External Surrogate Motion Research Articles

Four-Dimensional Computed Tomography-Based Correlation of Respiratory Motion of Lung Tumors With Implanted Fiducials and an External Surrogate

PurposeOur purpose was to assess the suitability of airway-implanted internal fiducial markers and an external surrogate of respiratory motion for motion management during radiation therapy of lung tumors. Methods and MaterialsWe analyzed 4-dimensional computed tomography scans acquired during radiation therapy simulation for 28 patients with lung tumors who had anchored fiducial markers bronchoscopically implanted inside small airways in or near the tumor in a prospective trial. We used a linear mixed model to build population-based correlative models of tumor and surrogate motion. The first 24 of the 28 patients were used to build correlative models, and 4 of the 28 consecutive patients were excluded and used as an internal validation cohort. Of the 24 patients from the model building cohort, all were used for the models based on the internal fiducial. The external surrogate was completely visualized in 11 patients from the model building cohort, so only those were used for the models based on the external surrogate. Furthermore, we determined the predicted residual error sum of squares for our correlative models, which may serve as benchmarks for future research. ResultsThe motion of the internal fiducials was significantly associated with the tumor motion in the anterior-posterior (P < .0001) and superior-inferior (SI) directions (P < .0001). We also observed a strong correlation of the external surrogate anterior-posterior motion to the tumor dominant SI motion (P < .0001). In the validation cohort, the internal fiducial SI motion was the only reliable predictor of lung tumor motion. ConclusionsThe internal fiducials appear to be more reliable predictors of lung tumor motion than the external surrogate. The suitability of such airway-implanted internal fiducial markers for advanced motion management techniques should be further investigated. Although the external surrogate seems to be less reliable, its wide availability and noninvasive application support its clinical utility, albeit the greater uncertainty will need to be compensated for.

Intrafraction Accuracy and Efficiency of a Surface Imaging System for Deep Inspiration Breath Hold in Ablative Pancreatic Cancer Treatment

Beam gating with deep inspiration breath hold (DIBH) has been widely used for motion management in radiotherapy. It commonly relies on some external surrogate for internal target motion, while the exact internal motion is unknown. In this study, we tracked internal targets and characterized the residual motion during DIBH treatment for pancreatic cancer patients guided by a surface imaging system. We also report the treatment time statistics.Eight consecutive patients with pancreatic cancer treated with surface-guided DIBH volumetrically-modulated arc therapy in 2019-2020 were selected for this study, each with some radiopaque markers (fiducial markers or surgical clips) implanted near or inside the target. Five patients were treated in 25 fractions, two were treated in 15 fractions, and one was treated in 10 fractions. The surface imaging system monitored a three-dimensional surface with gating thresholds of 3 mm translation and 3° rotation. During dose delivery, kV images were automatically taken every 20˚ or 40˚ gantry rotation, from which internal markers were identified. The marker displacement from their initial positions and the residual motion amplitudes were calculated. For the analysis of treatment efficiency, the treatment time of every session was obtained from imaging and treatment timestamps recorded at the treatment console.Even though the external surface was limited to 3 mm / 3°, significant residual internal target motion was observed for some patients. The residual motion ranged 3-21 mm. The average target displacement relative to the initial position was 0-3 mm for the cohort. The magnitude of instantaneous displacement was > 5 mm in 13% of the images analyzed. The average treatment time was 16 min with standard deviation 3 min.Internal target motion may differ significantly from external surrogate motion in DIBH treatment, even with surface image guidance. Radiographic verification of target position at the beginning and during each fraction is essential for accurate delivery of radiotherapy.

Intrafraction tumor motion during deep inspiration breath hold pancreatic cancer treatment.

PurposeBeam gating with deep inspiration breath hold (DIBH) has been widely used for motion management in radiotherapy. Normally it relies on some external surrogate for estimating the internal target motion, while the exact internal motion is unknown. In this study, we used the intrafraction motion review (IMR) application to directly track an internal target and characterized the residual motion during DIBH treatment for pancreatic cancer patients through their full treatment courses.Methods and MaterialsEight patients with pancreatic cancer treated with DIBH volumetric modulated arc therapy in 2017 and 2018 were selected for this study, each with some radiopaque markers (fiducial or surgical clips) implanted near or inside the target. The Varian Real‐time Position Management (RPM) system was used to monitor the breath hold, represented by the anterior‐posterior displacement of an external surrogate, namely reflective markers mounted on a plastic block placed on the patient's abdomen. Before each treatment, a cone beam computed tomography (CBCT) scan under DIBH was acquired for patient setup. For scan and treatment, the breath hold reported by RPM had to lie within a 3 mm window. IMR kV images were taken every 20° or 40° gantry rotation during dose delivery, resulting in over 5000 images for the cohort. The internal markers were manually identified in the IMR images. The residual motion amplitudes of the markers as well as the displacement from their initial positions located in the setup CBCT images were analyzed.ResultsEven though the external markers indicated that the respiratory motion was within 3 mm in DIBH treatment, significant residual internal target motion was observed for some patients. The range of average motion was from 3.4 to 7.9 mm, with standard deviation ranging from 1.2 to 3.5 mm. For all patients, the target residual motions seemed to be random with mean positions around their initial setup positions. Therefore, the absolute target displacement relative to the initial position was small during DIBH treatment, with the mean and the standard deviation 0.6 and 2.9 mm, respectively.ConclusionsInternal target motion may differ from external surrogate motion in DIBH treatment. Radiographic verification of target position at the beginning and during each fraction is necessary for precise RT delivery. IMR can serve as a useful tool to directly monitor the internal target motion.

Impact of sampling interval in training data acquisition on intrafractional predictive accuracy of indirect dynamic tumor-tracking radiotherapy.

To explore the effect of sampling interval of training data acquisition on the intrafractional prediction error of surrogate signal-based dynamic tumor-tracking using a gimbal-mounted linac. Twenty pairs of respiratory motions were acquired from 20 patients (ten lung, five liver, and five pancreatic cancer patients) who underwent dynamic tumor-tracking with the Vero4DRT. First, respiratory motions were acquired as training data for an initial construction of the prediction model before the irradiation. Next, additional respiratory motions were acquired for an update of the prediction model due to the change of the respiratory pattern during the irradiation. The time elapsed prior to the second acquisition of the respiratory motion was 12.6 ± 3.1 min. A four-axis moving phantom reproduced patients' three dimensional (3D) target motions and one dimensional surrogate motions. To predict the future internal target motion from the external surrogate motion, prediction models were constructed by minimizing residual prediction errors for training data acquired at 80 and 320 ms sampling intervals for 20 s, and at 500, 1,000, and 2,000 ms sampling intervals for 60 s using orthogonal kV x-ray imaging systems. The accuracies of prediction models trained with various sampling intervals were estimated based on training data with each sampling interval during the training process. The intrafractional prediction errors for various prediction models were then calculated on intrafractional monitoring images taken for 30 s at the constant sampling interval of a 500 ms fairly to evaluate the prediction accuracy for the same motion pattern. In addition, the first respiratory motion was used for the training and the second respiratory motion was used for the evaluation of the intrafractional prediction errors for the changed respiratory motion to evaluate the robustness of the prediction models. The training error of the prediction model was 1.7 ± 0.7 mm in 3D for all sampling intervals. The intrafractional prediction error for the same motion pattern was 1.9 ± 0.7 mm in 3D for an 80 ms sampling interval, which increased larger than 1 mm in 10.0% of prediction models trained at a 2,000 ms sampling interval with a significant difference (P < 0.01) and up to 2.5% for the other sampling intervals without a significant difference (P > 0.05). The intrafractional prediction error for the changed respiratory motion pattern increased to 5.1 ± 2.4 mm in 3D for an 80 ms sampling interval; however, there was not a significant difference in the robustness of the prediction model between the 80 ms sampling interval and other sampling intervals (P > 0.05). Although the training error of the prediction model was consistent for the all sampling intervals, the prediction model using the larger sampling interval of the 2,000 ms increased the intrafractional prediction error for the same motion pattern. The realistic accuracy of the prediction model was difficult to estimate using the larger sampling interval during the training process. It is recommended to construct the prediction model at sampling interval ≤ 1,000 ms and to reconstruct the model during treatment.

Particle therapy of moving targets-the strategies for tumour motion monitoring and moving targets irradiation.

Particle therapy of moving targets is still a great challenge. The motion of organs situated in the thorax and abdomen strongly affects the precision of proton and carbon ion radiotherapy. The motion is responsible for not only the dislocation of the tumour but also the alterations in the internal density along the beam path, which influence the range of particle beams. Furthermore, in case of pencil beam scanning, there is an interference between the target movement and dynamic beam delivery. This review presents the strategies for tumour motion monitoring and moving target irradiation in the context of hadron therapy. Methods enabling the direct determination of tumour position (fluoroscopic imaging of implanted radio-opaque fiducial markers, electromagnetic detection of inserted transponders and ultrasonic tumour localization systems) are presented. Attention is also drawn to the techniques which use external surrogate motion for an indirect estimation of target displacement during irradiation. The role of respiratory-correlated CT [four-dimensional CT (4DCT)] in the determination of motion pattern prior to the particle treatment is also considered. An essential part of the article is the review of the main approaches to moving target irradiation in hadron therapy: gating, rescanning (repainting), gated rescanning and tumour tracking. The advantages, drawbacks and development trends of these methods are discussed. The new accelerators, called "cyclinacs", are presented, because their application to particle therapy will allow making a breakthrough in the 4D spot scanning treatment of moving organs.

TH-CD-207A-01: Evaluation of Lung Tumor Motion Management Strategy with Dynamic MRI

Purpose: The purpose of this study is to develop a method to track and examine the correlation between the 3D motion of a lung tumor and an external surrogate with dynamic MRI. Methods: Dynamic MRI was obtained from lung cancer patients. To examine the motion correlation between external surrogates and the tumor, we placed four fiducials on the patient's chest at different locations. We acquired a contiguous multi-slice 2D cine MRI (sagittal) to capture the lung and whole tumor, followed by a two-slice 2D cine MRI (sagittal) to simultaneously track the tumor and fiducials. To extract real-time motion, we first reconstructed a phase-binned 4D-MRI from the multi-slice dataset using body area as the respiratory surrogate and a groupwise registration technique. The reconstructed 4D-MRI provided 3D template tumor volumes. Real-time 3D tumor position was calculated by 3D-2D template matching, registering 3D tumor templates and the cine 2D frames from the two-slice tracking dataset. External surrogate 3D trajectories were derived via matching a 3D geometrical model of the fiducial to image features on the 2D cine (two-slice) tracking datasets. Thus, we could analyze the correlation between the 3D trajectories of the tumor and external fiducials. Results: We tested our method on four lung cancer patients. 3D tumor motion correlated with the external surrogate signal, but showed a noticeable phase mismatch. The 3D tumor trajectory showed significant cycle-to-cycle variation, while the external surrogate was not sensitive enough to capture such variations. Additionally, surrogate signals obtained from fiducials at different locations showed noticeable phase mismatch. Conclusion: Our preliminary data show that external surrogate motion has significant variance in relation to tumor motion. Consequently, surrogate-based therapy should be used with caution. Quantitative evaluation of conventional tumor motion management methods such as the internal-target-volume-based approach as well as external-surrogate-based gating, are underway. This work was supported by NIH/NCI under grant R21CA178455.

SU‐G‐JeP4‐08: External and Internal Motion Assessments with Amplitude and Time Based Reconstruction and AveCT

Purpose:To investigate the effects of external surrogate and tumour motion by observing the reconstructed phases and AveCT in an Amplitude and Time based 4DCT.Methods:Based on patient motion studies, Cos6 and sinusoidal motions were simulated as external surrogate and tumour motions in a motion phantom. The diaphragm and tumour motions may or may not display the same waveform therefore the same and different waveforms were programmed into the phantom, scanned and reconstructed based on Amplitude and Time. The AveCT and phases were investigated with these different scenarios. The AveCT phantom images were also compared with CBCT phantom images programmed with the same motions.Results:For the same surrogate and tumour sin motions, the phases (Amplitude and Time) and AveCT indicated similar motions based on the position of the BB at the slice and displayed contrast values respectively. For cos6 motions, due to the varied time the tumour spends at each position, the Amplitude and Time based phases differed. The AveCT images represented the actual tumour motions and the Time and Amplitude based phases were represented by the surrogate with varied times.Conclusion:Different external surrogate and tumour motions may result in different displayed image motions when observing the AveCT and reconstructed phases. During the 4DCT, the surrogate motion is readily available for observation of the amplitude and time of the diaphragm position. Following image reconstruction, the user may need to observe the AveCT in addition to the reconstructed phases to comprehend the time weightings of the tumour motion during the scan. This may also apply to 3D CBCT images where the displayed tumour position in the images is influenced by the long duration of the CBCT. Knowledge of the tumour motion represented by the greyscale of the AveCT may also assist in CBCT treatment beam verification matching.

Real time tracking in liver SBRT: comparison of CyberKnife and Vero by planning structure-based γ-evaluation and dose-area-histograms

The purpose of this study was to evaluate and compare two clinical tracking systems for radiosurgery with regard to their dosimetric and geometrical accuracy in liver SBRT: the robot-based CyberKnife and the gimbal-based Vero. Both systems perform real-time tumour tracking by correlating internal tumour and external surrogate motion. CyberKnife treatment plans were delivered to a high resolution 2D detector array mounted on a 4D motion platform, with the platform simulating (a) tumour motion trajectories extracted from the corresponding CyberKnife predictor log files and (b) the tumour motion trajectories with superimposed baseline-drift.Static reference and tracked dose measurements were compared and dosimetric as well as geometrical uncertainties analyzed by a planning structure-based evaluation. For (a), γ-passing rates inside the CTV (γ-criteria of 1% / 1 mm) ranged from 95% to 100% (CyberKnife) and 98% to 100% (Vero). However, dosimetric accuracy decreases in the presence of the baseline-drift. γ-passing rates for (b) ranged from 26% to 92% and 94% to 99%, respectively; i.e. the effect was more pronounced for CyberKnife. In contrast, the Vero system led to maximum dose deviations in the OAR between +1.5 Gy to +6.0 Gy (CyberKnife: +0.5 Gy to +3.5 Gy). Potential dose shifts were interpreted as motion-induced geometrical tracking errors. Maximum observed shift ranges were −1.0 mm to +0.7 mm (lateral) /−0.6 mm to +0.1 mm (superior–inferior) for CyberKnife and −0.8 mm to +0.2 mm /−0.8 mm to +0.4 mm for Vero. These values illustrate that CyberKnife and Vero provide high precision tracking of regular breathing patterns. Even for the modified motion trajectory, the obtained dose distributions appear to be clinical acceptable with regard to literature QA γ-criteria of 3% / 3 mm.

SU‐E‐J‐186: Using 4DCT‐Based Motion Modeling to Predict Motion and Duty Cycle On Successive Days of Gated Radiotherapy

Purpose:To determine if 4DCT‐based motion modeling and external surrogate motion measured during treatment simulation can enhance prediction of residual tumor motion and duty cycle during treatment delivery.Methods:This experiment was conducted using simultaneously recorded tumor and external surrogate motion acquired over multiple fractions of lung cancer radiotherapy. These breathing traces were combined with the XCAT phantom to simulate CT images. Data from the first day was used to estimate the residual tumor motion and duty cycle both directly from the 4DCT (the current clinical standard), and from external‐surrogate based motion modeling. The accuracy of these estimated residual tumor motions and duty cycles are evaluated by comparing to the measured internal/external motions from other treatment days.Results:All calculations were done for 25% and 50% duty cycles. The results indicated that duty cycle derived from 4DCT information alone is not enough to accurately predict duty cycles during treatment. Residual tumor motion was determined from the recorded data and compared with the estimated residual tumor motion from 4DCT. Relative differences in residual tumor motion varied from −30% to 55%, suggesting that more information is required to properly predict residual tumor motion. Compared to estimations made from 4DCT, in three out of four patients examined, the 30 seconds of motion modeling data was able to predict the duty cycle with better accuracy than 4DCT. No improvement was observed in prediction of residual tumor motion for this dataset.Conclusion:Motion modeling during simulation has the potential to enhance 4DCT and provide more information about target motion, duty cycles, and delivered dose. Based on these four patients, 30 seconds of motion modeling data produced improve duty cycle estimations but showed no measurable improvement in residual tumor motion prediction. More patient data is needed to verify this Result.I would like to acknowledge funding from MRA, VARIAN Medical Systems, Inc.

WE-AB-303-10: The Use of On-Board KV Imaging During Respiratory-Gated VMAT Delivery to Determine the Correlation and Phase Shift Between External Marker Motion and Internal Tumour Motion

Purpose: To determine the correlation between internal tumor motion and external surrogate motion and any potential phase shifts during respiratory-gated VMAT delivery using on-board kV imaging. Methods: A respiratory motion phantom was used to trace the motion of an internal target, embedded with a 3 cm moving target, and an external surrogate. On-board kV projections were acquired at 11 frames per second during VMAT delivery on a TrueBeam v.1.6 (Varian Medical Systems, Palo Alto, CA) operating in developer mode. Full motion encompassed treatment delivery using sinusoidal motion with 4 second period and 2 cm peak-to-peak amplitude, and real-patient breathing (RPB) motion was programmed to the phantom. Respiratory-gated treatment deliveries (30–60% gating window) on five additional RPB waveforms were programmed to investigate correlation within the gating window. To investigate the capability of during treatment kV imaging to detect phase shifts during gated delivery, a second phantom with independent motion was used. Eight controlled phase shifts in intervals of 0.4 seconds were added to the sinusoidal motion. For each phase shift, the Pearson linear correlation coefficient statistical test was performed for each set of respiratory traces. The phase shift was calculated by shifting the external trace, in time, until maximum correlation was obtained. Results: A strong internal and external correlation was obtained for both the free-breathing (sinusoidal R2=0.993 and RPB R2=0.990) and respiratory-gated cases (range from R2=0.986–0.996). There was no significant difference between the programmed shift and that acquired using on-board kV imaging (p=0.899, R-squared = 0.997). Conclusion: During treatment kV imaging has the capability to verify intrafractional anatomical position. It can be accurately used as a tool to quantify internal and external correlation, and determine phase shifts within the gating window, and ultimately, the accuracy of respiratory-gating treatment delivery.

SU‐E‐T‐183: Clinical Quality Assurance Workflow for Dynamic Tumor Tracking Radiation Dose Delivery

Purpose:One of the most important aspects of implementation of new treatment modalities is an ‘end‐to‐end’ verification of the treatment process. Radiation treatment based on dynamic tracking of a tumor is highly patient‐specific, therefore, special attention should be paid to quality assurance of the treatment delivery. Our goal was to design the clinical workflow that ensures accurate delivery of the planned dose using the Dynamic Target Tracking option of VeroTM (BrainLab,MHI) linac.Methods:A patient simulation is designed to include a pre‐treatment session to verify whether the system can reliably track the motion of the implanted marker and build the 4D model of the target motion. The external surrogate and target motion patterns are recorded in the ExactracTM log files. In this work, a spectrum of custom marker and external surrogate motion trajectories closely resembling the patient specific motion patterns was used. 1mm thick/11mm long VisicoilTM marker was placed 15 and 20mm from the center of the spherical tissue equivalent target (centroid to centroid distance) in the 4D motion phantom (CIRSTM). 3D conformal (3 mm block margin) SBRT plans were delivered to 2 moving targets in the phantom: 1) 20mm diameter target that allows ion chamber dose measurement and 2) 25mm target that allows using film to measure CAX dose (GafchromicTM EBT3 used). The measured dose was compared to the iPlanTM TPS results using MonteCarlo algorithm (1% variance, Dose‐to‐water).Results:On average, film shows 98.9% pass using gamma criterion for 2% and 2mm DTA, 94.3% match for 2% and 1 mm DTA, 98% pass for 1% and 2 mm DTA however only 88% points passing for 1% and 1 mm DTA. Ion chamber measurements agreed with the calculation within 1.5%.Conclusion:The clinical QA workflow was designed for SBRT delivery using real‐time tumor tracking on VeroTM linac.

Using an external surrogate for predictor model training in real-time motion management of lung tumors.

Precise prediction of respiratory motion is a prerequisite for real-time motion compensation techniques such as beam, dynamic couch, or dynamic multileaf collimator tracking. Collection of tumor motion data to train the prediction model is required for most algorithms. To avoid exposure of patients to additional dose from imaging during this procedure, the feasibility of training a linear respiratory motion prediction model with an external surrogate signal is investigated and its performance benchmarked against training the model with tumor positions directly. The authors implement a lung tumor motion prediction algorithm based on linear ridge regression that is suitable to overcome system latencies up to about 300 ms. Its performance is investigated on a data set of 91 patient breathing trajectories recorded from fiducial marker tracking during radiotherapy delivery to the lung of ten patients. The expected 3D geometric error is quantified as a function of predictor lookahead time, signal sampling frequency and history vector length. Additionally, adaptive model retraining is evaluated, i.e., repeatedly updating the prediction model after initial training. Training length for this is gradually increased with incoming (internal) data availability. To assess practical feasibility model calculation times as well as various minimum data lengths for retraining are evaluated. Relative performance of model training with external surrogate motion data versus tumor motion data is evaluated. However, an internal-external motion correlation model is not utilized, i.e., prediction is solely driven by internal motion in both cases. Similar prediction performance was achieved for training the model with external surrogate data versus internal (tumor motion) data. Adaptive model retraining can substantially boost performance in the case of external surrogate training while it has little impact for training with internal motion data. A minimum adaptive retraining data length of 8 s and history vector length of 3 s achieve maximal performance. Sampling frequency appears to have little impact on performance confirming previously published work. By using the linear predictor, a relative geometric 3D error reduction of about 50% was achieved (using adaptive retraining, a history vector length of 3 s and with results averaged over all investigated lookahead times and signal sampling frequencies). The absolute mean error could be reduced from (2.0 ± 1.6) mm when using no prediction at all to (0.9 ± 0.8) mm and (1.0 ± 0.9) mm when using the predictor trained with internal tumor motion training data and external surrogate motion training data, respectively (for a typical lookahead time of 250 ms and sampling frequency of 15 Hz). A linear prediction model can reduce latency induced tracking errors by an average of about 50% in real-time image guided radiotherapy systems with system latencies of up to 300 ms. Training a linear model for lung tumor motion prediction with an external surrogate signal alone is feasible and results in similar performance as training with (internal) tumor motion. Particularly for scenarios where motion data are extracted from fluoroscopic imaging with ionizing radiation, this may alleviate the need for additional imaging dose during the collection of model training data.

Improving the intra-fraction update efficiency of a correlation model used for internal motion estimation during real-time tumor tracking for SBRT patients: Fast update or no update?

Background and purposeFor tumor tracking, a correlation model is used to estimate internal tumor position based on external surrogate motion. When patients experience an internal/external surrogate drift, an update of the correlation model is required to continue tumor tracking. In this study, the accuracy of the internal tumor position estimation for both the clinical available update at discrete points in time (rebuild) and an in-house developed non-clinical online update approach was investigated. MethodsA dynamic phantom with superimposed baseline drifts and 14 SBRT patients, treated with real-time tumor tracking (RTTT) on the Vero system, were retrospectively simulated for three update scenarios, respectively no update, clinical rebuild and 0.5Hz automated online update of the correlation model. By comparing the target positions based on 0.5Hz verification X-ray images with the estimated internal tumor positions regarding all three update scenarios, 95th percentile modeling errors (ME95), incidences of full geometrical coverage of the CTV by a 5mm extended PTV (P5mm) and population-based PTV margins were calculated. Further, the treatment time reduction was estimated when switching from the clinical rebuild approach to the online correlation model update. ResultsFor dynamic phantom motion with baseline drifts up to 0.4mm/min, a 0.5Hz intra-fraction update showed a similar accuracy in terms of ME95 and P5mm compared to clinical rebuild. For SBRT patients treated on Vero with RTTT, accuracy was improved by 0.5Hz online update compared to the clinical rebuild protocol, yielding smaller PTV margins (from 3.2mm to 2.7mm), reduced ME95,3D (from 4.1mm to 3.4mm) and an increased 5th percentile P5mm (from 90.7% to 96.1%) for the entire patient group. Further, 80% of treatment sessions were reduced in time with on average 5.5±4.1(1SD)min. ConclusionWith a fast (0.5Hz) automated online update of the correlation model, an efficient RTTT workflow with improved geometrical accuracy was obtained.

Subject-specific real-time respiratory liver motion compensation method for ultrasound-MRI/CT fusion imaging.

Ultrasound-MRI/CT fusion imaging is widely used in minimal invasive surgeries, such as liver biopsy and tumor ablation. However, respiration-induced quasi-periodic liver motion and deformation cause unacceptable misalignment of the fusion images (i.e., fusion error). A subject-specific liver motion model based on skin-mounted position sensor and corresponding ultrasound liver image sequence was developed to compensate for liver motion. External surrogate respiratory motion signal is used to predict internal liver motion. An electromagnetic position sensor fixed on abdominal skin is introduced to track the respiratory motion, and 2D ultrasound images are used to track the liver motion synchronously. Based on these measurements, a subject-specific model describing the relationship of respiratory skin motion and internal liver motion is built and applied in real time (ultrasound-MRI/CT fusion imaging system) to predict and to compensate for the liver motion due to respiratory movement. Feasibility experiments and clinical trials were carried out on a phantom and eight volunteers. Qualitative and quantitative analyses and visual inspections performed by experienced clinicians show that the proposed model could effectively compensate for the liver motion, and the ratio of motion-compensated fusion error to the original varied from 10 % (0.96/9.40 mm) to 28 % (2.90/10.22 mm). An online liver motion modeling and compensation method was developed that provides surgeons with stable and accurate multimodality fusion images in real time.

Tumor tracking based on correlation models in scanned ion beam therapy: an experimental study

Accurate dose delivery to extra-cranial lesions requires tumor motion compensation. An effective compensation can be achieved by real-time tracking of the target position, either measured in fluoroscopy or estimated through correlation models as a function of external surrogate motion. In this work, we integrated two internal/external correlation models (a state space model and an artificial neural network-based model) into a custom infra-red optical tracking system (OTS). Dedicated experiments were designed and conducted at GSI (Helmholtzzentrum für Schwerionenforschung). A robotic breathing phantom was used to reproduce regular and irregular internal target motion as well as external thorax motion. The position of a set of markers placed on the phantom thorax was measured with the OTS and used by the correlation models to infer the internal target position in real-time. Finally, the estimated target position was provided as input for the dynamic steering of a carbon ion beam. Geometric results showed that the correlation models transversal (2D) targeting error was always lower than 1.3 mm (root mean square). A significant decrease of the dosimetric error with respect to the uncompensated irradiation was achieved in four out of six experiments, demonstrating that phase shifts are the most critical irregularity for external/internal correlation models.

MO‐F‐WAB‐05: Predictor Model Training for Real‐Time Motion Management of Lung Tumors

Purpose: Real‐time image guided radiotherapy requires tumor motion prediction to compensate for system specific latencies caused by image acquisition, motion estimation, hardware control and other factors. The performance of prediction models depends strongly on the amount of patient specific data available for training. This may be problematic if additional imaging dose is associated with the training data acquisition. To solve this problem, we investigate the use of chest wall motion data as a surrogate to train a linear predictor for lung tumor motion. Methods: A dataset containing both lung tumor and chest wall motion trajectories for 172 beams/fractions of lung radiotherapy (11 patients) is used. We implement a linear prediction model and estimate the average geometric tracking error as a function of latency. Each trajectory is split into a dataset used for training the predictor model and one used for verification. In a second study we compare two training scenarios for a fixed system latency value of 250 ms. In the first scenario tumor motion is used for training directly, while in the second scenario the external surrogate motion trajectory is used. Verification is performed in both scenarios with the tumor motion trajectory as the reference. Results: For latencies of 0–500ms, the average error without using a predictor increases approximately linearly to 4 mm and at 250 ms we observe 1.8 mm. Using the linear predictor model the error is reduced by an average of about 50%. Training the predictor model with external surrogate data provides similar results. Conclusion: A linear prediction model can reduce latency induced tracking errors by an average of 50% in real‐time image guided radiotherapy systems with system latencies of about 250 ms. External surrogate data is suitable to train the prediction model yielding similar prediction performance as training with lung tumor motion data directly. The project described was supported, in part, by Award Number R21CA156068 from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health.

Evaluation of the geometric accuracy of surrogate‐based gated VMAT using intrafraction kilovoltage x‐ray images

To evaluate the geometric accuracy of beam targeting in external surrogate-based gated volumetric modulated arc therapy (VMAT) using kilovoltage (kV) x-ray images acquired during dose delivery. Gated VMAT treatments were delivered using a Varian TrueBeam STx Linac for both physical phantoms and patients. Multiple gold fiducial markers were implanted near the target. The reference position was created for each implanted marker, representing its correct position at the gating threshold. The gating signal was generated from the RPM system. During the treatment, kV images were acquired immediately before MV beam-on at every breathing cycle, using the on-board imaging system. All implanted markers were detected and their 3D positions were estimated using in-house developed software. The positioning error of a marker is defined as the distance of the marker from its reference position for each frame of the images. The overall error of the system is defined as the average over all markers. For the phantom study, both sinusoidal motion (1D and 3D) and real human respiratory motion was simulated for the target and surrogate. In the baseline case, the two motions were synchronized for the first treatment fraction. To assess the effects of surrogate-target correlation on the geometric accuracy, a phase shift of 5% and 10% between the two motions was introduced. For the patient study, intrafraction kV images of five stereotactic body radiotherapy (SBRT) patients were acquired for one or two fractions. For the phantom study, a high geometric accuracy was achieved in the baseline case (average error: 0.8 mm in the superior-inferior or SI direction). However, the treatment delivery is prone to geometric errors if changes in the target-surrogate relation occur during the treatment: the average error was increased to 2.3 and 4.7 mm for the phase shift of 5% and 10%, respectively. Results obtained with real human respiratory curves show a similar trend. For a target with 3D motion, the technique is able to detect geometric errors in the left-right (LR) and anterior-posterior (AP) directions. For the patient study, the average intrafraction positioning errors are 0.8, 0.9, and 1.4 mm and 95th percentile errors are 1.7, 2.1, and 2.7 mm in the LR, AP, and SI directions, respectively. The correlation between external surrogate and internal target motion is crucial to ensure the geometric accuracy of surrogate-based gating. Real-time guidance based on kV x-ray images overcomes the potential issues in surrogate-based gating and can achieve accurate beam targeting in gated VMAT.

Correlation between External Surrogate Motion and Lung Tumor Motion: From a Statistical Point of View

Correlation between External Surrogate Motion and Lung Tumor Motion: From a Statistical Point of View

Correlation between external and internal respiratory motion: a validation study

In motion-compensated image-guided radiotherapy, accurate tracking of the target region is required. This tracking process includes building a correlation model between external surrogate motion and the motion of the target region. A novel correlation method is presented and compared with the commonly used polynomial model. The CyberKnife system (Accuray, Inc., Sunnyvale/CA) uses a polynomial correlation model to relate externally measured surrogate data (optical fibres on the patient's chest emitting red light) to infrequently acquired internal measurements (X-ray data). A new correlation algorithm based on ɛ -Support Vector Regression (SVR) was developed. Validation and comparison testing were done with human volunteers using live 3D ultrasound and externally measured infrared light-emitting diodes (IR LEDs). Seven data sets (5:03-6:27 min long) were recorded from six volunteers. Polynomial correlation algorithms were compared to the SVR-based algorithm demonstrating an average increase in root mean square (RMS) accuracy of 21.3% (0.4 mm). For three signals, the increase was more than 29% and for one signal as much as 45.6% (corresponding to more than 1.5 mm RMS). Further analysis showed the improvement to be statistically significant. The new SVR-based correlation method outperforms traditional polynomial correlation methods for motion tracking. This method is suitable for clinical implementation and may improve the overall accuracy of targeted radiotherapy.

SU‐FF‐J‐59: Adaptive Gating of Radiotherapy Treatment Based On Marker‐Less Tracking of Lung Tumors with KV Fluoroscopy

Purpose: We investigated using kV fluoroscopy for adapting the respiratory gating window prior to each radiation treatment fraction. The gate adaptation was based on marker‐less tracking of the target and correlating the internal motion trace with the external breathing signal. Method and Materials: A phantom study was performed with the on‐board kV imager of the Novalis Tx machine. A dynamic chest phantom with a water‐filled sphere placed inside a movable lung density insert was utilized. Breathing traces from patients that underwent lung SBRT were used to drive the target and external surrogate motion. The Varian RPM gating system was employed to obtain an external breathing signal, while acquiring kV fluoroscopy images. The moving target was tracked and an internal signal was generated. One kV acquisition was used for adapting the gating window based on the correlation of the internal and external signal and another one for verifying the gated position of the target. Gating at end‐exhale is the most commonly used technique in the clinic, while gating at mid‐ventilation makes it more likely that the tolerance window for residual motion includes the tumor during the entire breathing cycle. The two gating windows were investigated and the results were compared. Results: No significant difference in the gated target position errors was observed between the two gating windows. The target was in the expected position about 90% of the time for each one. The maximum deviation outside the tolerance window for residual motion was 1.9mm for the end‐exhale and 2.2mm for the mid‐motion gating windows. Conclusion: The adapt and verify procedure examined in this study, if applied in real‐time, has the potential to improve the accuracy of lung cancer radiotherapy by facilitating the reduction of safety margins used for moving tumors.Conflict of Interest: Varian Medical Systems, Inc.