Validation and optimization of dual-energy CT for accurate dose calculation in photon radiotherapy of brain metastases.

Clear representation of anatomy is essential in the assessment of pathology in computed tomography (CT). With the introduction of photon-counting CT (PCCT) and more advanced iterative reconstruction (IR) algorithms into clinical practice, there is potential to improve low-contrast detectability in CT protocols. As such, it is necessary to perform task-based assessments to optimize protocols and compare image quality between PCCT and energy-integrating CT, like dual-energy CT (DECT) and single-energy CT (SECT). This work aimed to assess low-contrast detectability in abdominal protocols used in clinical PCCT, DECT, and SECT, using both model and humanobservers. Data were acquired with the standard resolution scan mode on a PCCT (NAEOTOM Alpha, Siemens Healthineers, Forchheim, Germany) and a DECT/SECT (SOMATOM Force, Siemens Healthineers, Forchheim, Germany). Detectability was investigated in the CTP 515 low-contrast module of the Catphan 600 phantom, which was surrounded by a fat annulus to simulate an abdomen and resulted in a water equivalent diameter of 298 mm. Supra-slice contrast rods with a nominal 1.0% contrast and diameters of 4, 6, 9, and 15 mm were used. Factory abdominal protocols were adjusted to acquire images with various tube potentials (70, 90, 120, and 140 kV in PCCT; 70/150Sn and 80/150Sn kV in DECT; 100 and 120 kV in SECT), virtual monoenergetic image (VMI) energy levels (40 to 140 keV in PCCT and DECT), doses (5, 10 mGy in PCCT; 10 mGy in DECT and SECT), and IR settings (Br40 kernel, no quantum IR (QIR) and QIR levels 1 to 4 in PCCT; advanced modeled IR (ADMIRE) level 3 in DECT and SECT). Mixed DECT (linear blending of the images at two tube voltages) images were also reconstructed. The noise power spectrum and task transfer function of each scan protocol were quantified; the detectability index for each protocol was also determined using in-house implementations of model observers (non-prewhitening matched filters with internal noise, NPWI, and with an eye filter and internal noise, NPWEI) and human observers (in-house four-alternative forced choice, scoring with 95% confidence intervals). Results show that the image noise is minimized at a VMI energy corresponding to the applied spectrum's mean energy in PCCT and with VMI settings of 70 and 80 keV for 70/150Sn and 80/150Sn tube potential pairs, respectively, in DECT. With respect to the human observer detectability index calculations, the normalized root-mean-square error for the NPWI and NPWEI model observers was 5% and 12%, respectively. PCCT VMI improves low-contrast detectability. Additionally, detectability can be matched between PCCT protocols by increasing the QIR strength level when reducing the dose. Not only does PCCT VMI outperform DECT VMI, but also DECT VMI outperforms DECT mixed imaging in improving low-contrastdetectability. Low-contrast detectability is optimized when the appropriate VMI energy level is selected in PCCT and DECT to minimize image noise. PCCT improves low-contrast detectability and may allow for dose reduction in abdominal protocols compared to both DECT and SECT. The non-prewhitening model observer with internal noise better quantified low-contrast detectability without the inclusion of an eyefilter.

Purpose: To reduce uncertainties in CT data conversion to stopping power ratio (SPR) and tissue composition when using single energy CT (SECT), this study presents a first approach of exploiting dual energy CT (DECT) data for Monte Carlo (MC) based dose calculation in particle therapy. Methods: DECT image data can be converted into an electron density and effective atomic number image. With both tissue parameters elemental weights of 71 tabulated tissues were predicted by dedicated linear fits for each element. The mass density was derived via a single linear fit of the electron density from adipose on. Dose calculations with the MC system TOPAS were performed using monoenergetic pencil beams (protons and carbons) stopped in 12 selected tissues which showed considerable differences in composition predictions from single and dual energy CT. Ranges were compared with predictions from the novel DECT approach and the standard SECT. Results: Predictions of mass density, carbon and oxygen elemental weights profit highly from the additional DECT information. Maximum differences to the true mass density could be reduced from 3.0% (SECT) to 0.9% (DECT) for soft tissue and from 0.7% (SECT) to 0.1% (DECT) for bone tissue. Furthermore, DECT shows a significant improvement also in common tissues: range uncertainties in cartilage were reduced from 1.7% (SECT) to 0.2% (DECT), in yellow marrow from 1.7% (SECT) to 0.5% (DECT), in liver from 0.9% (SECT) to 0.2% (DECT), in brain cerebroisal fluid from 2.2% (SECT) to 0.0% (DECT) and in femur the predicted range deviates by 0.8% (SECT) to 0.0% (DECT) from the true value. Conclusion: DECT can provide more accurate material compositions and mass density as compared to single energy CT. Therewith, DECT information can significantly reduce range uncertainties in particle therapy and can lead to a reduction in currently applied range margins in proton and carbon ion therapy.

Dual-energy computed tomography (DECT) is increasingly used in radiotherapy delineation due to its enhanced soft tissue contrast. DECT also supports direct dose calculation. However, as most current DECT scanners allow for use in only certain body regions, conventional single-energy computed tomography (SECT) is still needed for some patients. A safe clinical introduction of DECT thus requires a combined workflow. This study therefore investigates whether a unified Hounsfield look-up table (HLUT) can be applied across SECT and DECT reconstructions. Patient/material and methods: A Gammex Advanced Electron Density phantom containing tissue-equivalent inserts was scanned using SECT (70-140 kVp and Sn100-Sn140 kVp, Sn meaning tin-filtered) and dual-spiral DECT to identify matching HLUTs for three SECT methods, including a standard reconstruction (only 120 kVp; Method 1), and kVp-independent reconstructions providing mass density (MD; Method 2) or relative electron density (RED; Method 3). Dose agreement was subsequently tested on two anthropomorphic phantoms. For each SECT method, DECT reconstructions were compared through voxel-wise analysis of computed tomography (CT) numbers, and by performing dose calculations in three anatomical regions: head, thorax, and abdomen/pelvis. Across all three SECT methods, DECT reconstructions with acceptable clinical CT number agreement were identified. Corresponding dose calculations between SECT- and DECT-based plans showed minimal differences. This phantom study demonstrates that a unified HLUT can be applied across SECT and DECT using standard 120 kVp, MD, or RED reconstructions. This approach may streamline clinical workflows and support a safe and practical transition to DECT-based treatment planning.

Dual-energy computed tomography (CT) can provide a range of image information beyond conventional CT through virtual monoenergetic images (VMIs). The purpose of this study was to investigate the impact of material decomposition in detector-based spectral CT on radiomics features and effectiveness of using deep learning-based image synthesis to improve the reproducibility of radiomics features. In this paper, spectral CT image data from 45 esophageal cancer patients were collected for investigation retrospectively. First, we computed the correlation coefficient of radiomics features between conventional kilovoltage peak (kVp) CT images and VMI. Then, a wavelet loss-enhanced CycleGAN (WLL-CycleGAN) with paired loss terms was developed to synthesize virtual monoenergetic CT images from the corresponding conventional single-energy CT (SECT) images for improving radiomics reproducibility. Finally, the radiomic features in 6 different categories, including gray-level co-occurrence matrix (GLCM), gray-level difference matrix (GLDM), gray-level run-length matrix (GLRLM), gray-level size-zone matrix (GLSZM), neighborhood gray-tone difference matrix (NGTDM), and wavelet, were extracted from the gross tumor volumes from conventional single energy CT, synthetic virtual monoenergetic CT images, and virtual monoenergetic CT images. Comparison between errors in the VMI and synthetic VMI (sVMI) suggested that the performance of our proposed deep learning method improved the radiomic feature accuracy. Material decomposition of dual-layer dual-energy CT (DECT) can substantially influence the reproducibility of the radiomic features, and the degree of impact is feature dependent. The average reduction of radiomics errors for 15 patients in testing sets was 96.9% for first-order, 12.1% for GLCM, 12.9% for GLDM, 15.7% for GLRLM, 50.3% for GLSZM, 53.4% for NGTDM, and 6% for wavelet features. The work revealed that material decomposition has a significant effect on the radiomic feature values. The deep learning-based method reduced the influence of material decomposition in VMIs and might improve the robustness and reproducibility of radiomic features in esophageal cancer. Quantitative results demonstrated that our proposed wavelet loss-enhanced paired CycleGAN outperforms the original CycleGAN.

Purpose: AAPM TG43, the low dose rate (LDR) brachytherapy dosimetry protocol, ignores the effects of tissue composition and densities. Given the photoelectric effectˈs dominance at the low photon energies in use, the dosimetry is dependent on the elemental composition of tissues. AAPM TG186 was mandated to investigate model based dose calculations (MBDC) such as Monte Carlo (MC) simulations as an alternative to TG43. In MBDC, tissue densities and composition are extracted from single energy CT (SECT) data, which can lead to tissue missassignment. The novel dual energy CT (DECT) technology which is now becoming available may improve tissue segmentation. The study aim is to assess the dose calculation accuracy improvement provided by DECT based tissue segmentation compared to SECT. Methods: A Siemens DECT scanner was modelled with ImaSim, a novel CT image simulation tool. Density and atomic number maps of a virtual phantom containing 23 human tissues were extracted from DECT images taken at 80 kVp and 140 kVp. SECT images taken at 120 kVp were also used for segmentation using 3 and 7 tissues. MC dose calculations were performed with 103‐Pd and 125‐I seeds inserted in the phantom for the three segmentation schemes and compared to a reference calculation. Results: SECT segmentation with 3 tissues performs better than the TG43 approach but does not provide adequate dose calculation accuracy. Increasing the number of tissues to 7 significantly improves accuracy, although errors of more than 10% are observed for certain missassigned tissues. Using DECT segmentation brings accuracy within ±5%. Conclusions: DECT provides superior tissue segmentation resulting in high dose calculation accuracy. While SECT segmentation is outperformed by DECT, dose calculation accuracy is found to be tolerable when using seven tissues, supporting the implementation of MBDC based on widely available SECT.

Metal artifact reduction in cervix brachytherapy with titanium applicators using dual‐energy CT through virtual monoenergetic images and an iterative algorithm: A phantom study

To compare the radiation exposure from single-energy CT (SECT) against rapid kV-switching dual-energy CT (DECT) imaging in both adults and children when resulting image data offer equivalent lesion identification power. Lesions in an adult and a 10-year-old-child body phantom were imitated using iodine solutions of different concentrations. Phantoms were subjected to several SECT and DECT thoracic and abdominal scans using a rapid kV-switching DECT scanner. The contrast-to-noise ratio (CNR) of each lesion was measured on resulting SECT images and virtual monoenergetic images (VMI) available from DECT. The SECT scans that resulted in CNR values similar to the maximum CNR observed in VMIs derived from corresponding DECT scans were identified. SECT and DECT scans with equivalent lesion-discriminating power were compared regarding the associated radiation dose burden. Doses to the lung, breast, and esophagus from thoracic imaging and doses to the liver, kidneys, and stomach from abdominal imaging were determined through Monte Carlo simulations of SECT and DECT exposures. Compared to SECT imaging of the adult body phantom, organ doses from DECT were found to be 5-11% lower in thoracic imaging and 44-45% lower in abdominal imaging. Compared to SECT imaging of the 10-year-old body phantom, organ doses from DECT were found to be 2.8-3.4 times higher in thoracic imaging and 1.5-1.6 times higher in abdominal imaging. The use of rapid kV-switching DECT instead of SECT imaging may be associated with a similar or lower dose burden in adults but a noticeably higher dose burden in children. Question How does the radiation exposure from single-energy and dual-energy CT imaging compare when both techniques provide equivalent lesion identification power? Findings Rapid kV-switching dual-energy CT compared to single-energy CT may result in a similar or lower radiation dose in adults, but higher radiation dose in children. Clinical relevance Rapid kV-switching dual-energy CT imaging in children should be preferred over single-energy CT imaging only in cases where the additional information provided is crucial for an effective diagnosis.

ID: 114 The impact of dual-energy CT tissue segmentation for low-dose rate prostate brachytherapy Monte Carlo dose calculations

Low keV virtual monoenergetic (VME) images are effective in enhancing vessel opacification but require dual-energy CT (DECT), limiting widespread clinical use. Recent advancements in deep learning (DL) enable the generation of VME images from single-energy CT (SECT). However, the performance of the methods has not been evaluated in any clinical use case. The purpose of this work was to assess both objective and subjective image quality of deep learning-based VME images derived from heterogeneous SECT data for pulmonary angiography. In this retrospective study, 52 sets of SECT pulmonary angiography images were processed using a deep learning method to estimate material basis images. 40keV VME images were generated from heterogeneous SECT data using a pretrained physics-constrained Deep-En-Chroma DL model. Two thoracic radiologists, blinded to the image reconstruction method, evaluated pulmonary vessel opacification and overall image quality on DL-VME and SECT images using 5-point Likert scales. Objective image quality was assessed by measuring enhanced vessel contrast and contrast-to-noise ratio (CNR). Statistical analysis was performed using paired t tests and Mann-Whitney U tests. Compared with SECT, DL-VME images demonstrated significantly higher subjective image quality score and vessel opacification score (P≤0.008). DL-VME yielded a higher average contrast for emboli (1085 vs. 331HU, P<0.001) and improved CNR (17.8 vs. 11.1, P<0.001). Results of subgroup analysis indicate no significant variation in VME performance across patient sex, scanner model, radiation dose, and tube potential. The vessel opacification scores of both VME and SECT demonstrate dependence on patient weight, with VME providing better vessel opacity for both lighter and heavier patients. A measure of 40keV DL-VME derived from SECT effectively enhances both vessel opacification and image quality in CT pulmonary angiography. The image quality advantage of DL-VME over SECT remains robust across variations in data acquisition and patient variables.

Metal artefacts pose a common problem in single energy computed tomography (SECT) images used for radiotherapy. Virtual monoenergetic (VME) images constructed with dual energy computed tomography (DECT) scans can be used to reduce beam hardening artefacts. Dual energy metal artefact reduction is compared and combined with iterative metal artefact reduction (iMAR) to determine optimal imaging strategies for patients with metal prostheses. SECT and DECT scans were performed on a Siemens Somatom AS-64 Slice CT scanner. Images were acquired of a modified CIRS pelvis phantom with 6, 12, 20mm diameter stainless steel rods and VME images reconstructed at 100, 120, 140 and 190keV. These were post-reconstructed with and without the iMAR algorithm. Artefact reduction was measured using: (1) the change in Hounsfield Unit (HU) with and without metal artefact reduction (MAR) for 4 regions of interest; (2) the total number of artefact pixels, defined as pixels with a difference (between images with metal rod and without) exceeding a threshold; (3) the difference in the mean pixel intensity of the artefact pixels. DECT, SECT + iMAR and DECT + iMAR were compared. Both SECT + iMAR and DECT + iMAR offer successful MAR for phantom simulating unilateral hip prosthesis. DECT gives minimal artefact reduction over iMAR alone. Quantitative metrics are advantageous for MAR analysis but have limitations that leave room for metric development.

ObjectivesTo study whether photon-counting computed tomography (PCCT) can improve CT number accuracy and precision and reduce patient size dependence compared to dual-energy CT (DECT) virtual monoenergetic imaging (VMI) and single-energy CT (SECT).MethodsClinical PCCT, DECT, and SECT scanners were used to image a multi-energy quality assurance phantom and tissue-equivalent inserts with/without an outer nested annulus, representing 2 object sizes (18 and 33 cm). CT numbers were converted to linear attenuation coefficients (LAC) and regions of interest applied. Theoretical monoenergetic LAC were calculated from known elemental compositions as a ground truth. Percent differences in mean LAC between phantom sizes, between mean and theoretical LAC, and its coefficient of variation (COV) were calculated.ResultsMean LAC percent differences between small and larger phantoms were highest in DECT (within −3% to 9%) and SECT (within 1%-5%), particularly at higher calcium and iodine concentrations, while being relatively constant in PCCT over material concentrations and VMI energies (within ±2%). The COV in mean LAC was consistently lower (about 2-5 times) in PCCT relative to DECT and SECT for calcium in the large phantom. With consideration of the theoretical uncertainties of 2%, both PCCT and DECT showed comparable agreement to theoretical LAC.ConclusionsPCCT VMI produces CT numbers with less dependence on patient size and increased precision in large object sizes than DECT VMI and SECT.Advances in knowledgeClinical PCCT provides less variable CT numbers than DECT and SECT with less sensitivity to the imaged object size.

4. Dual energy CT: Iterative metal artifact reduction for radiotherapy

To determine the iodine contrast-to-noise ratio (CNR) for abdominal computed tomography (CT) when using energy domain noise reduction and virtual monoenergetic dual-energy (DE) CT images and to compare the CNR to that attained with single-energy CT at 80, 100, 120, and 140 kV. This HIPAA-compliant study was approved by the institutional review board with waiver of informed consent. A syringe filled with diluted iodine contrast material was placed into 30-, 35-, and 45-cm-wide water phantoms and scanned with a dual-source CT scanner in both DE and single-energy modes with matched scanner output. Virtual monoenergetic images were generated, with energies ranging from 40 to 110 keV in 10-keV steps. A previously developed energy domain noise reduction algorithm was applied to reduce image noise by exploiting information redundancies in the energy domain. Image noise and iodine CNR were calculated. To show the potential clinical benefit of this technique, it was retrospectively applied to a clinical DE CT study of the liver in a 59-year-old male patient by using conventional and iterative reconstruction techniques. Image noise and CNR were compared for virtual monoenergetic images with and without energy domain noise reduction at each virtual monoenergetic energy (in kiloelectron volts) and phantom size by using a paired t test. CNR of virtual monoenergetic images was also compared with that of single-energy images acquired with 80, 100, 120, and 140 kV. Noise reduction of up to 59% (28.7 of 65.7) was achieved for DE virtual monoenergetic images by using an energy domain noise reduction technique. For the commercial virtual monoenergetic images, the maximum iodine CNR was achieved at 70 keV and was 18.6, 16.6, and 10.8 for the 30-, 35-, and 45-cm phantoms. After energy domain noise reduction, maximum iodine CNR was achieved at 40 keV and increased to 30.6, 25.4, and 16.5. These CNRs represented improvement of up to 64% (12.0 of 18.6) with the energy domain noise reduction technique. For single-energy CT at the optimal tube potential, iodine CNR was 29.1 (80 kV), 21.2 (80 kV), and 11.5 (100 kV). For patient images, 39% (24 of 61) noise reduction and 67% (0.74 of 1.10) CNR improvement were observed with the energy domain noise reduction technique when compared with standard filtered back-projection images. Iodine CNR for adult abdominal CT may be maximized with energy domain noise reduction and virtual monoenergetic DE CT.

BackgroundTo investigate the feasibility of using dual-energy CT (DECT) for tissue segmentation and kilovolt (kV) dose calculations in pre-clinical studies and assess potential dose calculation accuracy gain.MethodsTwo phantoms and an ex-vivo mouse were scanned in a small animal irradiator with two distinct energies. Tissue segmentation was performed with the single-energy CT (SECT) and DECT methods. A number of different material maps was used. Dose calculations were performed to verify the impact of segmentations on the dose accuracy.ResultsDECT showed better overall results in comparison to SECT. Higher number of DECT segmentation media resulted in smaller dose differences in comparison to the reference. Increasing the number of materials in the SECT method yielded more instability. Both modalities showed a limit to which adding more materials with similar characteristics ceased providing better segmentation results, and resulted in more noise in the material maps and the dose distributions. The effect was aggravated with a decrease in beam energy. For the ex-vivo specimen, the choice of only one high dense bone for the SECT method resulted in large volumes of tissue receiving high doses. For the DECT method, the choice of more than one kind of bone resulted in lower dose values for the different tissues occupying the same volume. For the organs at risk surrounded by bone, the doses were lower when using the SECT method in comparison to DECT, due to the high absorption of the bone. SECT material segmentation may lead to an underestimation of the dose to OAR in the proximity of bone.ConclusionsThe DECT method enabled the selection of a higher number of materials thereby increasing the accuracy in dose calculations. In phantom studies, SECT performed best with three materials and DECT with seven for the phantom case. For irradiations in preclinical studies with kV photon energies, the use of DECT segmentation combined with the choice of a low-density bone is recommended.

Comparison of image quality, contrast administration, and radiation doses in pediatric abdominal dual-layer detector dual-energy CT using propensity score matching analysis