Search

Purpose To describe and characterize fast-kV switching, dual-energy (DE) imaging implemented within the on-board imager of a commercial linear accelerator for markerless tumor tracking (MTT). Methods and Materials Fast-kV switching, DE imaging provides for rapid switching between programmed tube voltages (ie, 60 and 120 kVp) from one image frame to the next. To characterize this system, the weighting factor used for logarithmic subtraction and signal difference-to-noise ratio were analyzed as a function of time and frame rate. MTT was evaluated using a thorax motion phantom and fast kV, DE imaging was compared versus single energy (SE) imaging over 360 degrees of rotation. A template-based matching algorithm was used to track target motion on both DE and SE sequences. Receiver operating characteristics were used to compare tracking results for both modalities. Results The weighting factor was inversely related to frame rate and stable over time. After applying the frame rate–dependent weighting factor, the signal difference-to-noise ratio was consistent across all frame rates considered for simulated tumors ranging from 5 to 25 mm in diameter. An analysis of receiver operating characteristics curves showed improved tracking with DE versus SE imaging. The area under the curve for the 10-mm target ranged from 0.821 to 0.858 for SE imaging versus 0.968 to 0.974 for DE imaging. Moreover, the residual tracking errors for the same target size ranged from 2.02 to 2.18 mm versus 0.79 to 1.07 mm for SE and DE imaging, respectively. Conclusions Fast-kV switching, DE imaging was implemented on the on-board imager of a commercial linear accelerator. DE imaging resulted in improved MTT accuracy over SE imaging. Such an approach may have application for MTT of patients with lung cancer receiving stereotactic body radiation therapy, particularly for small tumors where MTT with SE imaging may fail.

Aim/objectives/background The American College of Radiology (ACR) and the American Society for Radiation Oncology (ASTRO) have jointly developed the following practice parameter for image-guided radiation therapy (IGRT). IGRT is radiation therapy that employs imaging to maximize accuracy and precision throughout the entire process of treatment delivery with the goal of optimizing accuracy and reliability of radiation therapy to the target, while minimizing dose to normal tissues. Methods The ACR-ASTRO Practice Parameter for IGRT was revised according to the process described on the ACR website ("The Process for Developing ACR Practice Parameters and Technical Standards," www.acr.org/ClinicalResources/Practice-Parametersand-Technical-Standards) by the Committee on Practice Parameters of the ACR Commission on Radiation Oncology in collaboration with the ASTRO. Both societies then reviewed and approved the document. Results This practice parameter is developed to serve as a tool in the appropriate application of IGRT in the care of patients with conditions where radiation therapy is indicated. It addresses clinical implementation of IGRT including personnel qualifications, quality assurance standards, indications, and suggested documentation. Conclusions This practice parameter is a tool to guide clinical use of IGRT and does not make recommendations on site-specific IGRT directives. It focuses on the best practices and principles to consider when using IGRT effectively, especially with the significant increase in imaging data that is now available with IGRT. The clinical benefit and medical necessity of the imaging modality and frequency of IGRT should be assessed for each patient.

Purpose To predict biological effects of targeted alpha therapy (TAT) in preclinical studies, dosimetry calculations based on the micro-level distributions of emitters are essential. Due to the saturation of the tumor antigenic sites and bonding breaks by decay, some of Alpha-immuno-conjugate and decay daughters may inevitably be transported by convection and diffusion along with blood or lymphatic circulation. This results in highly nonuniform and unsteady distributions of irradiation sources. Since the micro-level distribution of emitters cannot be measured and obtained in patients with current technology, a modeling toolset to give more insight of the internal dose could be an alternative. Methods A multi-physics model based on a Monte Carlo microdosimetry technique and computational fluid dynamics (CFD) modeling was developed and applied to multiple internal irradiation sources. The CFD model tracks the path of the radionuclides and the dose model is capable of evaluating the time-dependent absorbed dose to the target. Results The conceptual model is capable of handling complex nonuniform irradiation sources in vasculature. The results from the simulations indicate that the assumption of homogeneous and motionless distribution of the administered activity used in the conventional dose calculation tends to significantly underestimate or overestimate the absorbed dose to the vascular system in various scenarios. Conclusion Modeling the in vivo transport of radionuclides has the potential to improve the accuracy of TAT dose estimates. It could be the first step to develop a simulation tool set for assessing absorbed dose to tumor or normal tissues and predict the corresponding biological responses in the future.

Purpose Although stereotactic body radiation therapy (SBRT) is an attractive noninvasive approach for liver irradiation, it presents specific challenges associated with respiration-induced liver motion, daily tumor localization due to liver deformation, and poor visualization of target with respect to adjacent normal liver in computed tomography (CT). We aim to identify potential hazards and develop a set of mitigation strategies to improve the safety of our liver SBRT program, using failure mode and effect analysis (FMEA). Materials and methods A multidisciplinary group consisting of two physicians, three physicists, two dosimetrists, and two therapists was formed. A process map covering ten major stages of the liver SBRT program from the initial diagnosis to posttreatment follow-up was generated. A total of 102 failure modes (FM), together with their causes and effects, were identified. The occurrence (O), severity (S), and lack of detectability (D) were independently scored using a scale from 1 (lowest risk) to 10 (largest risk). The ranking was done using the risk probability number (RPN) defined as the product of average O, S, and D numbers for each mode. Two fault tree analyses were performed. The failure modes with the highest RPN values as well as highest severity score were considered for investigation and a set of mitigation strategies was developed to address these. Results The median RPN (RPNmed ) values for all modes ranged from of 9 to 105 and the highest median S score (Smed ) was 8. Fourteen FMs were identified to be significant by both RPNmed and Smed (top ten RPNmed ranked and highest Smed FMs) and 12 of them were considered for risk mitigation efforts. The remaining two were omitted due to either sufficient checks already in place, or lack of practical mitigation strategies. Implemented measures consisted of five physics tasks, two physician tasks, and three workflow changes. Conclusions The application of FMEA to our liver SBRT program led to the identification of potential FMs and allowed improvement measures to enhance the safety of our clinical practice.

A shallow neural network was trained to accurately calculate the microdosimetric parameters, 〈z 1〉 and 〈z 1 2〉 (the first and second moments of the single-event specific energy spectra, respectively) for use in alpha-particle microdosimetry calculations. The regression network of four inputs and two outputs was created in MATLAB and trained on a data set consisting of both previously published microdosimetric data and recent Monte Carlo simulations. The input data consisted of the alpha-particle energies (3.97–8.78 MeV), cell nuclei radii (2–10 μm), cell radii (2.5–20 μm), and eight different source-target configurations. These configurations included both single cells in suspension and cells in geometric clusters. The mean square error (MSE) was used to measure the performance of the network. The sizes of the hidden layers were chosen to minimize MSE without overfitting. The final neural network consisted of two hidden layers with 13 and 20 nodes, respectively, each with tangential sigmoid transfer functions, and was trained on 1932 data points. The overall training/validation resulted in a MSE = 3.71 × 10−7. A separate testing data set included input values that were not seen by the trained network. The final test on 892 separate data points resulted in a MSE = 2.80 × 10−7. The 95th percentile testing data errors were within ±1.4% for 〈z 1〉 outputs and ±2.8% for 〈z 1 2〉 outputs, respectively. Cell survival was also predicted using actual versus neural network generated microdosimetric moments and showed overall agreement within ±3.5%. In summary, this trained neural network can accurately produce microdosimetric parameters used for the study of alpha-particle emitters. The network can be exported and shared for tests on independent data sets and new calculations.

Purpose Our peer-review program previously consisted of weekly chart rounds performed before the end of the first week of treatment. In order to perform peer review before the start of treatment when possible, we implemented daily prospective contouring and planning rounds (CPR). Methods and materials At the time of computed tomography simulation, patients were categorized by the treating physician into 5 treatment groups based on urgency and complexity (ie, standard, urgent, palliative nonemergent, emergent, and special procedures). A scoring system was developed to record the outcome of case presentations, and the results of the CPR case presentations were compared with the time period 2.5 years before CPR implementation, for which peer review was performed retrospectively. Results CPR was implemented on October 1, 2015, and a total of 4759 patients presented for care through May 31, 2018. The majority were in the standard care path (n = 3154; 66.3%). Among the remainder of the charts, 358 (7.5%), 430 (9.0%), and 179 (3.8%) cases were in the urgent, nonemergent palliative, and emergent care paths, respectively. The remaining patients were in the special procedures group, representing brachytherapy and stereotactic radiosurgery. A total of 125 patients (2.6%) required major changes and were re-presented after the suggested modifications, 102 patients (2.1%) had minor recommendations that did not require a repeat presentation, and 247 cases (5.2%) had minor documentation-related recommendations that did not require editing of the contours. In the 2.5 years before the implementation, records of a total of 1623 patients were reviewed, and only 9 patients (0.6%) had minor recommendation for change. The remainder was noted as complete agreement. Conclusions Contouring and planning rounds were successfully implemented at our clinic. Pretreatment and, most often, preplanning review of contours and directives allows for a more detailed review and changes to be made early on in the treatment planning process. When compared with historical case presentations, the CPR method made our peer review more thorough and improved standardization.

PURPOSE/OBJECTIVE(S) Dual energy (DE) fluoroscopy is being considered for markerless tumor tracking (MTT) of lung tumors. The principal advantage of this technique is that it can be used to remove overlaying bony anatomy, and hence improve the tracking accuracy vs. single energy (SE) fluoroscopy. Previous studies using phantoms have demonstrated the superiority of DE vs. SE imaging for MTT. The goal of this study is to clinically evaluate MTT with fast-kV switching DE imaging on a cohort of SBRT lung cancer patients using the on-board imager (OBI) of a commercial linear accelerator. MATERIALS/METHODS In this Institutional Review Board (IRB)-approved prospective study, DE images were acquired for 10 SBRT lung cancer patients. For each patient, DE images were obtained over 180o arc using fast-kV switching implemented on the OBI of a commercial linear accelerator. This fast-kV switching technique produced x-ray pulses that alternate between programmed tube voltages (i.e., 120 and 60 kVp) at a rate of 15 Hz. Weighted logarithmic subtraction was performed offline on consecutive high-low energy projections to produce soft tissue images. For comparison, SE image sequences were obtained by using the 120 kVp images. Separately, a template was derived from the contoured gross tumor volume (GTV) on the CT simulation scan. A template-based matching algorithm was then used to track target motion on both DE and SE image sequences. Successful matching was defined as any instance in which the algorithm matched the template on the respective fluoroscopic image. Ground truth positions were estimated from both the DE and SE images using Bayesian inference. RESULTS A total of 2278 SE and DE image frames were analyzed. DE imaging resulted in significantly improved tracking vs. SE on 884/2278 (38.8%) of image frames. In those frames, the average tracking error was 1.78 +/- 2.02 mm vs. 2.21 +/- 2.38 mm, for DE vs. SE imaging, respectively (P 0.7 mm. No correlation was observed between tumor volume/location and improved tracking accuracy with DE imaging. Additionally, the fraction of images that could not be tracked with SE was 91/2278 (4.0%). The addition of DE imaging decreased this number to 33/2278 (1.4%) (P < 0.001). CONCLUSION This is the first prospective study to evaluate fast-kV switching DE imaging for MTT in a cohort of SBRT lung cancer patients. This study has demonstrated that by removal of overlapping bony anatomy, DE imaging increases tracking accuracy and decreases the number of images where tracking fails. Future work includes optimization of template parameters to further improve DE tracking accuracy.

Purpose/Objective(s) Symptomatic radiation pneumonitis (RP) is a common cause of treatment related toxicity in patients receiving radiation therapy for lung cancer. Machine learning techniques have been utilized to determine predictors of RP, but most of these models have used dosimetric features to generate results. The identification of pretreatment factors related to the development of symptomatic RP and implementation of these factors into a machine learning model would be clinically useful to guide treatment decisions, risk estimation, and risk reduction of development of symptomatic RP. Materials/Methods We retrospectively reviewed data from 278 lung cancer patients that underwent stereotactic body radiation therapy (SBRT). Pretreatment factors including patient demographics, pulmonary function data, and tumor characteristics were previously analyzed for the development of symptomatic RP, defined as CTCAE v4.0 ≥ Grade 2. A programming environment was used to generate classification machine learning models based on this dataset. Models were tested for accuracy in classification, area-under-the-curve (AUC), and special attention was paid to the false negative rate. The performance of several different algorithms using various combinations of predictors was evaluated including Decision Trees, Discriminant Analysis, Logistic Regression, Naive Bayes Classifiers, Support Vector Machines, Nearest Neighbor Classifiers and Ensemble Classifiers. Each model utilized 5-fold cross-validation to prevent overfitting. Results Of the 278 patients receiving SBRT, 42 patients (15.1%) developed symptomatic RP. A RUSBoosted Trees Ensemble Classifier was found to be the most accurate model tested. Features identified as significant risk factors for symptomatic RP in prior univariate and multivariate analyses were found to be the most important predictors in the machine learning models tested. These features included prior radiation treatment to the thorax or lung resection, location of tumor in the right lower lobe, a higher T Stage and age. When this model was used to classify cases of symptomatic RP, it had an AUC of 0.86 and an accuracy of 87%, correctly predicting 36 out of 42 pneumonitis cases. The model had a false negative rate of 14% and false positive rate of 13%. Conclusion A machine learning model was developed to identify pretreatment factors predictive of symptomatic RP. The advantage of this model when compared to others with a similar goal is that it depends solely on pretreatment factors, and therefore could be applied in a clinical setting prior to radiation therapy. Next steps include testing with an independent dataset to confirm the model's accuracy and prospective studies to verify the model's clinical utility.

The goal of this study is to fully automate the treatment planning and delivery process of hippocampal-sparing whole brain irradiation (HS-WBRT) by combining a RapidPlan (RP) knowledge-based planning model and HyperArc (HA) technology. Additionally, this study compares the dosimetric performance of RapidPlan-HyperArc (RP-HA) treatment plans with RP plans and volumetric modulated arc therapy (VMAT) plans. Ten patients previously treated with HS-WBRT using conventional VMAT were re-planned using RP-HA technique and RP model for HS-WBRT. Treatment plans were generated for 30Gy in 3Gy fractions using 6MV photon beam on a TrueBeam linear accelerator (Varian Medical Systems, Palo Alto, CA) equipped with high definition multileaf collimator (HDMLC). Target coverage, homogeneity index (HI), Paddick Conformity index (CI), dose to organs-at-risk (OARs) provided by the 3 planning modalities were compared, and a paired t-test was performed. Total number of monitor units (MU), effective planning time and beam-on-time time were reported and evaluated for each plan. RP-HA plans achieved on average a 4% increase in D98% of PTV, a 26% improvement in HI, a 2.3% increase in CI, when compared to RP plans. Furthermore, RP-HA plans provided on average 11% decrease in D100% of hippocampi when compared to VMAT plans. All RP-HA plans were generated in less than 30 minutes while RP plans took 40 minutes and VMAT plans required on average 9 hours to complete. Regarding beam-on-time time, it was estimated that RP-HA plans take on average 5 minutes to deliver while RP and VMAT plans require 6.5 and 10 minutes, respectively. RP-HA method provides fully automated planning and delivery for HS-WBRT. The auto-generated plans together with automated treatment delivery allow standardization of plan quality, increased efficiency and ultimately improved patient care.

Radiation therapy (RT) of lung tumors requires an accurate and real-time localization of the tumor while the patient is being treated. Tumor motion caused by breathing can impact dose delivery in patients with lung cancer, leading to poor disease management and damage to surrounding normal tissues. A major challenge in tumor tracking using fluoroscopic imaging is to track the tumor while it is occluded by the overlapping bones (ribs and spine). In this work, we propose a series of three modeling strategies combining template matching for initial probability estimation and Kalman filter for sequential measurement updates. The images were acquired using a fast-kV switching real-time fluoroscope utilizing a dynamic thorax motion phantom. We utilized three distinct physics-based state representations to predict the motion of the tumor: Velocity, Acceleration, and Spring Motion. The Kalman filter with Spring Motion physics performed best with the average RMSE rates of 0.44 mm improving template matching alone with RMSE 1.81 cm. Additionally, occlusion and noise were simulated with general improvements from the Spring Motion Kalman filter. The proposed method will enable more accurate and precise lung radiotherapy using existing hardware and workflow. Our future work is focused on the clinical implementation of this method.

We investigated skin dose enhancements of brass mesh bolus (BMB) and a recently developed transparent polymer-gel bolus (PGB) for clinically relevant breast treatment delivery techniques. The dose enhancement of the breast surface with BMB and PGB were compared to that of tissue-equivalent bolus. Three breast treatment plans were generated on CT scans of an anthropomorphic chest phantom: tangential step-and-shoot 3D conformal (3DCRT) planned using Field-in-Field (FiF), tangential sliding-window 3DCRT using Electronic Compensator (EC), and volumetric modulated arc therapy (VMAT). All plans were created using 6 MV photons and a prescription dose (Rx) of 180 cGy per fraction. Skin doses of all 3 plans were measured with radiochromic films, separately delivered in triplicate. Each plan was delivered to the phantom without bolus, and then with BMB (1 or 2 layers; 3 or 10 mm tissue-equivalent), PGB, and Superflab (3, 5, and 10 mm tissue-equivalent). Doses were determined by reading the radiochromic films with a flatbed scanner, and analyzing the images using a calibration curve for each specific batch. For all bolus types and plans, surface doses averaged over the 3 measurements were between 88.4% and 107.4% of Rx. Without bolus, average measured skin doses were between 51.2% and 64.2% of Rx. Skin doses with BMB and PGB were comparable to that with tissue-equivalent bolus. Over all 3 treatment delivery techniques, using BMB resulted in average skin doses of 92.8% and 102.1% for 1- and 2 layers, respectively, and using PGB results in average skin doses of 94.8%, 98.2%, and 99.7% for 3, 5, and 10-mm tissue-equivalent, respectively. The average measured skin doses with BMB and PGB agreed within ± 3% compared to the tissue-equivalent thickness bolus. We concluded that BMB and PGB are clinically equivalent in skin dose enhancement for breast treatment as the 3, 5, and 10 mm tissue-equivalent bolus.

We performed Monte Carlo simulations in order to determine, by means of microdosimetry calculations, tumour control probability (TCP) curves for treatments with

While some institutions deliver multiple fractions per implant for MRI-based planning, it is common for only one fraction to be delivered per implant with CT-based cervical brachytherapy. The purpose of this study was to compare physician costs, hospital costs, and overall costs for cervical cancer patients treated with either CT-based or MRI-based high-dose-rate (HDR) cervical brachytherapy to determine if MRI-based brachytherapy as described can be financially feasible.We identified 40 consecutive patients treated with curative intent cervical brachytherapy. Twenty patients underwent CT-based HDR brachytherapy with five fractions delivered in five implants on nonconsecutive days in an outpatient setting with the first implant placed with a Smit sleeve under general anesthesia. Twenty patients received MRI-based HDR brachytherapy with four fractions delivered in two implants, each with MRI-based planning, performed 1-2 weeks apart with an overnight hospital admission for each implant. We used Medicare reimbursements to assess physician costs, hospital costs, and overall cost.The median cost of MRI-based brachytherapy was $14,248.75 (interquartile range [IQR]: $13,421.32-$15,539.74), making it less costly than CT-based brachytherapy with conscious sedation (i.e., $18,278.85; IQR: $17,323.13-$19,863.03, p0.0001) and CT-based brachytherapy with deep sedation induced by an anesthesiologist (i.e., $27,673.44; IQR: $26,935.14-$29,511.16, p0.0001). CT-based brachytherapy with conscious sedation was more costly than CT-based brachytherapy with deep sedation (p0.001).MRI-based brachytherapy using the described treatment course was less costly than both methods of CT-based brachytherapy. Cost does not need to be a barrier for MRI-based cervical brachytherapy, especially when delivering multiple fractions with the same application.

Purpose Radiation therapy (RT)–specific aspects of a patient’s cancer care commonly are documented and scheduled through a radiation oncology electronic health record (rEHR). However, patients who receive RT also receive multidisciplinary care from providers who use the hospital EHR (hEHR). We created an electronic interface to integrate our hEHR and rEHR to improve communication of the RT aspects of care between our department and the rest of the hospital. The objective of this study was to assess the impact of rEHR and hEHR integration on the accessibility of the RT-specific aspects of patient care to providers. Methods and Materials We performed a preintegration and postintegration survey of 175 staff members at our academic cancer center. Respondents rated the importance and accessibility of several RT encounters and documents on a Likert scale. The Wilcoxon-Mann-Whitney, χ2, and Fisher’s exact tests were used to compare preintegration and postintegration responses. Results There were 32 and 19 responses to the pre- and postintegration surveys, respectively. rEHR items most commonly reported to be at least moderately important were the dates of first treatment (n = 29 [91%]), last treatment (n = 29 [91%]), brachytherapy (n = 22 [69%]), radiosurgery (n = 22 [69%]), and computed tomography simulation (n = 21 [66%]). A drastic improvement was found in most items made visible in the hEHR through the interface. Conclusion By integrating our hEHR and rEHR, we improved the communication of patient care between the RT department and the multidisciplinary team. Institutions should pursue and support integration of the EHRs to improve the quality of care provided to patients with cancer.

Purpose : Adaptive magnetic resonance imaging (MRI)-based brachytherapy results in improved local control and decreased high-grade toxicities compared to historical controls. Incorporating MRI into the workflow of a department can be a major challenge when initiating an MRI-based brachytherapy program. This project aims to describe the goals, challenges, and solutions when initiating an MRI-based cervical cancer brachytherapy program at our institution. Material and methods : We describe the 6-month multi-disciplinary planning phase to initiate an MRI-based brachytherapy program. We describe the specific challenges that were encountered prior to treating our first patient. Results : We describe the solutions that were realized and executed to solve the challenges that we faced to establish our MRI-based brachytherapy program. We emphasize detailed coordination of care, planning, and communication to make the workflow feasible. We detail the imaging and radiation physics solutions to safely deliver MRI-based brachytherapy. The focus of these efforts is always on the delivery of optimal, state of the art patient care and treatment delivery within the context of our available institutional resources. Conclusions : Previous publications have supported a transition to MRI-based brachytherapy, and this can be safely and efficiently accomplished as described in this manuscript.

Purpose To develop a novel method to monitor external anatomical changes in head and neck cancer patients in order to triage possible adaptive radiotherapy needs Methods The presented approach aims to provide information on internal anatomical changes based on variations observed in external anatomy. Setup Cone Beam Computed Tomography (CBCT) images are processed to produce an accurate external contour of the patient's skin. After registering the CBCTs to the reference planning CT, the external contours from each CBCT are transferred to the initial– first week – CBCT. Contour radii, defined as the distances between an external contour and the isocenter projection in each CBCT slice, are calculated for each scan over the full 360-degrees. The changes in external anatomy are then quantified by the difference in radial distance between the external contours of any secondary CBCT relative to the initial CBCT. Finally, the radial difference is displayed in cylindrical coordinates as a 2D intensity map to highlight regions of interests with significant changes. Weekly CBCT scans from 15 head and neck patients were retrospectively analyzed to demonstrate the utility of this approach as a proof of principle. External changes suggested by the 2D radial difference map of an example patient after 23 fractions were then correlated with the changes in the gross tumor volumes and organs at risks. The resulting dosimetric effects were evaluated. An interactive standalone software application has been developed to facilitate the generation and the interpretation of the 2D intensity map Results The 2D radial difference maps provided qualitative and quantitative information, such as the location and the magnitude of external contour changes and the rate at which these deviations occur. Out of the 15 patients, 10 presented clear evidence of general external volume shrinkage due to weight loss, and nine patients had at least one site of local shrinkage. Only two patients showed no signs of anatomical change during their entire treatment course. For the example patient, the mean (±σ) radial difference was 6.7 (±3.0) mm for the left parotid and 7.3 (±2.5) mm for the right parotid. The mean dose to the left and right parotids increased from 20.1 Gy to 30 Gy and from 16.3 Gy to 29.6 Gy respectively Conclusion This novel method provides an efficient tool to visualize 3D external anatomical changes on a single 2D map. It quickly pinpoints the location of differences in anatomy during the course of radiotherapy, which can help physicians determine if a treatment plan needs to be adapted. The interactive graphic user interface developed in this study will be evaluated in an adaptive radiotherapy workflow for head and neck patients in a future prospective trial. This article is protected by copyright. All rights reserved.

PURPOSE: 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. METHODS: 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 relative electron density (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. RESULTS: 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. Relative electron density values measured for our method resulted in a mean percentage error of 0.0 ± 1.8%. CONCLUSIONS: 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.

PURPOSE To present a novel method, based on convolutional neural networks (CNN), to automate weighted log subtraction (WLS) for dual-energy (DE) fluoroscopy to be used in conjunction with markerless tumor tracking (MTT). METHODS A CNN was developed to automate WLS (aWLS) of DE fluoroscopy to enhance soft tissue visibility. Briefly, this algorithm consists of two phases: training a CNN architecture to predict pixel-wise weighting factors followed by application of WLS subtraction to reduce anatomical noise. To train the CNN, a custom phantom was built consisting of aluminum (Al) and acrylic (PMMA) step wedges. Per-pixel ground truth (GT) weighting factors were calculated by minimizing the contrast of Al in the step wedge phantom to train the CNN. The pretrained model was then utilized to predict pixel-wise weighting factors for use in WLS. For comparison, the weighting factor was manually determined in each projection (mWLS). A thorax phantom with five simulated spherical targets (5-25 mm) embedded in a lung cavity, was utilized to assess aWLS performance. The phantom was imaged with fast-kV dual-energy (120 and 60 kVp) fluoroscopy using the on-board imager of a commercial linear accelerator. DE images were processed offline to produce soft tissue images using both WLS methods. MTT was compared using soft tissue images produced with both mWLS and aWLS techniques. RESULTS Qualitative evaluation demonstrated that both methods achieved soft tissue images with similar quality. The use of aWLS increased the number of tracked frames by 1-5% compared to mWLS, with the largest increase observed for the smallest simulated tumors. The tracking errors for both methods produced agreement to within 0.1 mm. CONCLUSIONS A novel method to perform automated WLS for DE fluoroscopy was developed. Having similar soft tissue quality as well as bone suppression capability as mWLS, this method allows for real-time processing of DE images for MTT.

Dual-energy (DE) imaging using planar imaging with an on-board imager (OBI) is being considered in radiotherapy. We describe here a custom phantom designed to optimize DE imaging parameters using the OBI of a commercial linear accelerator. The phantom was constructed of lung-, tissue- and bone-equivalent material slabs. Five simulated tumors located at two different depths were encased in the lung-equivalent materials. Two slabs with bone-equivalent material inserts were constructed to simulate ribs, which overlap the simulated tumors. DE bone suppression was performed using a weighted logarithmic subtraction based on an iterative method that minimized the contrast between simulated bone- and lung-equivalent materials. The phantom was subsequently used to evaluate different combinations of high-low kV x-ray pairs of images based on the signal-difference-to-noise ratio (SDNR) metric. The results show a strong correlation between tumor visibility and selected energy pairs, where higher energy separation leads to larger SDNR values. To evaluate the effect of image post-processing methods on tumor visibility, an anti-correlated noise reduction (ACNR) technique and adaptive kernel scatter correction method were applied to subsequent DE images. Application of the ACNR technique approximately doubled the SDNR values, hence increasing tumor visibility, while scatter correction had little effect on SDNR values. This phantom allows for quick image acquisition and optimization of imaging parameters and weighting factors. Optimized DE imaging increases soft tissue visibility and may allow for markerless motion tracking of lung tumors.

Objective: To investigate the effects of adaptive radiotherapy on dosimetric, clinical, and toxicity outcomes for patients with head and neck cancer undergoing chemoradiotherapy with intensity-modulated radiotherapy. Methods: Fifty-one patients with advanced head and neck cancer underwent definitive chemoradiotherapy with the original plan optimized to deliver 70.2 Gy. All patients were resimulated at a median dose of 37.8 Gy (range, 27.0-48.6 Gy) due to changes in tumor volume and/or patient weight loss (>15% from baseline). Thirty-four patients underwent adaptive replanning for their boost planning (21.6 Gy). The dosimetric effects of the adaptive plan were compared to the original plan and the original plan copied on rescan computed tomography. Acute and late toxicities and tumor local control were assessed. Gross tumor volume reduction rate was calculated. Results: With adaptive replanning, the maximum dose to the spinal cord, brain stem, mean ipsilateral, and contralateral parotid had a median reduction of −4.5%, −3.0%, −6.2%, and −2.5%, respectively (median of 34 patients). Median gross tumor volume and boost planning target volume coverage improved by 0.8% and 0.5%, respectively. With a median follow-up time of 17.6 months, median disease-free survival and overall survival was 14.8 and 21.1 months, respectively. Median tumor volume reduction rate was 35.2%. For patients with tumor volume reduction rate ≤35.2%, median disease-free survival was 8.7 months, whereas it was 16.9 months for tumor volume reduction rate >35.2%. Four patients had residual disease after chemoradiotherapy, whereas 64.7% (20 of 34) of patients achieved locoregional control. Conclusion: Implementation of adaptive radiotherapy in head and neck cancer offers benefits including improvement in tumor coverage and decrease in dose to organs at risk. The tumor volume reduction rate during treatment was significantly correlated with disease-free survival and overall survival.

Purpose: To evaluate a method for reducing metal artifacts, arising from dental fillings, on cone-beam computed tomography images. Materials and Methods: A projection interpolation algorithm is applied to cone-beam computed tomography images containing metal artifacts from dental fillings. This technique involves identifying metal regions in individual cone-beam computed tomography projections and interpolating the surrounding values to remove the metal from the projection data. Axial cone-beam computed tomography images are then reconstructed, resulting in a reduction in the streak artifacts produced by the metal. Both phantom and patient imaging data are used to evaluate this technique. Results: The interpolation substitution technique successfully reduced metal artifacts in all cases. Corrected images had fewer or no streak artifacts compared to their noncorrected counterparts. Quantitatively, regions of interest containing the artifacts showed reduced variance in the corrected images versus the uncorrected images. Average pixel values in regions of interest around the metal object were also closer in value to nonmetal regions after artifact reduction. Artifact correction tended to perform better on patient images with less complex metal objects versus those with multiple large dental fillings. Conclusion: The interpolation substitution is potentially an efficient and effective technique for reducing metal artifacts caused by dental fillings on cone-beam computed tomography image. This technique may be effective in reducing such artifacts in patients with head and neck cancer receiving daily image-guided radiotherapy.

We performed Monte Carlo simulations in order to determine, by means of microdosimetry calculations, tumour control probability (TCP) curves for treatments with 225Ac-PSMA of metastatic castration resistant prostate cancer (mCRPC). Realistic values of cell radiosensitivity, nucleus size and lesion size were used for calculations. As the cell radiosensitivity decreased, the nucleus size decreased and the lesion size increased, the absorbed dose to reach a given TCP increased. The widest variations occurred with regard to the cell radiosensitivity. For the Monte Carlo simulations, in order to address a non-uniform PSMA expression, different 225Ac-PSMA distributions were considered. The effect of these different PSMA distributions resulted in small variations in the TCP curves (maximum variation of 5%). Absorbed doses to reach a TCP of 0.9 for a uniform 225Ac-PSMA distribution, considering a relative biological effectiveness (RBE) of 5, ranged between 35.0 Gy and 116.5 Gy. The lesion absorbed doses per administered activity reported in a study on treatments with 225Ac-PSMA of mCRPC ranged between 1.3 Gy MBq−1 and 9.8 Gy MBq−1 for a RBE = 5. For a 70 kg-patient to whom 100 kBq kg−1 of 225Ac-PSMA are administered, the range of lesion absorbed doses would be between 9.1 Gy and 68.6 Gy. Thus, for a single cycle of 100 kBq kg−1, a number of lesions would not receive an absorbed dose high enough to reach a TCP of 0.9.

Template-based matching algorithms are currently being considered for markerless motion tracking of lung tumors. These algorithms use tumor templates derived from the planning CT scan, and track the motion of the tumor on single energy fluoroscopic images obtained at the time of treatment. In cases where bone may obstruct the view of the tumor, dual energy fluoroscopy may be used to enhance soft tissue contrast. The goal of this study is to predict which tumors will have a high degree of accuracy for markerless motion tracking based on radiomic features obtained from the planning CT scan, using peak-to-sidelobe ratio (PSR) as a surrogate of tracking accuracy. In this study, CT imaging data of 8 lung cancer patients were obtained and analyzed through the open source IBEX program to generate 2,287 radiomic features. Agglomerative hierarchical clustering was used to narrow down these features into 145 clusters comprised of the highest correlation to PSR. The features among the clusters with the least inter-correlation were then chosen to limit redundancy in the data. The results of this study demonstrated a number of radiomic features that are positively correlated to PSR. The features with the highest degree of correlation included complexity, orientation and range. This approach may be used to determine patients for whom markerless motion tracking would be beneficial.

We performed Monte Carlo simulations in order to determine by means of microdosimetry calculations the average number of hits to the cell nucleus required to reach a tumour control probability (TCP) of 0.9, [Formula: see text], for the source geometry of a nucleus embedded in a homogeneous distribution of

Purpose We transitioned from a low-dose-rate (LDR) to a high-dose-rate (HDR) prostate brachytherapy program. The objective of this study was to describe our experience developing a prostate HDR program, compare the LDR and HDR dosimetry, and identify the impact of several targeted interventions in the HDR workflow to improve efficiency. Methods and Materials We performed a retrospective cohort study of patients treated with LDR or HDR prostate brachytherapy. We used iodine-125 seeds (145 Gy as monotherapy, and 110 Gy as a boost) and preoperative planning for LDR. For HDR, we used iridium-192 (13.5 Gy × 2 as monotherapy and 15 Gy × 1 as a boost) and computed tomography–based planning. Over the first 18 months, we implemented several targeted interventions into our HDR workflow to improve efficiency. To evaluate the progress of the HDR program, we used linear mixed-effects models to compare LDR and HDR dosimetry and identify changes in the implant procedure and treatment planning durations over time. Results The study cohort consisted of 122 patients (51 who received LDR and 71 HDR). The mean D90 was similar between patients who received LDR and HDR (P = .28). HDR mean V100 and V95 were higher (P < .0001), but mean V200 and V150 were lower (P < .0001). HDR rectum V100 and D1cc were lower (P < .0001). The HDR mean for the implant procedure duration was shorter (54 vs 60 minutes; P = .02). The HDR mean for the treatment planning duration dramatically improved with the implementation of targeted workflow interventions (3.7 hours for the first quartile to 2.0 hours for the final quartile; P < .0001). Conclusions We successfully developed a prostate HDR brachytherapy program at our institution with comparable dosimetry to our historic LDR patients. We identified several targeted interventions that improved the efficiency of treatment planning. Our experience and workflow interventions may help other institutions develop similar HDR programs.

Purpose/Objective(s) To evaluate dosimetric differences between MRI-only and MRI-CT planning with a titanium tandem and ovoid applicator to determine if all imaging and planning goals can be achieved with MRI only. Materials/Methods We evaluated 10 patients who underwent MRI-CT-based cervical brachytherapy with a titanium tandem and ovoid applicator. High-risk clinical target volume and organs at risk were contoured on the 3D T2 MRI, which were transferred to the co-registered CT, where the applicator was identified. Retrospectively, three planners independently delineated the applicator on the axial 3D T2 MRI while blinded to the CT. Identical dwell position times in the delivered plan were loaded. Dose-volume histogram parameters were compared to the previously delivered MRI-CT plan. Results There were no significant differences in dose to D90 or D98 of the high-risk clinical target volume with MRI vs. MRI-CT planning. MRI vs. MRI-CT planning resulted in mean D0.1cc bladder of 8.8 ± 3.4 Gy vs. 8.5 ± 3.2 Gy (p = 0.29) and D2cc bladder of 6.2 ± 1.4 Gy vs. 6.0 ± 1.4 Gy (p = 0.33), respectively. Mean D0.1cc rectum was 5.7 ± 1.2 Gy vs. 5.3 ± 1.2 Gy (p = 0.03) and D2cc rectum 4.0 ± 0.8 Gy vs. 4.2 ± 1.0 Gy (p = 0.18), respectively. Mean D0.1cc sigmoid was 5.2 ± 1.3 Gy vs. 5.4 ± 1.6 Gy (p = 0.23) and D2cc sigmoid 3.9 ± 1.0 Gy vs. 4.0 ± 1.1 Gy (p = 0.18), respectively. Conclusion There were no clinically significant dosimetric differences between the MRI and MRI-CT plans. This study demonstrates that cervical brachytherapy with a titanium applicator can be planned with MRI alone, which is now our clinical standard.

Purpose: Dual-energy (DE) imaging is a method that suppresses bony anatomy on planar kV images while enhancing soft tissue contrast. This technique has been used specifically in the chest to improve the visualization of small lung tumors. However, DE imaging may also provide quantitative information that has not been previously investigated. In this study, the aim was to establish a theoretical relationship between DE image contrast and tumor thickness and to observe this trend in phantom experiments. Methods: A phantom consisting cork (used to simulate lung), tissue-equivalent material, and pork ribs was constructed to test for a relationship between DE image contrast and simulated tumor thickness. Fifteen phantom setups were used with various thicknesses of cork and tissue-equivalent material. For each setup, high (120 kVp) and low (60 kVp) energy planar images were acquired and DE images were produced. The image contrast between the simulated tumor and surrounding tissue was then plotted against the known thicknesses and a linear regression was performed. Results: A linear regression of the contrast data vs simulated tumor thickness resulted in a slope of −0.0454 with an R 2 = 0.9904. The expected uncertainty in the thickness measurements using the regression parameters and DE contrast standard deviation was 0.13 cm. Conclusions: Phantom data exhibited a linear relationship between DE image contrast and simulated tumor thickness. Future studies will investigate patient-specific parameters so that this method can be used clinically to evaluate tumor thickness from planar kV images. Such an approach may have benefits for both adaptive and heavy ion therapies.

Purpose: A multi-institutional planning study was performed to evaluate the frequency that current guidelines established by Radiation Therapy Oncology Group (RTOG) protocols and other literature for lung stereotactic body radiotherapy (SBRT) treatments are followed. Methods: A total of 300 patients receiving lung SBRT treatments in four different institutions were retrospectively reviewed. The treatments were delivered using Linac based SBRT (160 patients) or image guided robotic radiosurgery (140). Most tumors were located peripherally (250/300). Median fractional doses and ranges were 18 Gy (8–20 Gy), 12 Gy (6–15 Gy), and 10 Gy (5–12 Gy) for three, four, and five fraction treatments, respectively. The following planning criteria derived from RTOG trials and the literature were used to evaluate the treatment plans: planning target volumes, PTVV 100 ≥ 95% and PTVV 95 ≥ 99%; conformality indices, CI100% < 1.2 and CI50% range of 2.9–5.9 dependent on PTV; total lung-ITV: V 20Gy < 10%, V 12.5Gy < 15%, and V 5Gy < 37%; contralateral lung V 5Gy < 26%; and maximum doses for spinal cord, esophagus, trachea/bronchus, and heart and great vessels. Populations were grouped by number of fractions, and dosimetric criteria satisfaction rates (CSRs) were reported. Results: Five fraction regimens were the most common lung SBRT fractionation (46%). The median PTV was 27.2 cm3 (range: 3.8–419.5 cm3). For all plans: mean PTVV 100 was 94.5% (±5.6%, planning CSR: 69.8%), mean PTVV 95 was 98.1% (±4.1%, CSR: 69.5%), mean CI100% was 1.14 (±0.21, CSR: 79.1%, and 16.5% within minor deviation), and mean CI50% was 5.63 (±2.8, CSR: 33.0%, and 28.0% within minor deviation). When comparing plans based on location, peripherally located tumors displayed higher PTVV 100 and PTVV 95 CSR (71.5% and 71.9%, respectively) than centrally located tumors (61.2% and 57.1%, respectively). Overall, the planning criteria were met for all the critical structure such as lung, heart, spinal cord, esophagus, and trachea/bronchus for at least 85% of the patients. Conclusions: Among the various parameters that were used to evaluate the SBRT plans, the CI100% and CI50% were the most challenging criteria to meet. Although the CSRs of organs at risk were higher among all cases, their proximity to the PTV was a significant factor.

Objective: To develop decision trees predicting for tumor volume reduction in patients with head and neck (H&N) cancer using pretreatment clinical and pathological parameters. Methods: Forty-eight patients treated with definitive concurrent chemoradiotherapy for squamous cell carcinoma of the nasopharynx, oropharynx, oral cavity, or hypopharynx were retrospectively analyzed. These patients were rescanned at a median dose of 37.8 Gy and replanned to account for anatomical changes. The percentages of gross tumor volume (GTV) change from initial to rescan computed tomography (CT; %GTVΔ) were calculated. Two decision trees were generated to correlate %GTVΔ in primary and nodal volumes with 14 characteristics including age, gender, Karnofsky performance status (KPS), site, human papilloma virus (HPV) status, tumor grade, primary tumor growth pattern (endophytic/exophytic), tumor/nodal/group stages, chemotherapy regimen, and primary, nodal, and total GTV volumes in the initial CT scan. The C4.5 Decision Tree induction algorithm was implemented. Results: The median %GTVΔ for primary, nodal, and total GTVs was 26.8%, 43.0%, and 31.2%, respectively. Type of chemotherapy, age, primary tumor growth pattern, site, KPS, and HPV status were the most predictive parameters for primary %GTVΔ decision tree, whereas for nodal %GTVΔ, KPS, site, age, primary tumor growth pattern, initial primary GTV, and total GTV volumes were predictive. Both decision trees had an accuracy of 88%. Conclusions: There can be significant changes in primary and nodal tumor volumes during the course of H&N chemoradiotherapy. Considering the proposed decision trees, radiation oncologists can select patients predicted to have high %GTVΔ, who would theoretically gain the most benefit from adaptive radiotherapy, in order to better use limited clinical resources.

Purpose: We evaluated the performance of organ contour propagation from a planning computed tomography to cone-beam computed tomography with deformable image registration by comparing contours to manual contouring. Materials and Methods: Sixteen patients were retrospectively identified based on showing considerable physical change throughout the course of treatment. Multiple organs in the 3 regions (head and neck, prostate, and pancreas) were evaluated. A cone-beam computed tomography from the end of treatment was registered to the planning computed tomography using rigid registration, followed by deformable image registration. The contours were copied on cone-beam computed tomography image sets using rigid registration and modified by 2 radiation oncologists. Contours were compared using Dice similarity coefficient, mean surface distance, and Hausdorff distance. Results: The mean physician-to-physician Dice similarity coefficient for all organs was 0.90. When compared to each physician’s contours, the overall mean for rigid was 0.76 (P < .001), and it was improved to 0.79 (P < .001) for deformable image registration. Comparing deformable image registration to physicians resulted in a mean Dice similarity coefficient of 0.77, 0.74, and 0.84 for head and neck, prostate, and pancreas groups, respectively; whereas, the physician-to-physician mean agreement for these sites was 0.87, 0.90, and 0.93 (P < .001, for all sites). The mean surface distance for physician-to-physician contours was 1.01 mm, compared to 2.58 mm for rigid-to-physician contours and 2.24 mm for deformable image registration-to-physician contours. The mean physician-to-physician Hausdorff distance was 11.32 mm, and when compared to any physician’s contours, the mean for rigid and deformable image registration was 12.1 mm and 12.0 mm (P < .001), respectively. Conclusion: The physicians had a high level of agreement via the 3 metrics; however, deformable image registration fell short of this level of agreement. The automatic workflows using deformable image registration to deform contours to cone-beam computed tomography to evaluate the changes during treatment should be used with caution.

Purpose The purpose of this study was to report early outcomes and assess the learning curve in a new MRI-based cervical brachytherapy program. Methods We accrued 33 patients prospectively, and only patients with ≥3 months' followup (n = 27) were assessed for disease control and toxicity. Eras were defined as first half and second half for the intracavitary (IC)-only era (n = 13 each), and the intracavitary/interstitial (IC/IS) era was separated by difference in applicator availability (n = 7). Dose to 90% of the high-risk clinical target volume (D90 HR-CTV) and minimum dose to the maximally irradiated 2 cubic centimeters (D2cc) to organs at risk were used to assess dosimetry. Statistics were performed with t tests and Kaplan–Meier method. Results Median followup was 14.7 months. Median treatment duration was 50.5 vs. 57 days for patients treated with external beam radiation therapy at our institution vs. an outside institution (p = 0.03). One-year local control, noncervical pelvic control, distant metastasis–free rate, and overall survival were 84.0%, 96.0%, 78.5%, and 91.3%, respectively. When comparing the first half and second half eras of IC only, there were no differences in median D90 HR-CTV or D2cc of the bladder, rectum, or sigmoid. Comparing the entire IC era to the IC/IS era, median D90 HR-CTV trended higher from 88.0 Gy to 92.9 Gy (p = 0.11). D2cc rectum decreased from 69.3 Gy to 62.6 Gy (p = 0.01), and D2cc bladder trended lower from 87.5 Gy to 83.6 Gy (p = 0.09). Conclusions There was no significant difference between the first half and second half eras with IC-only MRI-based brachytherapy. Incorporation of an IC/IS applicator generated the greatest dosimetric improvement. Early results of the MRI-based brachytherapy program are favorable.

Purpose: To evaluate the use of a Fast Fourier Transform (FFT) based pattern‐matching algorithm for two‐dimensional translational and rotational medical image registration.Methods: The FFT pattern matching algorithm is based on the Fourier shift theorem. Briefly, image registration is accomplished by obtaining the Fourier Transform (FT) of two images, taking the normalized cross‐correlation of the two FT, and performing an inverse FT on this correlation matrix. This results in a Dirchlet delta function that has a maximum value at a location corresponding to the translational shift between the two images. Rotational registration can also be achieved by performing this algorithm on the polar transformation of the FT images. The FT registration method was evaluated through the use of clinical images with induced translational and rotational shifts. Results: Over a range of induced shifts of +/−10 mm in both the x and y directions, and induced rotations of +/−10 degrees, all recovered rotations were within 0.1 degree of the induced rotation, and all recovered translations were within 0.5 mm of the induced translation. The computational time of the FT registration on a 1024×1024 image was approximately 2.23 sec. Conclusions: An FFT based image registration algorithm is computationally efficient and provides a high degree of accuracy for two dimensional image registrations. The FFT registration approach provides a distinct analytical solution and does not rely on iterative methods to converge on a solution. In addition, the discrete nature of the FFT means that the accuracy of the solution is directly related to the size of the pixels in the images. The equivalent of sub‐pixel registration can be achieved by simply resizing the image to a larger matrix (i.e. 512×512 to 1024×1024).

Purpose: To evaluate the efficiency gains obtained from using a Graphics Processing Unit (GPU) to perform a Fourier Transform (FT) based image registration.Methods: Fourier‐based image registration involves obtaining the FT of the component images, and analyzing them in Fourier space to determine the translations and rotations of one image set relative to another. An important property of FT registration is that by enlarging the images (adding additional pixels), one can obtain translations and rotations with sub‐pixel resolution. The expense, however, is an increased computational time. GPUs may decrease the computational time associated with FT image registration by taking advantage of their parallel architecture to perform matrix computations much more efficiently than a Central Processor Unit (CPU). In order to evaluate the computational gains produced by a GPU, images with known translational shifts were utilized. A program was written in the Interactive Data Language (IDL; Exelis, Boulder, CO) to performCPU‐based calculations. Subsequently, the program was modified using GPU bindings (Tech‐X, Boulder, CO) to perform GPU‐based computation on the same system. Multiple image sizes were used, ranging from 256×256 to 2304×2304. The time required to complete the full algorithm by the CPU and GPU were benchmarked and the speed increase was defined as the ratio of the CPU‐to‐GPU computational time. Results: The ratio of the CPU‐to‐ GPU time was greater than 1.0 for all images, which indicates the GPU is performing the algorithm faster than the CPU. The smallest improvement, a 1.21 ratio, was found with the smallest image size of 256×256, and the largest speedup, a 4.25 ratio, was observed with the largest image size of 2304×2304. Conclusions: GPU programming resulted in a significant decrease in computational time associated with a FT image registration algorithm. The inclusion of the GPU may provide near real‐time, sub‐pixel registration capability.

Purpose: To evaluate the efficacy of dual-energy (DE) vs single-energy (SE) fluoroscopic imaging of lung tumors using a markerless template-based tracking algorithm. Methods: Ten representative patient breathing patterns were programmed into a Quasar™ motion phantom. The phantom was modified by affixing pork ribs to the surface, and a cedar insert with a small spherical volume was used to simulate lung and tumor, respectively. Sequential 60 kVp (6 mA) and 120 kVp (1.5 mA) fluoroscopic sequences were acquired. Frame-by-frame weighted logarithmic subtraction was performed resulting in a DE fluoroscopic sequence. A template-based algorithm was then used to track tumor motion throughout the DE and SE fluoroscopy sequences. Tracking coordinates were evaluated against ground-truth tumor locations. Fluoroscopic images were also acquired for two lung cancer patients, neither of which had implanted fiducials. Results: For phantom imaging, a total of 1925 frames were analyzed. The algorithm successfully tracked the target on 99.9% (1923/1925) of DE frames vs 90.7% (1745/1925) SE images (p 5 mm displacements from the ground truth vs 0% for DE fluoroscopy. Conclusions: This work indicates the potential for markerless tumor tracking utilizing DE fluoroscopy. With DE imaging, the algorithm showed improved detectability vs SE fluoroscopy and was able to accurately track the tumor in nearly all cases.