Head-to-Head Comparison of Dual-Source and Split-Beam Filter Multi-Energy CT versus SPECT/CT for Assessing Lobar Lung Perfusion in Emphysema

Abstract

Purpose

To evaluate dual-source and split-beam filter multi-energy chest CT in assessing pulmonary perfusion on a lobar level in patients with lung emphysema, using perfusion SPECT as the reference standard.

Materials and Methods

Patients with emphysema evaluated for lung volume reduction therapy between May 2016 and February 2021 were retrospectively included. All patients underwent SPECT and either dual-source or split-beam filter (SBF) multi-energy CT. To calculate the fractional lobar lung perfusion (FLLP), SPECT acquisitions were co-registered with chest CT scans (hereafter, SPECT/CT) and semi-manually segmented. For multi-energy CT scans, lung lobes were automatically segmented using a U-Net model. Segmentations were manually verified. The FLLP was derived from iodine maps computed from the multi-energy data. Statistical analysis included Pearson and intraclass correlation coefficients and Bland-Altman analysis.

Results

Fifty-nine patients (30 male, 29 female; 31 underwent dual-source CT, 28 underwent SBF CT; mean age for all patients, 67 years ± 8 [SD]) were included. Both multi-energy methods significantly correlated with the SPECT/CT acquisitions for all individual lobes (P < .001). Pearson correlation concerning all lobes combined was significantly better for dual-source (r = 0.88) than for SBF multi-energy CT (r = 0.78; P = .006). On the level of single lobes, Pearson correlation coefficient differed for the right upper lobe only (dual-source CT, r = 0.88; SBF CT, r = 0.58; P = .008).

Conclusion

Dual-source and SBF multi-energy CT accurately assessed lung perfusion on a lobar level in patients with emphysema compared with SPECT/CT. The overall correlation was higher for dual-source multi-energy CT.

Keywords: Chronic Obstructive Pulmonary Disease, Comparative Studies, Computer Applications, CT Spectral Imaging, Image Postprocessing, Lung, Pulmonary Perfusion

© RSNA, 2023

Summary

Dual-source and split-beam filter multi-energy CT accurately assessed lobar lung perfusion in patients with lung emphysema when compared with perfusion SPECT as the reference standard.

Key Points

• ■ Dual-source and split-beam filter multi-energy CT accurately assessed lung perfusion on a lobar level compared with perfusion SPECT (Pearson r = 0.88 and 0.78, respectively), thus providing both structural and functional information.

• ■ Regarding all lobes combined, the correlation with perfusion SPECT was higher for dual-source CT versus split-beam filter CT (P = .006).

• ■ While lung function parameters showed no observable effect, the individual lung lobes significantly influenced the disparities between SPECT and multi-energy CT perfusion measurements.

Introduction

Conservative therapy for hyperinflation in chronic obstructive pulmonary disease is based on two main pillars: nonpharmacologic intervention (eg, smoking cessation, respiratory rehabilitation) and pharmacologic treatment, including bronchodilators (eg, anticholinergics, β-2-sympathomimetics) and anti-inflammatory agents (eg, corticosteroids, phosphodiesterase-4-inhibitors, antibiotics) (1). However, these options are of limited benefit to patients with advanced emphysema.

Besides lung transplantation, lung volume reduction therapy (LVRT) is a promising therapeutic option for well-selected patients. It has been shown to improve survival, lung function, and disease burden (2–5). LVRT can be performed surgically and endoscopically. Endoscopic lung volume reduction is mainly performed via bronchoscopic insertion of valves or, less commonly, endobronchial coils, stents, steam, or hydrogel foam. Unlike the surgical approach, endobronchial valve implantation removes the entire lobe from gas exchange by inducing lobar atelectasis. Consequently, precise pre-interventional planning of the target volume is crucial.

Factors like lung perfusion and emphysema distribution affect the clinical outcome. Therefore, optimal patient selection is paramount, and patients undergoing assessment for LVRT require both structural CT and pulmonary perfusion evaluation. The latter is usually determined with scintigraphy or SPECT/CT (6–10).

A newer method, iodine mapping derived from multi-energy CT, has also been shown to reliably assess lung perfusion compared with planar scintigraphy (11) and SPECT/CT (12,13). Technical implementations in current commercial CT systems include dual-source multi-energy CT, where two x-ray source and detector systems work at different voltages, and split-beam filter (SBF) multi-energy CT, where a two-material tin and gold filter splits the single-source x-ray beam into high- and low-energy spectra. Comparisons between these techniques, especially concerning clinical applications, are scarce.

With lung perfusion derived from multi-energy data more readily available, the necessary information for LVRT could be obtained from a single examination, saving time and money. Moreover, patient radiation exposure could be reduced because single- and multi-energy CT scans are comparable in the applied dose, and SPECT/CT could become dispensable (14).

To our knowledge, a prior study has not been conducted to quantitatively compare dual-source and SBF CT with SPECT/CT in assessing lung perfusion on a lobar level and focused primarily on patients with emphysema. We closed this gap with the present study by first comparing lobar perfusion assessed with SPECT/CT versus multi-energy CT for two established techniques (dual-source and SBF CT) and, second, by comparing the accuracy of these two multi-energy techniques.

Materials and Methods

The institutional ethics board approved this retrospective study (approval number: 21–2457–104). The need for written informed consent was waived because the study was retrospective and no additional imaging or patient testing was needed. All procedures involving patient data followed the ethical standards of the institutional and national research committees and the Declaration of Helsinki. No industry support was received for the study. Only the authors had control of the data and information submitted for publication.

Patients

We included consecutive patients who underwent both multi-energy chest CT and SPECT between May 2016 and February 2021. Patients were referred for evaluation for endoscopic LVRT or suspected pulmonary artery embolism. Patients in the latter group were included only if they also had emphysema. In the case of more than one CT or SPECT study per patient, studies chronologically closer to each other were selected. Patients for whom the subsequent analysis was erroneous (eg, unsuccessful segmentation of the lung lobes) were excluded. Only pre-intervention scans were considered.

Imaging

Dual-source CT.— Contrast-enhanced chest CT was performed with the patient in the supine position during an end-inspiration breath hold using a second-generation dual-source CT scanner (Somatom Definition Flash; Siemens Healthineers). Iohexol (Accupaque 350; GE Healthcare Buchler) was used as an intravenous contrast medium and administered at a flow rate of 3–5 mL/sec and at a dose of 1.5 mL per kilogram of body weight. The scan was triggered using a bolus-tracking technique with a delay of 5 seconds after a region of interest in the main pulmonary artery reached the threshold of 100 HU. Scans were performed with the vendor's default settings: 100 kVp (tube A) and 140 kVp with a tin filter (tube B); quality reference tube current–time product was 150 mAs (tube A) and 128 mAs (tube B), respectively, using the combined angular and longitudinal automatic tube current modulation technique (CARE Dose 4D; Siemens); the pitch factor was 0.55.

SBF CT.— Contrast-enhanced SBF CT was performed with the patient in the supine position during an end-inspiration breath hold using a Somatom Definition AS+ scanner (Siemens) with the same parameters concerning contrast agent application as described for dual-source CT. Scans were performed with the vendor's default settings. The x-ray beam (120 kV) was split using filters made of tin (high energy) and gold (low energy). Quality reference tube current–time product was 352 mAs, and the pitch factor was 0.25.

SPECT.— For lung perfusion imaging, patients underwent slow intravenous injection of 156 MBq ± 22 (SD) technetium-99m (^99mTc)-labeled macroaggregated albumin (^99mTc HSA-B20; ROTOP; or ^99mTc Pulmonic; Curium) in the supine position. SPECT of the lungs was performed using a double-head gamma camera (Symbian Intevo Bold, Siemens; AnyScan SC, Mediso; Discovery NM 630, GE Healthcare) equipped with low-energy high-resolution collimators.

SPECT was performed with 64 projections (12 seconds per projection) using a 128 × 128 matrix (15). Images acquired using the Symbia Intevo Bold scanner were reconstructed using an iterative algorithm (Flash 3D; four iterations and eight subsets) and a three-dimensional (3D) Gaussian postreconstruction filter with 8.0-mm full-width at half-maximum without attenuation correction. Images acquired using the AnyScan SC or Discovery NM 630 scanners were reconstructed using HERMES HybridRecon 3.2.1 (HERMES Medical Solutions) with an ordered subset expectation maximization algorithm (five iterations, 16 subsets) and a 3D Gaussian postreconstruction filter with 9.0-mm full-width at half-maximum collimator and scatter corrections. Patients were instructed to breathe shallowly during the acquisition.

Image Postprocessing

Multi-energy CT.— The same processing steps were performed for all multi-energy CT scans. Multiplanar reconstructions for anatomic and structural analysis were generated from weighted average images (60% from high-energy data, 40% from low-energy data) applying lung and soft-tissue kernels. Iodine perfusion maps were calculated from the multi-energy data using dedicated postprocessing software (syngo.via CT Dual Energy, vVB30A; Siemens). The preset parameters for pulmonary perfused blood volume tissue recognition were a minimum attenuation of -960 HU and a maximum attenuation of -600 HU (default settings for evaluating normally aerated lung). The calculated perfusion maps and corresponding structural chest CT scans were reconstructed with a section thickness of 1.0 mm. All data were exported in Digital Imaging and Communications in Medicine format and pseudonymized using syngo.via.

Data were then transferred to an on-site Linux Ubuntu (version 20.04; Canonical Foundation) machine with an EPYC 7742 64-Core Processor (AMD) and four A100-SXM4 40-GB graphics processing units (NVIDIA) where all further processing was done. A custom routine was implemented using Python (version 3.8.12; Python Software Foundation). All code for data processing is available at github.com/qstro.

Structural scans and perfusion maps were converted to Neuroimaging Informatics Technology Initiative file format using SimpleITK (16). Lung lobes (right upper, right middle, right lower, left upper, left lower) were automatically segmented from the structural scans using the lungmask Python library and a pretrained U-shape 3D convolutional neural network available at github.com/JoHof/lungmask (17). All segmentations were visually verified and manually corrected where needed using ITK-SNAP (itksnap.org) (18). These segmentations were used as masks, and the perfusion per lobe was calculated as a fraction of the whole-lung perfusion (fractional lobar lung perfusion [FLLP]). Figure 1 shows exemplar images of the multi-energy CT data.

Figure 1:

Split-beam filter CT images in a 59-year-old man show three-dimensional frontal view chest image of the lobar segmentation after automated segmentation and manual optimization (left), axial section of a structural scan with applied lung window settings calculated from the multi-energy data set with superimposed two-dimensional lobar segmentation mask (middle), and multi-energy–derived iodine map (right). Red plane indicates z-axis position of the axial sections.

SPECT/CT.— SPECT perfusion images were co-registered with structural chest CT scans (hereafter referred to as SPECT/CT) with an automated rigid transformation and quantified with Conformité Européenne–marked quantification software (Hybrid 3D 3.0.1–Lung Lobe Quantification; HERMES Medical Solutions). Co-registrations were quality checked by board-certified radiologists and nuclear medicine physicians (S.H., J.G., O.W.H.; 9, 16, and 23 years of experience, respectively) and manually corrected, if necessary. The right and left lung volumes were calculated from the CT image using an automatic segmentation algorithm with attenuation thresholds according to the manufacturer's specifications to discriminate between the lung and soft tissue. The lung volumes were split into lobar partial volumes by marking seven interpolation points in one section for each fissure and checking the correct automatic delineation. The resulting lobar regions of interest were transferred to the SPECT images to measure the amount of retained tracer and to automatically calculate the relative perfusion distribution for each lobar region as a fraction of 1.

Statistical Analysis

Tests were performed using the SciPy (version 1.7.1) (19), statsmodels (version 0.13.0) (20), and pingouin (version 0.5.2) (21) Python libraries. Clinical data were compared with descriptive statistics, t tests for continuous data, and χ² tests for categorical data. The Levene test was used to assess the equality of variances. Agreement of FLLPs between dual-source CT and SPECT/CT and between SBF CT and SPECT/CT was evaluated with Bland-Altman analysis, linear regression, and Pearson and intraclass correlation coefficients (ICCs [2,1]; two-way random effects model, absolute agreement, single measure) (22). ICCs were calculated because the Pearson correlation coefficient indicates only the linear relationship between two data sets, not their absolute agreement. The Pearson correlation coefficients between dual-source CT and SBF CT were statistically compared after applying the Fisher z transformation and considering the different sample sizes. We used a linear mixed model to compare the mean difference of FLLPs. The main effects were lung lobe and multi-energy system. The interaction effect was lobe × system. Patients were added as random factors to account for multiple measurements within one patient. We report unadjusted and Bonferroni-adjusted P values of the pairwise post hoc comparisons. We further tested confounding clinical variables with multivariable regression. The dependent variable was the difference between SPECT and multi-energy CT FLLPs. Independent variables were multi-energy system, lung lobe, and pulmonary function test parameters. For all analyses, two-tailed tests were used. Results with a type I error < .05 were considered significant.

Results

Patient Characteristics

Fifty-nine patients met the inclusion criteria (30 male, 29 female; mean age for all patients, 67 years ± 8 [mean age for male patients, 67 years ± 8; mean age for female patients, 66 years ± 8]). Table 1 shows detailed characteristics of study patient groups, separated by the multi-energy technique. The forced expiratory volume per second (P = .004), vital capacity (P = .02), and residual volume (P = .008) of the pulmonary function tests significantly differed between the multi-energy groups, with better results for the dual-source group. We found no evidence of differences in any other clinical variables between the two groups.

Table 1:

Patient Characteristics

All patients had emphysematous lung disease (eg, chronic obstructive pulmonary disease). In addition, some patients had pulmonary vascular disease (eg, pulmonary hypertension, pulmonary embolism, chronic thromboembolic pulmonary hypertension).

Automated segmentation of the multi-energy CT scans was possible in all included patients. In 11 of 59 patients, manual correction was needed, mainly because of necessary extrapolation when fissures were incomplete. Pearson correlation coefficients of lobar volumes between the multi-energy CT and SPECT/CT segmentations were 0.97 for the dual-source group and 0.96 for the SBF CT group.

Quantification of Pulmonary Perfusion

Lobar contribution.— The middle lobe contributed the least to the total lung perfusion in both multi-energy CT groups. When lung perfusion was normalized by the lobar volume (FLLP divided by the fraction of volume of the respective lobe in relation to the volume of the entire lung), this effect was neutralized entirely (Table 2), suggesting that no lung lobe contributed significantly to lung perfusion in relation to its volume.

Table 2:

Mean Lobar Perfusion

Multi-energy CT versus SPECT/CT: results for all lobes combined.— For both multi-energy methods, evaluation of lung perfusion correlated significantly with SPECT/CT (Pearson r = 0.88 for dual-source CT, Pearson r = 0.78 for SBF CT; P < .001 for both) (Fig 2), with the dual-source technique demonstrating significantly better results compared with the split-beam technique (P = .006). The ICC (2,1) was 0.87 (95% CI: 0.83, 0.90) for dual-source and 0.78 (95% CI: 0.71, 0.84) for SBF CT.

Figure 2:

Scatterplots show perfusion results for all lobes combined (lobes are equal to five points per patient) for the fractional lobar lung perfusion per multi-energy CT and SPECT/CT scans. (A) SPECT/CT versus dual-source CT. (B) SPECT/CT versus split-beam filter CT. Left upper corner indicates Pearson correlation coefficient r and P values. Regression line and 95% CI are shown.

Multi-energy CT versus SPECT/CT: results for single lobes.— Both multi-energy methods showed high correlations with SPECT/CT when comparing FLLPs. However, the results for the dual-source scanner were better in some instances. Pearson correlation between dual-source multi-energy CT and SPECT/CT was significantly higher than between SBF CT and SPECT/CT in the right upper lobe (r = 0.88 vs r = 0.58, P = .008). For all other lobes, we found no evidence of differences. For SBF CT, the left upper lobe showed the highest correlation (r = 0.82), but this was not significantly better than the dual-source scanner (r = 0.71, P = .32). Figure 3 and Table 3 show detailed results.

Figure 3:

Scatterplots show perfusion results for each lobe. The fractional contribution to the total lung perfusion for the multi-energy CT and SPECT/CT scans are separated by pulmonary lobe: (A) right upper lobe, (B) right middle lobe, (C) right lower lobe, (D) left upper lobe, and (E) left lower lobe. Regression line and 95% CI are shown. Bland-Altman plots (including 95% CIs as translucent bands) for the same data are shown to compare multi-energy CT with SPECT/CT acquisitions. The agreement between the two methods is represented by assigning the mean of the two measurements (SPECT/CT, multi-energy CT) as the x-axis value and the difference between the two values (SPECT/CT minus multi-energy CT) as the y-axis value. Left two columns: dual-source CT. Right two columns: split-beam filter CT. The x- and y-axis ranges are fixed to allow for easier comparison between lobes.

Table 3:

Statistical Comparison of Lobar Lung Perfusion

Multivariable regression was applied to rule out possible confounding variables. We selected forced expiratory volume per second, residual volume, and vital capacity of the pulmonary function tests because these parameters differed significantly between the dual-source and SBF CT collectives. Forced expiratory volume per second (P > .99), vital capacity (P = .97), residual volume (P = .98), and multi-energy system (P = .9) had no impact, whereas a significant influence on the difference between SPECT/CT and multi-energy CT perfusion measurements was observed for the different lung lobes (P < .001).

Radiation Dose

The effective dose for the multi-energy CT examinations was calculated by multiplying the dose-length product by the conversion factor 0.0146 mSv × mGy⁻¹ × cm⁻¹ (23). Radiation exposure from ^99mTc macroaggregated albumin was estimated to be 0.011 mSv/MBq. The mean effective dose for the multi-energy scans, including localizer and bolus tracking, was 5.50 mSv ± 2.15 for dual-source CT and 3.36 mSv ± 1.06 for SBF CT. The volume CT dose index was 8.96 mGy for the dual-source group and 6.54 mGy for the SBF CT group. The mean effective dose for the SPECT scans (without CT) was 1.7 mSv ± 0.24.

Discussion

Information about lobar lung perfusion is essential for planning LVRT in patients with emphysema. Multi-energy CT could enable a “one-stop shop” approach to acquiring functional and anatomic data in one session. To our knowledge, our study is the first to compare lung perfusion on a lobar level assessed with dual-source versus SBF CT, with SPECT/CT as the reference standard. Regarding the included number of patients, our study is among the largest to compare multi-energy CT and SPECT/CT quantitatively concerning lobar lung perfusion. Our results support the following conclusions: (a) dual-source and SBF CT reliably assess lung perfusion on a lobar level in patients with lung emphysema, as shown by Pearson correlations of 0.88 and 0.78 and ICCs of 0.87 and 0.78, respectively; and (b) perfusion measurements are more accurate for the dual-source technique.

To optimize patient selection during pre-interventional assessment for LVRT, thin-section chest CT and perfusion SPECT must be performed for anatomic evaluation and functional evaluation of the lung parenchyma, respectively (6,9).

Diagnostic thin-section CT is essential for therapy planning because the radiologist must quantify lung emphysema and assess its distribution, including the extent of heterogeneity. Furthermore, fissure status, bronchiectasis, small-airway disease, pneumonia, pleural effusion, previous surgery, and signs of pulmonary hypertension or pleuroparenchymal illness must be evaluated. Exclusion criteria, such as foci suspicious for malignancy, must be ruled out (10).

Perfusion analysis is also mandatory because removing lung parenchyma that is still well perfused and thus obviously involved in the gas exchange must be avoided (7,8). Perfusion scintigraphy or perfusion SPECT have been the standard for this purpose (10,24). However, these methods distinguish only among the three lung zones (upper, middle, and lower) and not among lung lobes. Only SPECT/CT qualifies for assessment of lobar lung perfusion; however, evaluation is tedious and, so far, not part of the clinical routine. Endoscopic valve implantation removes the entire lobe from gas exchange. Thus, pre-interventional information about the contribution of each lobe to lung perfusion as a surrogate for gas exchange is of utmost importance. Multi-energy CT imaging might be able to provide both structural and functional imaging (25). This technique is already used in acute and chronic pulmonary embolism workup. In this setting, however, mere subjective evaluation of the distribution of lung perfusion is performed without acknowledging lobar anatomy.

We show that multi-energy CT can provide all necessary data to streamline the evaluation of patients for LVRT, including quantification of lung perfusion on a lobar level. The calculation of lobar perfusion is conveniently done automatically without major intervention by using freely available open-source software libraries (the script is available at github.com/qstro; note that it is not approved as a medical product and is provided only for research purposes).

Literature regarding the quantitative comparison of multi-energy CT and SPECT/CT on a lobar level is scarce. The study by Si-Mohamed et al (13) focused mainly on patients with pulmonary hypertension. The study by Lapointe et al (26) included only five patients with lung cancer, an order of magnitude smaller than our study sample. Similar to our study, Jeyin et al (27) focused on patients with emphysema, showing a correlation between dual-source CT and SPECT/CT between 0.78 and 0.84. Unfortunately, in all these studies, only Pearson correlation coefficients and Bland-Altman plots were calculated. Pearson correlation measures only linear relationships, not absolute agreement. We calculated the ICC (2,1) to address this limitation (22).

The accuracy of dual-source data was better than that of SBF CT (all lung lobes combined). This difference is thought to have been caused by the different techniques: Dual-source CT uses two x-ray source and detector systems at different voltages. In SBF CT, a two-material filter (eg, made from tin and gold) splits the single-source x-ray beam into high- and low-energy spectra. Spectral separation is worse for the split-beam technique, leading to a lower contrast-to-noise ratio compared with dual-source CT (28,29).

Almeida et al (29) found that dual-source scanners measure electron densities (relative to water) and effective atomic numbers closer to the phantom reference than split-beam scanners. Another study comparing virtual unenhanced attenuation values derived from multi-energy CT showed that SBF CT attenuation had a higher variation between consecutive studies of the same patients compared with dual-source CT; in an interscanner comparison, this study also showed that the attenuation differed significantly between dual-source and split-beam scanners, suggesting that images from these scanners cannot be reliably compared (30). Petritsch et al (31) compared the two techniques for diagnosing acute pulmonary artery embolism and reported lower iodine distribution map quality and higher radiation dose for the split-beam acquisitions. This finding is supported by studies investigating multi-energy CT in head and neck imaging (32).

One could argue that the differences in effective dose might have aggravated this tendency. It was shown that CT number is overestimated by multi-energy CT at very low dose levels (33). However, no bias was observed when comparing the split-beam data with SPECT/CT and between the dual-source and SBF CT results. Furthermore, a potential overestimation of perfusion in low-dose acquisitions would not have influenced the FLLP because they are relative measures. Thus, it is unlikely that differences in dose levels biased our results.

Our study had some limitations. First, nonenhanced chest CT is recommended to assess emphysema burden and its distribution because contrast material increases lung density (6,10). However, intravenous contrast agent administration facilitates the evaluation of LVRT eligibility criteria, like mediastinal and hilar lymph node status, and is mandatory for multi-energy lung perfusion assessment. This gap might be closed by calculating virtual unenhanced images from the multi-energy data set. Further research is necessary to assess whether virtual nonenhanced images are suitable for emphysema quantification.

Second, the comparability of our dual-source and SBF CT groups might have been limited by the different workflows for lobar segmentation between multi-energy CT and SPECT/CT. No commercial application was available at our institution for the lobar segmentation and quantification of multi-energy data. However, the resulting segmentations were checked and, if necessary, manually corrected by board-certified radiologists and nuclear medicine physicians. Comparability of the segmentations was ensured by calculating Pearson correlation coefficients of lobar volumes between SPECT/CT and multi-energy segmentations.

Third, the variance between dual-source or SBF multi-energy CT and SPECT/CT was not negligible in some cases (levels of agreement between -0.17 and 0.14). There could be various reasons for this. Multi-energy and SPECT techniques inherently differ. In multi-energy CT, the pulmonary perfusion is estimated from the distribution of an iodinated contrast agent assessed by a fast scan during peak pulmonary opacification. In contrast, ^99mTc-labeled macroaggregated albumin scintigraphy is based on the fixation of particles in the pulmonary capillaries during the first lung passage, and the scanning process extends over a substantially longer period. Also, functional parameters might have caused perfusion differences because inhalational status differs between multi-energy CT (inspiratory hold) and SPECT (free breathing), and hemodynamics are subject to change. Both parameters influence lung perfusion (34,35).

Still, a mean difference of -0.04 to 0.03 seems tolerable for clinical decision-making because clinicians are aware of a certain degree of variance in measurements. Hence, therapy decisions are usually based on several parameters. In the case of LVRT, the determination of the target lobe is not exclusively based on quantitative pulmonary perfusion but also on lobar volume, gray-scale– and densitometry-based lung tissue analysis, and collateral ventilation.

In conclusion, our study showed that dual-source and SBF CT can be used to accurately assess lung perfusion on a lobar level in patients with lung emphysema compared with SPECT/CT. Multi-energy CT could streamline LVRT assessment by saving radiation dose, time, and money. However, studies with larger patient samples are needed to consolidate an eventual difference in accuracy between dual-source and SBF CT. Furthermore, the performance of multi-energy CT should be correlated with the clinical outcome of endoscopic LVRT.