Optimal virtual monoenergetic image in "TwinBeam" dual-energy CT for organs-at-risk delineation based on contrast-noise-ratio in head-and-neck radiotherapy

Technical note: TIGRE-DE for the creation of virtual monoenergetic images from dual-energy cone-beam CT

BACKGROUND

Dual-energy (DE)-CBCT represents a promising imaging modality that can produce virtual monoenergetic (VM) CBCT images. VM images, which provide enhanced contrast and reduced imaging artifacts, can be used to assist in soft-tissue visualization during image-guided radiotherapy.

PURPOSE

This work reports the development of TIGRE-DE, a module in the open-source TIGRE toolkit for the performance of DE-CBCT and the production of VM CBCT images. This module is created to make DE-CBCT tools accessible in a wider range of clinical and research settings.

METHODS

We developed an add-on (TIGRE-DE) to the TIGRE toolkit that performs DE material decomposition. To verify its performance, sequential CBCT scans at 80 and 140 kV of a Catphan 604 phantom were decomposed into equivalent thicknesses of aluminum (Al) and polymethyl-methylacrylate (PMMA) basis materials. These basis material projections were used to synthesize VM projections for a range of x-ray energies, which were then reconstructed using the Feldkamp-Davis-Kress (FDK) algorithm. Image quality was assessed by computing Hounsfield units (HU) and contrast-to-noise ratios (CNR) for the material inserts of the phantom and comparing with the constituent 80 and 140 kV images.

RESULTS

All VM images generated using TIGRE-DE showed good general agreement with the theoretical HU values of the material inserts of the phantom. Apart from the highest-density inserts imaged at the extremes of the energy range, the measured HU values agree with theoretical HUs within the clinical tolerance of ±50 HU. CNR measurements for the various inserts showed that, of the energies selected, 60 keV provided the highest CNR values. Moreover, 60 keV VM images showed average CNR enhancements of 63% and 66% compared to the 80 and 140 kV full-fan protocols.

CONCLUSIONS

TIGRE-DE successfully implements DE-CBCT material decomposition and VM image creation in an accessible, open-source platform.

Split-filter dual energy computed tomography radiotherapy: From calibration to image guidance

Background and purpose

Dual-energy computed tomography (DECT) is an emerging technology in radiotherapy (RT). Here, we investigate split-filter DECT throughout the RT treatment chain as compared to single-energy CT (SECT).

Materials and methods

DECT scans were acquired with a tin-gold split-filter at 140 kV resulting in a low- and high-energy CT reconstruction (recon). Ten cancer patients (four head-and-neck (HN)​, three rectum​, two anal/pelvis and one abdomen) were DECT scanned without and with iodine administered. A cylindrical and an anthropomorphic HN phantom were scanned with DECT and 120 kV SECT. The DECT images generated were: 120 kV SECT-equivalent (CTmix), virtual monoenergetic images (VMIs), iodine map, virtual non-contrast (VNC), effective atomic number (Zeff), and relative electron density (ρe,w). The clinical utility of these recons was investigated for calibration, delineation, dose calculation and image-guided RT (IGRT).

Results

A calibration curve for 75 keV VMI had a root-mean-square-error (RMSE) of 34 HU in closest agreement with the RSME of SECT calibration. This correlated with a phantom-based dosimetric agreement to SECT of γ1%1mm > 98%. A 40 keV VMI recon was most promising to improve tumor delineation accuracy with an average evaluation score of 1.6 corresponding to "partial improvement". The dosimetric impact of iodine was in general < 2%. For this setup, VNC vs. non-contrast CTmix based dose calculations are considered equivalent. SECT- and DECT-based IGRT was in agreement within the setup uncertainty.

Conclusions

DECT-based RT could be a feasible alternative to SECT providing additional recons to support the different steps of the RT workflow.

High-Pitch Multienergy Coronary CT Angiography in Dual-Source Photon-Counting Detector CT Scanner at Low Iodinated Contrast Dose

OBJECTIVES

The aim of this study was to evaluate the high-helical pitch, multienergy (ME) scanning mode of a clinical dual-source photon-counting detector (PCD) computed tomography (CT) and the benefit of virtual monoenergetic images (VMIs) for low-contrast-dose coronary CT angiography (CTA).

MATERIALS AND METHODS

High-pitch (3.2) ME coronary CTA was performed in PCD-CT in 27 patients using low contrast dose (30 mL of iohexol 350 mg/mL) and in 26 patients at routine contrast dose (60 mL). Low-energy-threshold 120 kV images (also known as T3D images) and 50 kiloelectron volts (50 keV) and 100 kiloelectron volts (100 keV) VMIs were reconstructed using a 1024 × 1024 matrix and 0.6-mm slices. The CT numbers, noise, and contrast-to-noise ratio (CNR) were measured in the ascending aorta (AA), left main coronary artery (LMCA), and distal left anterior descending (LAD) artery. Confidence in grading luminal stenosis with calcific plaque, noncalcific plaque, and stent was evaluated by 2 independent readers on a 0-100 scale (0 the lowest), and a CAD-RADS score was assigned. Image contrast enhancement, sharpness, noise, artifacts, and overall image quality were rated using a 5-point ordinal scale (1 the lowest).

RESULTS

The radiation doses (CTDI) in low- and routine-contrast cohorts were 2.5 ± 0.6 mGy and 3.1 ± 1.7 mGy, respectively ( P = 0.12). At all measured locations, the mean CT number was >300 HU in 120 kV (LMCA 382.9 ± 76.2, distal LAD 341.0 ± 53.9, AA 399.5 ± 76.1) and 50 keV images (LMCA 667.5 ± 139.9, distal LAD 578.1 ± 121.5, AA 700.8 ± 142.5) in the low-contrast cohort, with a 96% increase in CT numbers for 50 keV over 120 kV. The CT numbers were significantly higher ( P < 0.0001) in 50 keV than 120 kV and 100 keV VMI. The CNR was also significantly ( P < 0.0001) higher in 50 keV than 120 kV and 100 keV images in all vessels. Confidence in the assessment of luminal stenosis in the presence of calcific plaque was significantly higher ( P = 0.001) with the addition of 100 keV VMI (median score, 100) than using 50 keV alone (median score, 70) and 120 kV (median score, 70) for reader 1, but no significant differences were seen for reader 2 who had same median scores of 100 for all image types. The confidence in the assessment of luminal stenosis within a stent improved with the use of 100 keV images for both readers (reader 1: median scores for 50 + 100 keV = 100, 50 keV = 82.5, 120 kV = 82.5; reader 2: 50 + 100 keV = 100, 50 keV = 90, 120 kV = 90). There were no significant differences in confidence scores for assessment of luminal stenosis from noncalcific plaques for both readers. The reader-averaged qualitative scores for vascular enhancement and overall image quality were significantly higher for 50 keV VMI than for 120 kV images in both low- and routine-contrast dose cohorts. The image sharpness was nonsignificantly higher at 50 keV VMI than 120 kV images, and the artifact score was comparable for 50 keV VMI and 120 kV images. The noise was higher in 50 keV VMI than in 120 kV images.

CONCLUSIONS

High-pitch ME PCD-CT mode produced diagnostic quality coronary CTA images at low radiation and iodinated contrast doses. The availability of ME VMIs significantly improved the CNR, overall image quality, and confidence in assessment of luminal stenosis in the presence of calcific plaques and stents, and resulted in change of CAD-RADS categories in 9 patients.

Dual-energy CT based mass density and relative stopping power estimation for proton therapy using physics-informed deep learning

Proton therapy requires accurate dose calculation for treatment planning to ensure the conformal doses are precisely delivered to the targets. The conversion of CT numbers to material properties is a significant source of uncertainty for dose calculation. The aim of this study is to develop a physics-informed deep learning (PIDL) framework to derive accurate mass density and relative stopping power maps from dual-energy computed tomography (DECT) images. The PIDL framework allows deep learning (DL) models to be trained with a physics loss function, which includes a physics model to constrain DL models. Five DL models were implemented including a fully connected neural network (FCNN), dual-FCNN (DFCNN), and three variants of residual networks (ResNet): ResNet-v1 (RN-v1), ResNet-v2 (RN-v2), and dual-ResNet-v2 (DRN-v2). An artificial neural network (ANN) and the five DL models trained with and without physics loss were explored to evaluate the PIDL framework. Two empirical DECT models were implemented to compare with the PIDL method. DL training data were from CIRS electron density phantom 062M (Computerized Imaging Reference Systems, Inc., Norfolk, VA). The performance of DL models was tested by CIRS adult male, adult female, and 5-year-old child anthropomorphic phantoms. For density map inference, the physics-informed RN-v2 was 3.3%, 2.9% and 1.9% more accurate than ANN for the adult male, adult female, and child phantoms. The physics-informed DRN-v2 was 0.7%, 0.6%, and 0.8% more accurate than DRN-v2 without physics training for the three phantoms, respectfully. The results indicated that physics-informed training could reduce uncertainty when ANN/DL models without physics training were insufficient to capture data structures or derived significant errors. DL models could also achieve better image noise control compared to the empirical DECT parametric mapping methods. The proposed PIDL framework can potentially improve proton range uncertainty by offering accurate material properties conversion from DECT.

Learning-based synthetic dual energy CT imaging from single energy CT for stopping power ratio calculation in proton radiation therapy

OBJECTIVE

Dual energy CT (DECT) has been shown to estimate stopping power ratio (SPR) map with a higher accuracy than conventional single energy CT (SECT) by obtaining the energy dependence of photon interactions. This work presents a learning-based method to synthesize DECT images from SECT image for proton radiotherapy.

METHODS

The proposed method uses a residual attention generative adversarial network. Residual blocks with attention gates were used to force the model to focus on the difference between DECT images and SECT images. To evaluate the accuracy of the method, we retrospectively investigated 70 head-and-neck cancer patients whose DECT and SECT scans were acquired simultaneously. The model was trained to generate both a high and low energy DECT image based on a SECT image. The generated synthetic low and high DECT images were evaluated against the true DECT images using leave-one-out cross-validation. To evaluate our method in the context of a practical application, we generated SPR maps from synthetic DECT (sDECT) using a dual-energy based stoichiometric method and compared the SPR maps to those generated from DECT. A dosimetric comparison for dose obtained from DECT was performed against that derived from sDECT.

RESULTS

The mean of mean absolute error, peak signal-to-noise ratio and normalized cross-correlation for the synthetic high and low energy CT images was 36.9 HU, 29.3 dB, 0.96 and 35.8 HU, 29.2 dB, and 0.96, respectively. The corresponding SPR maps generated from synthetic DECT showed an average normalized mean square deviation of about 1% with reduced noise level and artifacts than those from original DECT. Dose-volume histogram (DVH) metrics for the clinical target volume agree within 1% between the DECT and sDECT calculated dose.

CONCLUSION

Our method synthesized accurate DECT images and showed a potential feasibility for proton SPR map generation.

ADVANCES IN KNOWLEDGE

This study investigated a learning-based method to synthesize DECT images from SECT image for proton radiotherapy.

Tumorous tissue characterization using integrated 18F-FDG PET/dual-energy CT in lung cancer: Combining iodine enhancement and glycolytic activity

Positron emission tomography/computed tomography (PET/CT) with 18F-fluorodeoxyglucose (18F-FDG) has become the method of choice for tumor staging in lung cancer patients with improved diagnostic accuracy for the evaluation of lymph node involvement and distant metastasis. Due to its spectral capabilities, dual-energy CT (DECT) employs a material decomposition algorithm enabling precise quantification of iodine concentrations in distinct tissues. This technique enhances the characterization of tumor blood supply and has demonstrated promising results for the assessment of therapy response in patients with lung cancer. Several studies have demonstrated that DECT provides additional value to the PET-based evaluation of glycolytic activity, especially for the evaluation of therapy response and follow-up of patients with lung cancer. The combination of PET and DECT in a single scanner system enables the simultaneous assessment of glycolytic activity and iodine enhancement, offering further insight to the characterization of tumorous tissues. Recently a new approach of a novel integrated PET/DECT was investigated in a pilot study on patients with non-small cell lung cancer (NSCLC). The study showed a moderate correlation between PET-based standard uptake values (SUV) and DECT-based iodine densities in the evaluation of lung tumorous tissue but with limited assessment of lymph nodes. The following review on tumorous tissue characterization using PET and DECT imaging describes the strengths and limitations of this novel technique.

Improving radiation physics, tumor visualisation, and treatment quantification in radiotherapy with spectral or dual-energy CT

Over the past decade, spectral or dual‐energy CT has gained relevancy, especially in oncological radiology. Nonetheless, its use in the radiotherapy (RT) clinic remains limited. This review article aims to give an overview of the current state of spectral CT and to explore opportunities for applications in RT.

In this article, three groups of benefits of spectral CT over conventional CT in RT are recognized. Firstly, spectral CT provides more information of physical properties of the body, which can improve dose calculation. Furthermore, it improves the visibility of tumors, for a wide variety of malignancies as well as organs‐at‐risk OARs, which could reduce treatment uncertainty. And finally, spectral CT provides quantitative physiological information, which can be used to personalize and quantify treatment.

Prospective Evaluation of the First Integrated Positron Emission Tomography/Dual-Energy Computed Tomography System in Patients With Lung Cancer

PURPOSE

The aim of this pilot study was to prospectively evaluate the first integrated positron emission tomography (PET)/dual-energy computed tomography (DECT) system performance in patients with non-small cell lung cancer (NSCLC).

MATERIALS AND METHODS

In this single-center, prospective trial, consecutive patients with NSCLC referred for a PET study between May 2017 and June 2018 were enrolled. All patients received contrast-enhanced imaging on a clinical PET/DECT system. Data analysis included PET-based standard uptake values (SUVmax) and DECT-based iodine densities of tumor masses, lymph nodes, and distant metastases. Results were analyzed using correlation tests and receiver operating characteristics curves.

RESULTS

The study population was composed of 21 patients (median age 62 y, 14 male patients). A moderate positive correlation was found between iodine density values (2.2 mg/mL) and SUVmax (10.5) in tumor masses (ρ=0.53, P<0.01). Iodine density values (2.3 mg/mL) and SUVmax (5.4) of lymph node metastases showed a weak positive correlation (ρ=0.23, P=0.14). In addition, iodine quantification analysis provided no added value in differentiating between pathologic and nonpathologic lymph nodes with an area under the curve (AUC) of 0.55 using PET-based SUVmax as the reference standard. A weak positive correlation was observed between iodine density (2.2 mg/mL) and SUVmax in distant metastases (14.9, ρ=0.23, P=0.52).

CONCLUSIONS

The application of an integrated PET/DECT system in lung cancer might provide additional insights in the assessment of tumor masses. However, the added value of iodine density quantification for the evaluation of lymph nodes and distant metastases seems limited.

Knowledge-based radiation treatment planning: A data-driven method survey

This paper surveys the data-driven dose prediction methods investigated for knowledge-based planning (KBP) in the last decade. These methods were classified into two major categories-traditional KBP methods and deep-learning (DL) methods-according to their techniques of utilizing previous knowledge. Traditional KBP methods include studies that require geometric or anatomical features to either find the best-matched case(s) from a repository of prior treatment plans or to build dose prediction models. DL methods include studies that train neural networks to make dose predictions. A comprehensive review of each category is presented, highlighting key features, methods, and their advancements over the years. We separated the cited works according to the framework and cancer site in each category. Finally, we briefly discuss the performance of both traditional KBP methods and DL methods, then discuss future trends of both data-driven KBP methods to dose prediction.

Head and neck single- and dual-energy CT: differences in radiation dose and image quality of 2nd and 3rd generation dual-source CT

OBJECTIVES

To compare radiation dose and image quality of single-energy (SECT) and dual-energy (DECT) head and neck CT examinations performed with second- and third-generation dual-source CT (DSCT) in matched patient cohorts.

METHODS

200 patients (mean age 55.1 ± 16.9 years) who underwent venous phase head and neck CT with a vendor-preset protocol were retrospectively divided into four equal groups (n = 50) matched by gender and BMI: second (Group A, SECT, 100-kV; Group B, DECT, 80/Sn140-kV), and third-generation DSCT (Group C, SECT, 100-kV; Group D, DECT, 90/Sn150-kV). Assessment of radiation dose was performed for an average scan length of 27 cm. Contrast-to-noise ratio measurements and dose-independent figure-of-merit calculations of the submandibular gland, thyroid, internal jugular vein, and common carotid artery were analyzed quantitatively. Qualitative image parameters were evaluated regarding overall image quality, artifacts and reader confidence using 5-point Likert scales.

RESULTS

Effective radiation dose (ED) was not significantly different between SECT and DECT acquisition for each scanner generation (p = 0.10). Significantly lower effective radiation dose (p < 0.01) values were observed for third-generation DSCT groups C (1.1 ± 0.2 mSv) and D (1.0 ± 0.3 mSv) compared to second-generation DSCT groups A (1.8 ± 0.1 mSv) and B (1.6 ± 0.2 mSv). Figure-of-merit/contrast-to-noise ratio analysis revealed superior results for third-generation DECT Group D compared to all other groups. Qualitative image parameters showed non-significant differences between all groups (p > 0.06).

CONCLUSION

Contrast-enhanced head and neck DECT can be performed with second- and third-generation DSCT systems without radiation penalty or impaired image quality compared with SECT, while third-generation DSCT is the most dose efficient acquisition method.

ADVANCES IN KNOWLEDGE

Differences in radiation dose between SECT and DECT of the dose-vulnerable head and neck region using DSCT systems have not been evaluated so far. Therefore, this study directly compares radiation dose and image quality of standard SECT and DECT protocols of second- and third-generation DSCT platforms.

Head and neck multi-organ segmentation on dual-energy CT using dual pyramid convolutional neural networks

Organ delineation is crucial to diagnosis and therapy, while it is also labor-intensive and observer-dependent. Dual energy CT (DECT) provides additional image contrast than conventional single energy CT (SECT), which may facilitate automatic organ segmentation. This work aims to develop an automatic multi-organ segmentation approach using deep learning for head-and-neck region on DECT. We proposed a mask scoring regional convolutional neural network (R-CNN) where comprehensive features are firstly learnt from two independent pyramid networks and are then combined via deep attention strategy to highlight the informative ones extracted from both two channels of low and high energy CT. To perform multi-organ segmentation and avoid misclassification, a mask scoring subnetwork was integrated into the Mask R-CNN framework to build the correlation between the class of potential detected organ's region-of-interest (ROI) and the shape of that organ's segmentation within that ROI. We evaluated our model on DECT images from 127 head-and-neck cancer patients (66 training, 61 testing) with manual contours of 19 organs as training target and ground truth. For large- and mid-sized organs such as brain and parotid, the proposed method successfully achieved average Dice similarity coefficient (DSC) larger than 0.8. For small-sized organs with very low contrast such as chiasm, cochlea, lens and optic nerves, the DSCs ranged between around 0.5 and 0.8. With the proposed method, using DECT images outperforms using SECT in almost all 19 organs with statistical significance in DSC (p<0.05). Meanwhile, by using the DECT, the proposed method is also significantly superior to a recently developed FCN-based method in most of organs in terms of DSC and the 95th percentile Hausdorff distance. Quantitative results demonstrated the feasibility of the proposed method, the superiority of using DECT to SECT, and the advantage of the proposed R-CNN over FCN on the head-and-neck patient study. The proposed method has the potential to facilitate the current head-and-neck cancer radiation therapy workflow in treatment planning.

Computed Tomography- (CT-) Based Virtual Surgery Planning for Spinal Intervertebral Foraminal Assisted Clinical Treatment

With the development of minimally invasive spine concepts and the introduction of new minimally invasive instruments, minimally invasive spine technology, represented by foraminoscopy, has flourished, and percutaneous foraminoscopy has become one of the most reliable minimally invasive procedures for the treatment of lumbar disc herniation. Percutaneous foraminoscopy is a safe and effective minimally invasive spinal endoscopic surgical technique. It fully protects the paravertebral muscles and soft tissues as well as the posterior column structure of the spine, provides precise treatment of the target nucleus pulposus tissue, with the advantages of less surgical trauma, fewer postoperative complications, and rapid postoperative recovery, and is widely promoted and used in clinical practice. In this paper, we can view the location, morphology, structure, alignment, and adjacency relationships by performing coronary, CT, and diagonal reconstruction along the attachment of the yellow ligaments and performing 3D reconstruction or processing techniques after performing CT scans. This allows clinicians to observe the laminoplasty and the stenosis of the vertebral canal in a more intuitive and overall manner. It has clinical significance for the display of the sublaminar spine as well as the physician's judgment of the disease and the choice of surgery.

Learning-Based Stopping Power Mapping on Dual-Energy CT for Proton Radiation Therapy

Purpose

Dual-energy computed tomography (DECT) has been used to derive relative stopping power (RSP) maps by obtaining the energy dependence of photon interactions. The DECT-derived RSP maps could potentially be compromised by image noise levels and the severity of artifacts when using physics-based mapping techniques. This work presents a noise-robust learning-based method to predict RSP maps from DECT for proton radiation therapy.

Materials and Methods

The proposed method uses a residual attention cycle-consistent generative adversarial network to bring DECT-to-RSP mapping close to a 1-to-1 mapping by introducing an inverse RSP-to-DECT mapping. To evaluate the proposed method, we retrospectively investigated 20 head-and-neck cancer patients with DECT scans in proton radiation therapy simulation. Ground truth RSP values were assigned by calculation based on chemical compositions and acted as learning targets in the training process for DECT datasets; they were evaluated against results from the proposed method using a leave-one-out cross-validation strategy.

Results

The predicted RSP maps showed an average normalized mean square error of 2.83% across the whole body volume and an average mean error less than 3% in all volumes of interest. With additional simulated noise added in DECT datasets, the proposed method still maintained a comparable performance, while the physics-based stoichiometric method suffered degraded inaccuracy from increased noise level. The average differences from ground truth in dose volume histogram metrics for clinical target volumes were less than 0.2 Gy for D_95% and D_max with no statistical significance. Maximum difference in dose volume histogram metrics of organs at risk was around 1 Gy on average.

Conclusion

These results strongly indicate the high accuracy of RSP maps predicted by our machine-learning–based method and show its potential feasibility for proton treatment planning and dose calculation.

Multiparametric dual-energy CT for distinguishing nasopharyngeal carcinoma from nasopharyngeal lymphoma

OBJECTIVES

To determine the optimal kiloelectron volt of noise-optimized virtual monoenergetic images [VMI (+)] for visualization of nasopharyngeal carcinoma (NPC) and nasopharyngeal lymphoma (NPL), and to explore the clinical value of quantitative parameters derived from dual-energy computed tomography (DECT) for distinguishing the two entities.

MATERIALS AND METHODS

Eighty patients including 51 with NPC and 29 with NPL were enrolled. The VMIs (+) at 40-80 keV with an interval of 10 keV were reconstructed by contrast enhanced images. The overall image quality and demarcation of lesion margins, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were assessed in VMIs (+) and polyenergetic images (PEI). Normalized iodine concentration (NIC), slope of the spectral Hounsfield unit curve (λ_HU) and effective atomic number (Z_eff) were calculated. Diagnostic performance was assessed by receiver operating characteristic (ROC) curve.

RESULTS

The 40 keV VMI (+) yielded highest overall image quality scores, demarcation of lesion margins scores, SNR and CNR. The values of NIC, λ_HU and Z_eff in NPL were higher than those in NPC (P <  0.001). Multivariate logistic regression model combining NIC, λ_HU and Z_eff showed the best performance for distinguishing NPC from NPL (AUC: 0.947, sensitivity: 93.1 % and specificity: 92.2 %).

CONCLUSION

VMI (+) reconstruction at 40 keV was optimal for visualizing NPC and NPL. Quantitative parameters derived from DECT were helpful for differentiating NPC from NPL.

Deep learning in medical image registration: a review

This paper presents a review of deep learning (DL)-based medical image registration methods. We summarized the latest developments and applications of DL-based registration methods in the medical field. These methods were classified into seven categories according to their methods, functions and popularity. A detailed review of each category was presented, highlighting important contributions and identifying specific challenges. A short assessment was presented following the detailed review of each category to summarize its achievements and future potential. We provided a comprehensive comparison among DL-based methods for lung and brain registration using benchmark datasets. Lastly, we analyzed the statistics of all the cited works from various aspects, revealing the popularity and future trend of DL-based medical image registration.

Cone-beam CT-derived relative stopping power map generation via deep learning for proton radiotherapy

PURPOSE

In intensity-modulated proton therapy (IMPT), protons are used to deliver highly conformal dose distributions, targeting tumors, and sparing organs-at-risk. However, due to uncertainties in both patient setup and relative stopping power (RSP) calculation, margins are added to the treatment volume during treatment planning, leading to higher doses to normal tissues. Cone-beam computed tomography (CBCT) images are taken daily before treatment; however, the poor image quality of CBCT limits the use of these images for online dose calculation. In this work, we use a deep-learning-based method to predict RSP maps from daily CBCT images, allowing for online dose calculation in a step toward adaptive radiation therapy.

METHODS

Twenty-three head-and-neck cancer patients were simulated using a Siemens TwinBeam dual-energy CT (DECT) scanner. Mixed-energy scans (equivalent to a 120 kVp single-energy CT scan) were converted to RSP maps for treatment planning. Cone-beam computed tomography images were taken on the first day of treatment, and the planning RSP maps were registered to these images. A deep learning network based on a cycle-GAN architecture, relying on a compound loss function designed for structural and contrast preservation, was then trained to create an RSP map from a CBCT image. Leave-one-out and holdout cross validations were used for evaluation, and mean absolute error (MAE), mean error (ME), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) were used to quantify the differences between the CT-based and CBCT-based RSP maps. The proposed method was compared to a deformable image registration-based method which was taken as the ground truth and two other deep learning methods. For one patient who underwent resimulation, the new planning RSP maps and CBCT images were used for further evaluation and validation.

RESULTS

The CBCT-based RSP generation method was evaluated on 23 head-and-neck cancer patients. From leave-one-out testing, the MAE between CT-based and CBCT-based RSP was 0.06 ± 0.01 and the ME was -0.01 ± 0.01. The proposed method statistically outperformed the comparison DL methods in terms of MAE and ME when compared to the planning CT. In terms of dose comparison, the mean gamma passing rate at 3%/3 mm was 94% when three-dimensional (3D) gamma index was calculated per plan and 96% when gamma index was calculated per field.

CONCLUSIONS

The proposed method provides sufficiently accurate RSP map generation from CBCT images, allowing for evaluation of daily dose based on CBCT and possibly allowing for CBCT-guided adaptive treatment planning for IMPT.

Auto-segmentation of organs at risk for head and neck radiotherapy planning: From atlas-based to deep learning methods

Radiotherapy (RT) is one of the basic treatment modalities for cancer of the head and neck (H&N), which requires a precise spatial description of the target volumes and organs at risk (OARs) to deliver a highly conformal radiation dose to the tumor cells while sparing the healthy tissues. For this purpose, target volumes and OARs have to be delineated and segmented from medical images. As manual delineation is a tedious and time-consuming task subjected to intra/interobserver variability, computerized auto-segmentation has been developed as an alternative. The field of medical imaging and RT planning has experienced an increased interest in the past decade, with new emerging trends that shifted the field of H&N OAR auto-segmentation from atlas-based to deep learning-based approaches. In this review, we systematically analyzed 78 relevant publications on auto-segmentation of OARs in the H&N region from 2008 to date, and provided critical discussions and recommendations from various perspectives: image modality - both computed tomography and magnetic resonance image modalities are being exploited, but the potential of the latter should be explored more in the future; OAR - the spinal cord, brainstem, and major salivary glands are the most studied OARs, but additional experiments should be conducted for several less studied soft tissue structures; image database - several image databases with the corresponding ground truth are currently available for methodology evaluation, but should be augmented with data from multiple observers and multiple institutions; methodology - current methods have shifted from atlas-based to deep learning auto-segmentation, which is expected to become even more sophisticated; ground truth - delineation guidelines should be followed and participation of multiple experts from multiple institutions is recommended; performance metrics - the Dice coefficient as the standard volumetric overlap metrics should be accompanied with at least one distance metrics, and combined with clinical acceptability scores and risk assessments; segmentation performance - the best performing methods achieve clinically acceptable auto-segmentation for several OARs, however, the dosimetric impact should be also studied to provide clinically relevant endpoints for RT planning.

Status and innovations in pre-treatment CT imaging for proton therapy

Pre-treatment CT imaging is a topic of growing importance in particle therapy. Improvements in the accuracy of stopping-power prediction are demanded to allow for a dose conformality that is not inferior to state-of-the-art image-guided photon therapy. Although range uncertainty has been kept practically constant over the last decades, recent technological and methodological developments, like the clinical application of dual-energy CT, have been introduced or arise at least on the horizon to improve the accuracy and precision of range prediction. This review gives an overview of the current status, summarizes the innovations in dual-energy CT and its potential impact on the field as well as potential alternative technologies for stopping-power prediction.

Fast-switching dual energy cone beam computed tomography using the on-board imager of a commercial linear accelerator

To evaluate fast-kV switching (FS) dual energy (DE) cone beam computed tomography (CBCT) using the on-board imager (OBI) of a commercial linear accelerator to produce virtual monoenergetic (VM) and relative electron density (RED) images. Using an polynomial attenuation mapping model, CBCT phantom projections obtained at 80 and 140 kVp with FS imaging, were decomposed into equivalent thicknesses of aluminum (Al) and polymethyl methacrylate (PMMA). All projections were obtained with the titanium foil and bowtie filter in place. Basis material projections were then recombined to create VM images by using the linear attenuation coefficients at the specified energy for each material. Similarly, RED images were produced by replacing the linear attenuation values of Al and PMMA by their respective RED values in the projection space. VM and RED images were reconstructed using Feldkamp-Davis-Kress (FDK) and an iterative algorithm (iCBCT, Varian Medical Systems). Hounsfield units (HU), contrast-to-noise ratio (CNR) and RED values were compared against known values. The results after VM-CBCT production showed good material decomposition and consistent HU_VM values, with measured root mean square errors (RMSE) from theoretical values, after FDK reconstruction, of 20.5, 5.7, 12.8 and 21.7 HU for 50, 80, 100 and 150 keV, respectively. The largest CNR improvements, when compared to polychromatic images, were observed for the 50 keV VM images. Image noise was reduced up to 28% in the VM-CBCT images after iterative image reconstruction. RED values measured for our method resulted in a mean percentage error of 0.0%  ±  1.8%. This study describes a method to generate VM-CBCT and RED images using FS-DE scans obtained using the OBI of a linac, including the effects of the bowtie filter. The creation of VM and RED images increases the dynamic range of CBCT images, and provides additional data that may be used for adaptive radiotherapy, and on table verification for radiotherapy treatments.