tube current and a larger FOV available with lower tube current. This incongruity in FOV limits the use of dual-source CT in the evaluation of larger body size and cases that require increased anatomical coverage. This difference can result in non-processing of tissues falling between smaller and larger FOV. Due to this, appropriate centering in the gantry during the acquisition is of paramount importance[13]. Another limitation that comes with dual-source systems is with echocardiogram gating. For faster, relatively motionless acquisition in heart and dynamic lung images, the application of dual energy mode is traded off with super quick acquisition due to higher pitch[44]. Similarly, in rapid kV switching systems, there is low photon output at low voltage, compromising image quality due to higher noise. To overcome this higher tube current is required, therefore increasing the total radiation dose. This can also be overcome with the use of deep learning models and iterative reconstruction algorithms[13].

Owing to the unique body composition and continuous growth of children, there is a requirement for age and weight-based protocol optimization. Children have relatively higher organ size, lower body fat, and higher tissue water content compared with adults. As a result of this, a lower tube voltage (70-80 kVp) can be used to reduce the radiation dose and boost the iodine contrast[14]. Lower tube voltage results in an increase in noise of diagnostic images. AI driven algorithms can be used for denoising; however, there is currently limited resource availability in this regard, especially in pediatric patients.

Children are more vulnerable to the effects of ionizing radiation during CT examinations. This increased sensitivity is due to their higher rate of mitosis and longer life expectancy[45]. Given the continuous changes in body composition during growth, there is a critical need to develop age-specific and body composition-specific imaging protocols to minimize unnecessary radiation exposure in line with the application of as low as reasonably possible. Moreover, there is limited validation of DECT protocols in neonates and pediatric populations who are critically ill in which unique challenges such as altered image quality and increased motion artifacts may impact diagnostic performance.

There is a long learning curve associated with the interpretation of iodine maps and the generation of VMIs from spectral images that are generated for the radiologists.

FUTURE ASPECTS

Deep learning models can be employed for the optimization of acquisition protocols, and further reduction of radiation dose can be achieved by deep learning image reconstruction. In a recent systematic review on the role of AI in the optimization of the radiation dose, a 36%-70% reduction in radiation dose was demonstrated without sacrificing image quality. The major leap in this direction can be achieved by training these models to develop the algorithms producing diagnostic quality images with ultra-low dose protocol using spectral imaging[15]. Another study showed that VMIs when combined with deep learning models for reducing the noise could offer a 19.6% reduction in radiation dose with maintenance of diagnostic quality images[46].

CONCLUSION

DECT has expanded the horizons of the currently available applications of SECT in the pediatric population. There is no significant increase in radiation exposure, making it safe for use in children. There are certain pitfalls that need to be kept in mind to derive the maximum benefit from it. Further, DECT has more common uses in adults that may not translate to actual clinical practice in pediatric imaging.