Validation of clinical acceptability of deep-learning-based automated segmentation of organs-at-risk for head-and-neck radiotherapy treatment planning.

Introduction: Organ-at-risk segmentation for head and neck cancer radiation therapy is a complex and time-consuming process (requiring up to 42 individual structure, and may delay start of treatment or even limit access to function-preserving care. Feasibility of using a deep learning (DL) based autosegmentation model to reduce contouring time without compromising contour accuracy is assessed through a blinded randomized trial of radiation oncologists (ROs) using retrospective, de-identified patient data. Methods: Two head and neck expert ROs used dedicated time to create gold standard (GS) contours on computed tomography (CT) images. 445 CTs were used to train a custom 3D U-Net DL model covering 42 organs-at-risk, with an additional 20 CTs were held out for the randomized trial. For each held-out patient dataset, one of the eight participant ROs was randomly allocated to review and revise the contours produced by the DL model, while another reviewed contours produced by a medical dosimetry assistant (MDA), both blinded to their origin. Time required for MDAs and ROs to contour was recorded, and the unrevised DL contours, as well as the RO-revised contours by the MDAs and DL model were compared to the GS for that patient. Results: Mean time for initial MDA contouring was 2.3 hours (range 1.6-3.8 hours) and RO-revision took 1.1 hours (range, 0.4-4.4 hours), compared to 0.7 hours (range 0.1-2.0 hours) for the RO-revisions to DL contours. Total time reduced by 76% (95%-Confidence Interval: 65%-88%) and RO-revision time reduced by 35% (95%-CI,-39%-91%). All geometric and dosimetric metrics computed, agreement with GS was equivalent or significantly greater (p<0.05) for RO-revised DL contours compared to the RO-revised MDA contours, including volumetric Dice similarity coefficient (VDSC), surface DSC, added path length, and the 95%-Hausdorff distance. 32 OARs (76%) had mean VDSC greater than 0.8 for the RO-revised DL contours, compared to 20 (48%) for RO-revised MDA contours, and 34 (81%) for the unrevised DL OARs. Conclusion: DL autosegmentation demonstrated significant time-savings for organ-at-risk contouring while improving agreement with the institutional GS, indicating comparable accuracy of DL model. Integration into the clinical practice with a prospective evaluation is currently underway.

Automated testing platform for radiotherapy treatment planning scripts.

Realizing the potential of user-developed automation software interacting with a treatment planning system (TPS) requires rigorous testing to ensure patient safety and data integrity. We developed an automated test platform to allow comparison of the treatment planning database before and after the execution of a write-enabled script interacting with a commercial TPS (Eclipse, Varian Medical Systems, Palo Alto, CA) using the vendor-provided Eclipse Scripting Application Programming Interface (ESAPI). The C#-application known as Write-Enable Script Testing Engine (WESTE) serializes the treatment planning objects (Patient, Structure Set, PlanSetup) accessible through ESAPI, and then compares the serialization acquired before and after the execution of the script being tested, documenting identified differences to highlight the changes made to the treatment planning data. The first two uses of WESTE demonstrated that the testing platform could acquire and analyze the data quickly (<4 s per test case) and facilitate the clinical implementation of write-enabled scripts.

Implementation of free breathing respiratory amplitude-gated treatments.

PURPOSE: The purpose of this study was to provide guidance in developing and implementing a process for the accurate delivery of free breathing respiratory amplitude-gated treatments. METHODS: A phase-based 4DCT scan is acquired at time of simulation and motion is evaluated to determine the exhale phases that minimize respiratory motion to an acceptable level. A phase subset average CT is then generated for treatment planning and a tracking structure is contoured to indicate the location of the target or a suitable surrogate over the planning phases. Prior to treatment delivery, a 4DCBCT is acquired and a phase subset average is created to coincide with the planning phases for an initial match to the planning CT. Fluoroscopic imaging is then used to set amplitude gate thresholds corresponding to when the target or surrogate is in the tracking structure. The final imaging prior to treatment is an amplitude-gated CBCT to verify both the amplitude gate thresholds and patient positioning. An amplitude-gated treatment is then delivered. This technique was commissioned using an in-house lung motion phantom and film measurements of a simple two-field 3D plan. RESULTS: The accuracy of 4DCBCT motion and target position measurements were validated relative to 4DCT imaging. End to end testing showed strong agreement between planned and film measured dose distributions. Robustness to interuser variability and changes in respiratory motion were demonstrated through film measurements. CONCLUSIONS: The developed workflow utilizes 4DCBCT, respiratory-correlated fluoroscopy, and gated CBCT imaging in an efficient and sequential process to ensure the accurate delivery of free breathing respiratory-gated treatments.

2D-3D registration for cranial radiation therapy using a 3D kV CBCT and a single limited field-of-view 2D kV radiograph.

PURPOSE: We present and evaluate a fully automated 2D-3D intensity-based registration framework using a single limited field-of-view (FOV) 2D kV radiograph and a 3D kV CBCT for 3D estimation of patient setup errors during brain radiotherapy. METHODS: We evaluated two similarity measures, the Pearson correlation coefficient on image intensity values (ICC) and maximum likelihood measure with Gaussian noise (MLG), derived from the statistics of transmission images. Pose determination experiments were conducted on 2D kV radiographs in the anterior-posterior (AP) and left lateral (LL) views and 3D kV CBCTs of an anthropomorphic head phantom. In order to minimize radiation exposure and exclude nonrigid structures from the registration, limited FOV 2D kV radiographs were employed. A spatial frequency band useful for the 2D-3D registration was identified from the bone-to-no-bone spectral ratio (BNBSR) of digitally reconstructed radiographs (DRRs) computed from the 3D kV planning CT of the phantom. The images being registered were filtered accordingly prior to computation of the similarity measures. We evaluated the registration accuracy achievable with a single 2D kV radiograph and with the registration results from the AP and LL views combined. We also compared the performance of the 2D-3D registration solutions proposed to that of a commercial 3D-3D registration algorithm, which used the entire skull for the registration. The ground truth was determined from markers affixed to the phantom and visible in the CBCT images. RESULTS: The accuracy of the 2D-3D registration solutions, as quantified by the root mean squared value of the target registration error (TRE) calculated over a radius of 3 cm for all poses tested, was ICCAP : 0.56 mm, MLGAP : 0.74 mm, ICCLL : 0.57 mm, MLGLL : 0.54 mm, ICC (AP and LL combined): 0.19 mm, and MLG (AP and LL combined): 0.21 mm. The accuracy of the 3D-3D registration algorithm was 0.27 mm. There was no significant difference in mean TRE for the 2D-3D registration algorithms using a single 2D kV radiograph with similarity measure and image view point. There was no significant difference in mean TRE between ICCLL , MLGLL , ICC (AP and LL combined), MLG (AP and LL combined), and the 3D-3D registration algorithm despite the smaller FOV used for the 2D-3D registration. While submillimeter registration accuracy was obtained with both ICC and MLG using a single 2D kV radiograph, combining the results from the two projection views resulted in a significantly smaller (P≤0.05) mean TRE. CONCLUSIONS: Our results indicate that it is possible to achieve submillimeter registration accuracy with both ICC and MLG using either single or dual limited FOV 2D kV radiographs of the head in the AP and LL views. The registration accuracy suggests that the 2D-3D registration solutions presented are suitable for the estimation of patient setup errors not only during conventional brain radiation therapy, but also during stereotactic procedures and proton radiation therapy where tighter setup margins are required.

Intravenous contrast-enhanced cone beam computed tomography (IVCBCT) of intrahepatic tumors and vessels.

PURPOSE: Liver tumors are challenging to visualize on cone beam computed tomography (CBCT) without intravenous (IV) contrast. Image guidance for liver cancer stereotactic body ablative radiation therapy (SABR) could be improved with the direct visualization of hepatic tumors and vasculature. This study investigated the feasibility of the use of IV contrast-enhanced CBCT (IV-CBCT) as a means to improve liver target visualization. METHODS AND MATERIALS: Patients on a liver SABR protocol underwent IV-CBCT before 1 or more treatment fractions in addition to a noncontrast CBCT. Image acquisition was initiated 0 to 30 seconds following injection and acquired over 60 to 120 seconds. "Stop and go" exhale breath-hold CBCT scans were used whenever feasible. Changes in mean CT number in regions of interest within visible vasculature, tumor, and adjacent liver were quantified between CBCT and IV-CBCT. RESULTS: Twelve pairs of contrast and noncontrast CBCTs were obtained in 7 patients. Intravenous-CBCT improved hepatic tumor visibility in breath-hold scans only for 3 patients (2 metastases, 1 hepatocellular carcinoma). Visible tumors ranged in volume from 124 to 564 mL. Small tumors in free-breathing patients did not show enhancement on IVCBT. CONCLUSIONS: Intravenous-CBCT may enhance the visibility of hepatic vessels and tumor in CBCT scans obtained during breath hold. Optimization of IV contrast timing and reduction of artifacts to improve tumor visualization warrant further investigation.

Quality assurance of asymmetric jaw alignment using 2D diode array.

PURPOSE: A method using a 2D diode array is proposed to measure the junction gap (or overlap) and dose with high precision for routine quality assurance of the asymmetric jaw alignment. METHODS: The central axis (CAX) of the radiation field was determined with a 15 × 15 cm(2) photon field at four cardinal collimator angles so that the junction gap (or overlap) can be measured with respect to the CAX. Two abutting fields having a field size of 15 cm (length along the axis parallel to the junction) × 7.5 cm (width along the axis perpendicular to the junction) were used to irradiate the 2D diode array (MapCHECK2) with 100 MU delivered at the photon energy of 6 MV. The collimator was slightly rotated at 15° with respect to the beam central axis to increase the number of diodes effective on the measurement of junction gap. The junction gap and dose measured in high spatial resolution were compared to the conventional methods using an electronic portal imaging device (EPID) and radiochromic film, respectively. In addition, the reproducibility and sensitivity of the proposed method to the measurements of junction gap and dose were investigated. RESULTS: The junction gap (or overlap) and dose measured by MapCHECK2 agreed well to those measured by the conventional methods of EPID and film (the differences ranged from -0.01 to 0 cm and from -1.34% to 0.6% for the gap and dose, respectively). No variation in the repeat measurements of the junction gap was found whereas the measurements of junction dose were found to vary in quite a small range over the days of measurement (0.21%-0.35%). While the sensitivity of the measured junction gap to the actual junction gap applied was the ideal value of 1 cm∕cm as expected, the sensitivity of the junction dose to the actual junction gap increased as the junction gap (or overlap) decreased (maximum sensitivity: 201.7%∕cm). CONCLUSIONS: The initial results suggest that the method is applicable for a comprehensive quality assurance of the asymmetric jaw alignment.

Quality assurance for image-guided radiation therapy utilizing CT-based technologies: a report of the AAPM TG-179.

PURPOSE: Commercial CT-based image-guided radiotherapy (IGRT) systems allow widespread management of geometric variations in patient setup and internal organ motion. This document provides consensus recommendations for quality assurance protocols that ensure patient safety and patient treatment fidelity for such systems. METHODS: The AAPM TG-179 reviews clinical implementation and quality assurance aspects for commercially available CT-based IGRT, each with their unique capabilities and underlying physics. The systems described are kilovolt and megavolt cone-beam CT, fan-beam MVCT, and CT-on-rails. A summary of the literature describing current clinical usage is also provided. RESULTS: This report proposes a generic quality assurance program for CT-based IGRT systems in an effort to provide a vendor-independent program for clinical users. Published data from long-term, repeated quality control tests form the basis of the proposed test frequencies and tolerances. CONCLUSION: A program for quality control of CT-based image-guidance systems has been produced, with focus on geometry, image quality, image dose, system operation, and safety. Agreement and clarification with respect to reports from the AAPM TG-101, TG-104, TG-142, and TG-148 has been addressed.

The Impact of Evolving Image-Guidance Processes on Initial Patient Setup for Lung Radiotherapy.

PURPOSE: The purpose of this study was to investigate whether the implementation of three different daily image-guidance processes has altered initial (pre-imaging) patient setup practice for thoracic radiotherapy patients. METHODS: A total of 1 997 daily cone-beam computed tomography (CBCT) images from 72 thoracic patients undergoing radical radiotherapy were retrospectively reviewed under ethics approval. Patients were analyzed in three consecutive cohorts grouped according to the image-guidance process used during radiotherapy. After initial alignment of skin marks and lasers: Process A (24 patients spanning 6 months), CBCT alignment with an action level of 3 mm, correction applied via manual couch adjustment, followed by a verification CBCT; Process B (22 patients, 5 months), CBCT alignment with an action level of 3 mm, correction applied via remote couch adjustment, followed by a verification CBCT; Process C (26 patients, 5 months), CBCT alignment with correction applied for all displacements via remote couch adjustment, with no verification scans required. Initial patient setup displacements from skin marks were determined by re-registering the initial alignment CBCT to the planning CT using automated spine matching. Patient setup displacements were compared between the three processes in the left-right (LR), cranial-caudal (CC), and anterior-posterior (AP) directions. RESULTS: The mean ± 1 standard deviation of initial patient setup displacements were calculated for each cohort: Process A, 1.2 ± 2.4 mm (LR), 0.6 ± 3.5 mm (CC), and -0.8 ± 2.0 mm (AP); Process B, 0.5 ± 2.7 mm (LR), 1.2 ± 3.4 mm (CC) and -1.7 ± 2.0 mm (AP); Process C, 1.0 ± 2.5 mm (LR), 0.1 ± 3.5 mm (CC), and -2.3 ± 2.2 mm (AP). The means systematic and random uncertainties were comparable between the processes, showing similar setup error distributions. CONCLUSION: Initial skin setup practices for thoracic radiotherapy patients remain unaffected across the three image-guidance processes. Pre-imaging alignment principles and performance by radiation therapists at our center remain consistent amid technological advances.

Interfraction and intrafraction changes in amplitude of breathing motion in stereotactic liver radiotherapy.

PURPOSE: Interfraction and intrafraction changes in amplitude of liver motion were assessed in patients with liver cancer treated with kV cone beam computed tomography (CBCT)-guided stereotactic body radiation therapy (SBRT). METHODS AND MATERIALS: A total of 314 CBCTs obtained with the patient in the treatment position immediately before and after each fraction, and 29 planning 4DCTs were evaluated in 29 patients undergoing six-fraction SBRT for unresectable liver cancer, with (n = 15) and without (n = 14) abdominal compression. Offline, the CBCTs were sorted into 10 bins, based on phase of respiration. Liver motion amplitude was measured using liver-to-liver alignment from the end-exhale and end-inhale CBCT and four-dimensional CT reconstructions. Inter- and intrafraction amplitude changes were measured from the difference between the pre-SBRT CBCTs relative to the planning four-dimensional CT, and from the pre-SBRT and post-SBRT CBCTs, respectively. RESULTS: Mean liver motion amplitude for all patients (range) was 1.8 (0.1-7.0), 8.0 (0.1-18.8), and 4.3 (0.1-12.1) mm in the mediolateral (ML), craniocaudal (CC), and anteroposterior (AP) directions, respectively. Mean absolute inter- and intrafraction liver motion amplitude changes were 1.0 (ML), 1.7 (CC), and 1.6 (AP) mm and 1.3 (ML), 1.6 (CC), and 1.9 (AP) mm, respectively. No significant correlations were found between intrafraction amplitude change and intrafraction time (range, 4:56-25:37 min:sec), and between inter- and intrafraction amplitude changes and liver motion amplitude. Intraobserver reproducibility (sigma, n = 29 fractions) was 1.3 (ML), 1.4 (CC), and 1.4 (AP) mm. CONCLUSIONS: For the majority of liver SBRT patients, the change in liver motion amplitude was minimal over the treatment course and showed no apparent relationships with the magnitude of liver motion and intrafraction time.

Investigating User Perspective on Training and Clinical Implementation of Volumetric Imaging.

PURPOSE: Volumetric imaging (VI) is gaining momentum as a clinical tool for image guidance. The purpose of this study was to explore radiation therapists' (RTs) perception of VI training and clinical implementation. The results provide an evaluation of current training for continuous improvement, identify educational needs for the future, and chart a path for higher level evaluation of the learning activities. METHODS: Sixty-five RTs who received initial training and subsequent front-end user experience with VI were contacted with an electronic posttraining questionnaire. Defined with assistance from a pilot group to encompass relevant aspects of training and VI, the survey consisted of 16 questions using a 4-point Likert scale and 1 open-ended comment box. The main sections of the survey were: demographics, current training satisfaction, user perception of competence with VI software manipulation, and assessment of future training needs for RTs. RESULTS: Forty-five returned surveys (69%) were used for analysis. Results were examined for all participants, and then separated by subgroups for comparison with the Fisher exact test. The subgroups used were years of RT experience (<2 years, 2-5 years, 6-10 years, >10 years), and length of hands-on practice with VI (<3 months, 3-6 months, 7-9 months, >9 months). There was a consensus that training was organized (93%), comprehensive (89%), and adequate for VI clinical application (91%). Although 62% of respondents were confident with soft-tissue visualization, 95% agreed with the need for further training. Proposed cross-sectional anatomy review was seen as beneficial in 73%, though analysis with subgroups (experience with VI [P = .042], experience as an RT [P = .019]) showed that less experienced staff perceived a supplemental review session valuable. A total of 93% had confidence using VI for image matching; however, subgroup analysis revealed that the more experienced RTs are less confident (P = .016). Although only 58% agreed that workflow is improved with VI, 93% felt that volumetric imaging increased accuracy of radiotherapy. Common themes for improvement of future training sessions included conducting training closer to VI unit placement, and increased hands-on use of the software complemented case studies. CONCLUSIONS: RTs felt the initial VI image training assisted them in gaining confidence in interpreting and analyzing information obtained from VI. They found this training organized, comprehensive, and adequate for their needs. Continuous training for implementation of new technology is essential to the success of image guidance methods.

Setup reproducibility for thoracic and upper gastrointestinal radiation therapy: Influence of immobilization method and on-line cone-beam CT guidance.

We report the setup reproducibility of thoracic and upper gastrointestinal (UGI) radiotherapy (RT) patients for 2 immobilization methods evaluated through cone-beam computed tomography (CBCT) image guidance, and present planning target volume (PTV) margin calculations made on the basis of these observations. Daily CBCT images from 65 patients immobilized in a chestboard (CB) or evacuated cushion (EC) were registered to the planning CT using automatic bony anatomy registration. The standardized region-of-interest for matching was focused around vertebral bodies adjacent to tumor location. Discrepancies >3 mm between the CBCT and CT datasets were corrected before initiation of RT and verified with a second CBCT to assess residual error (usually taken after 90 s of the initial CBCT). Positional data were analyzed to evaluate the magnitude and frequencies of setup errors before and after correction. The setup distributions were slightly different for the CB (797 scans) and EC (757 scans) methods, and the probability of adjustment at a 3-mm action threshold was not significantly different (p = 0.47). Setup displacements >10 mm in any direction were observed in 10% of CB fractions and 16% of EC fractions (p = 0.0008). Residual error distributions after CBCT guidance were equivalent regardless of immobilization method. Using a published formula, the PTV margins for the CB were L/R, 3.3 mm; S/I, 3.5 mm; and A/P, 4.6 mm), and for EC they were L/R, 3.7 mm; S/I, 3.3 mm; and A/P, 4.6 mm. In the absence of image guidance, the CB slightly outperformed the EC in precision. CBCT allows reduction to a single immobilization system that can be chosen for efficiency, logistics, and cost. Image guidance allows for increased geometric precision and accuracy and supports a corresponding reduction in PTV margin.

Accuracy of automatic couch corrections with on-line volumetric imaging.

The purpose of this study was to characterize automatic remote couch adjustment and to assess the accuracy of automatic couch corrections following localization with cone-beam CT (CBCT). Automatic couch movement was evaluated through passive reflector markers placed on a phantom, tracked with an optical tracking system (OTS). Repeated couch movements in the lateral, cranial/caudal, and vertical directions were monitored through the OTS to assess velocity and response time. In conjunction with CBCT, remote table movement for patient displacements following initial setup was available on four linear accelerators (Elekta Synergy). After the initial CBCT scan assessment, patients with isocenter displacements that exceeded clinical protocol tolerances were corrected using remote automatic couch movement. A verification CBCT scan was acquired after any remote movements. These verification CBCT datasets were assessed for the following time periods: one month post clinical installation, and six months later to monitor remote couch correction stability. Residual error analysis was evaluated using the verification scans. The mean +/- standard deviations (mu +/- sigma) of couch movement based on phantom measurements with the OTS were 0.16 +/- 0.48 mm, 0.32 +/- 0.30 mm, 0.11 +/- 0.12 mm in the L/R, A/P, and S/I couch directions, respectively. The fastest maximum velocity was observed in the inferior direction at 10.5 mm/s, and the slowest maximum velocity in the left direction at 3.6 mm/s. From 1134 verification CBCT registrations for 207 patients, the residual error for each translational direction from each month evaluated are reported. The mu was less than 0.3 mm in all directions, and sigma was in the order of 1 mm. At a 3 mm threshold, 21 of the 1134 fractions (2%) exceeded tolerance, attributed to patient intrafraction movement. Remote automatic couch movement is reliable and effective for adjusting patient position with a precision of approximately 1mm. Patient residual error observed in this study demonstrates that displacement is minimal after remote couch adjustment.

2D-3D registration for prostate radiation therapy based on a statistical model of transmission images.

PURPOSE: In external beam radiation therapy of pelvic sites, patient setup errors can be quantified by registering 2D projection radiographs acquired during treatment to a 3D planning computed tomograph (CT). We present a 2D-3D registration framework based on a statistical model of the intensity values in the two imaging modalities. METHODS: The model assumes that intensity values in projection radiographs are independently but not identically distributed due to the nonstationary nature of photon counting noise. Two probability distributions are considered for the intensity values: Poisson and Gaussian. Using maximum likelihood estimation, two similarity measures, maximum likelihood with a Poisson (MLP) and maximum likelihood with Gaussian (MLG), distribution are derived. Further, we investigate the merit of the model-based registration approach for data obtained with current imaging equipment and doses by comparing the performance of the similarity measures derived to that of the Pearson correlation coefficient (ICC) on accurately collected data of an anthropomorphic phantom of the pelvis and on patient data. RESULTS: Registration accuracy was similar for all three similarity measures and surpassed current clinical requirements of 3 mm for pelvic sites. For pose determination experiments with a kilovoltage (kV) cone-beam CT (CBCT) and kV projection radiographs of the phantom in the anterior-posterior (AP) view, registration accuracies were 0.42 mm (MLP), 0.29 mm (MLG), and 0.29 mm (ICC). For kV CBCT and megavoltage (MV) AP portal images of the same phantom, registration accuracies were 1.15 mm (MLP), 0.90 mm (MLG), and 0.69 mm (ICC). Registration of a kV CT and MV AP portal images of a patient was successful in all instances. CONCLUSIONS: The results indicate that high registration accuracy is achievable with multiple methods including methods that are based on a statistical model of a 3D CT and 2D projection images.

Inter- and intrafraction variability in liver position in non-breath-hold stereotactic body radiotherapy.

PURPOSE: The inter- and intrafraction variability of liver position was assessed in patients with liver cancer treated with kilovoltage cone-beam computed tomography (CBCT)-guided stereotactic body radiotherapy. METHODS AND MATERIALS: A total of 314 CBCT scans obtained in the treatment position immediately before and after each fraction were evaluated from 29 patients undergoing six-fraction, non-breath-hold stereotactic body radiotherapy for unresectable liver cancer. Off-line, the CBCT scans were sorted into 10 bins, according to the phase of respiration. The liver position (relative to the vertebral bodies) was measured using rigid alignment of the exhale CBCT liver with the exhale planning CT liver, following the alignment of the vertebrae. The interfraction liver position change was measured by comparing the pretreatment CBCT scans, and the intrafraction change was measured from the CBCT scans obtained immediately before and after each fraction. RESULTS: The mean amplitude of liver motion for all patients was 1.8 mm (range, 0.1-5.7), 8.0 mm (range, 0.1-18.8), and 4.3 mm (range 0.1-12.1) in the medial-lateral (ML), craniocaudal (CC), and anteroposterior (AP) directions, respectively. The mean absolute ML, CC, and AP interfraction changes in liver position were 2.0 mm (90th percentile, 4.2), 3.5 mm (90th percentile, 7.3), and 2.3 mm (90th percentile, 4.7). The mean absolute intrafraction ML, CC, and AP changes were 1.3 mm (90th percentile, 2.9), 1.6 mm (90th percentile, 3.6), and 1.5 mm (90th percentile, 3.1), respectively. The interfraction changes were significantly larger than the intrafraction changes, with a CC systematic error of 2.9 and 1.1 mm, respectively. The intraobserver reproducibility (sigma, n = 29 fractions) was 1.3 mm in the ML, 1.4 mm in the CC, and 1.6 mm in the AP direction. CONCLUSION: Interfraction liver position changes relative to the vertebral bodies are an important source of geometric uncertainty, providing a rationale for prefraction soft-tissue image guidance. The intrafraction change in liver position from the beginning to the end of each fraction was small for most patients.

Novel dosimetric phantom for quality assurance of volumetric modulated arc therapy.

The objective of this work is to assess the suitability and performance of a new dosimeter system with a novel geometry for the quality assurance (QA) of volumetric modulated arc therapy (VMAT). The new dosimeter system consists of a hollow cylinder (15 and 25 cm inner and outer diameters) with 124 diodes embedded in the phantom's cylindrical wall forming four rings of detectors. For coplanar beams, the cylindrical geometry and the ring diode pattern offer the advantage of invariant perpendicular incidence on the beam central axis for any gantry angle and also have the benefit of increasing the detector density as both walls of the cylinder sample the beam. Other advantages include real-time readout and reduced weight with the hollow phantom shape. A calibration method taking into account the variation in radiation sensitivity of the diodes as a function of gantry angle was developed and implemented. In this work, the new dosimeter system was used in integrating mode to perform composite dose measurements along the cylindrical surface supporting the diodes. The reproducibility of the dosimeter response and the angular dependence of the diodes were assessed using simple 6 MV photon static beams. The performance of the new dosimeter system for VMAT QA was then evaluated using VMAT plans designed for a head and neck, an abdominal sarcoma, and a prostate patient. These plans were optimized with 90 control points (CPs) and additional versions of each plan were generated by increasing the number of CPs to 180 and 360 using linear interpolation. The relative dose measured with the dosimeter system for the VMAT plans was compared to the corresponding TPS dose map in terms of relative dose difference (% deltaD) and distance to agreement (DTA). The dosimeter system's sensitivity to gantry rotation offset and scaling errors as well as setup errors was also evaluated. For static beams, the dosimeter system offered good reproducibility and demonstrated small residual diode angular dependence after calibration. For VMAT deliveries, the agreement between measured and calculated doses was good with > or = 86.4% of the diodes satisfying 3% of % deltaD or 2 mm DTA for the 180 CP plans. The phantom offered sufficient sensitivity for the detection of small gantry rotation offset (3 degrees) and scaling errors (1 degree) as well as phantom setup errors of 1 mm, although the results were plan dependent. With its novel geometry, the dosimeter system was also able to experimentally demonstrate the discretization effect of the number of CPs used in the TPS to simulate a continuous arc. These results demonstrate the suitability of the new dosimeter system for the patient-specific QA of VMAT plans and suggest that the dosimeter system can be an effective tool in the routine QA and commissioning of treatment machines capable of VMAT delivery and cone-beam CT image guidance.

Quantifying interfraction and intrafraction tumor motion in lung stereotactic body radiotherapy using respiration-correlated cone beam computed tomography.

PURPOSE: Stereotactic body radiation therapy (SBRT) is an effective treatment for medically inoperable Stage I non-small-cell lung cancer. However, changes in the patient's breathing patterns during the course of SBRT may result in a geographic miss or an overexposure of healthy tissues to radiation. However, the precise extent of these changes in breathing pattern is not well known. We evaluated the inter- and intrafractional changes in tumor motion amplitude (DeltaM) over an SBRT course. METHODS AND MATERIALS: Eighteen patients received image-guided SBRT delivered in three fractions; this therapy was done with abdominal compression in four patients. For each fraction, cone beam computed tomography (CBCT) was performed for tumor localization (+/- 3-mm tolerance) and then repeated to confirm geometric accuracy. Additional CBCT images were acquired at the midpoint and end of each SBRT fraction. Respiration-correlated CBCT (rcCBCT) reconstructions allowed retrospective assessment of inter- and intrafractional DeltaM by a comparison of tumor displacements in all four-dimensional CT and rcCBCT scans. The DeltaM was measured in mediolateral, superior-inferior, and anterior-posterior directions. RESULTS: A total of 201 rcCBCT images were analyzed. The mean time from localization of the tumor to the end-fraction CBCT was 35 +/- 7 min. Compared with the motion recorded on four-dimensional CT, the mean DeltaM was 0.4, 1.0, and 0.4 mm, respectively, in the mediolateral, superior-inferior, and anterior-posterior directions. On treatment, the observed DeltaM was, on average, <1 mm; no DeltaM was statistically different with respect to the initial rcCBCT. However, patients in whom abdominal compression was used showed a statistically significant difference (p < 0.05) in the variance of DeltaM with respect to the initial rcCBCT in the superior-inferior direction. CONCLUSIONS: The inter- and intrafractional DeltaM that occur during a course of lung SBRT are small. However, abdominal compression causes larger variations in the time spent on the treatment couch and in the inter- and intrafractional DeltaM values.

Automated 2D-3D registration of portal images and CT data using line-segment enhancement.

In prostate radiotherapy, setup errors with respect to the patient's bony anatomy can be reduced by aligning 2D megavoltage (MV) portal images acquired during treatment to a reference 3D kilovoltage (kV) CT acquired for treatment planning purposes. The purpose of this study was to evaluate a fully automated 2D-3D registration algorithm to quantify setup errors in 3D through the alignment of line-enhanced portal images and digitally reconstructed radiographs computed from the CT. The line-enhanced images were obtained by correlating the images with a filter bank of short line segments, or "sticks" at different orientations. The proposed methods were validated on (1) accurately collected gold-standard data consisting of a 3D kV cone-beam CT scan of an anthropomorphic phantom of the pelvis and 2D MV portal images in the anterior-posterior (AP) view acquired at 15 different poses and (2) a conventional 3D kV CT scan and weekly 2D MV AP portal images of a patient over 8 weeks. The mean (and standard deviation) of the absolute registration error for rotations around the right-lateral (RL), inferior-superior (IS), and posterior-anterior (PA) axes were 0.212 degree (0.214 degree), 0.055 degree (0.033 degree) and 0.041 degree (0.039 degree), respectively. The corresponding registration errors for translations along the RL, IS, and PA axes were 0.161 (0.131) mm, 0.096 (0.033) mm, and 0.612 (0.485) mm. The mean (and standard deviation) of the total registration error was 0.778 (0.543) mm. Registration on the patient images was successful in all eight cases as determined visually. The results indicate that it is feasible to automatically enhance features in MV portal images of the pelvis for use within a completely automated 2D-3D registration framework for the accurate determination of patient setup errors. They also indicate that it is feasible to estimate all six transformation parameters from a 3D CT of the pelvis and a single portal image in the AP view.

Statistical process control for IMRT dosimetric verification.

Patient-specific measurements are typically used to validate the dosimetry of intensity-modulated radiotherapy (IMRT). To evaluate the dosimetric performance over time of our IMRT process, we have used statistical process control (SPC) concepts to analyze the measurements from 330 head and neck (H&N) treatment plans. The objectives of the present work are to: (i) Review the dosimetric measurements of a large series of consecutive head and neck treatment plans to better understand appropriate dosimetric tolerances; (ii) analyze the results with SPC to develop action levels for measured discrepancies; (iii) develop estimates for the number of measurements that are required to describe IMRT dosimetry in the clinical setting; and (iv) evaluate with SPC a new beam model in our planning system. H&N IMRT cases were planned with the PINNACLE treatment planning system versions 6.2b or 7.6c (Philips Medical Systems, Madison, WI) and treated on Varian (Palo Alto, CA) or Elekta (Crawley, UK) linacs. As part of regular quality assurance, plans were recalculated on a 20-cm-diam cylindrical phantom, and ion chamber measurements were made in high-dose volumes (the PTV with highest dose) and in low-dose volumes (spinal cord organ-at-risk, OR). Differences between the planned and measured doses were recorded as a percentage of the planned dose. Differences were stable over time. Measurements with PINNACLE3 6.2b and Varian linacs showed a mean difference of 0.6% for PTVs (n=149, range, -4.3% to 6.6%), while OR measurements showed a larger systematic discrepancy (mean 4.5%, range -4.5% to 16.3%) that was due to well-known limitations of the MLC model in the earlier version of the planning system. Measurements with PINNACLE3 7.6c and Varian linacs demonstrated a mean difference of 0.2% for PTVs (n=160, range, -3.0%, to 5.0%) and -1.0% for ORs (range -5.8% to 4.4%). The capability index (ratio of specification range to range of the data) was 1.3 for the PTV data, indicating that almost all measurements were within +/-5%. We have used SPC tools to evaluate a new beam model in our planning system to produce a systematic difference of -0.6% for PTVs and 0.4% for ORs, although the number of measurements is smaller (n=25). Analysis of this large series of H&N IMRT measurements demonstrated that our IMRT dosimetry was stable over time and within accepted tolerances. These data provide useful information for assessing alterations to beam models in the planning system. IMRT is enhanced by the addition of statistical process control to traditional quality control procedures.

Improving image-guided target localization through deformable registration.

PURPOSE: To quantify the improvements in online target localization using kV cone beam CT (CBCT) with deformable registration. METHODS AND MATERIAL: Twelve patients treated under a 6 fraction liver cancer radiation therapy protocol were imaged in breath hold using kV CBCT at each treatment fraction. The images were imported into the treatment planning software and rigidly registered by fitting the liver, identified on the daily kV CBCT image, into the liver contours, previously drawn on the planning CT. The liver was then manually contoured on each CBCT image. Deformable registration was automatically performed, aligning the CT liver to the liver on each CBCT image using MORFEUS, a biomechanical model based deformable registration algorithm. The tumor, defined on planning CT, was mapped onto the CBCT, through MORFEUS. The center of mass (COM) displacement of the tumor was computed. RESULTS: The mean (SD) displacement magnitude (absolute value) of the COM following deformable registration was 0.08 (0.07), 0.10 (0.11), and 0.10 (0.17) cm in the left-right (LR), anterior-posterior (AP), and superior-inferior (SI) directions, respectively. The maximum displacement of the COM was 0.34, 0.65, and 0.97 cm in the LR, AP, and SI directions, respectively. Fifteen percent of the treatment fractions had a COM displacement of greater than 0.3 cm and 33% of patients had at least 1 fraction with a displacement of greater than 0.3 cm. The deformable registration, excluding the manual contouring of the liver, was performed in less than 1 minute, on average. DISCUSSION: Rigid registration of the liver volume between planning CT and verification kV CBCT localizes the tumor to within 0.3 cm for the majority (66%) of patients; however, larger offsets in tumor position can be observed due to liver deformation.

A quality assurance program for image quality of cone-beam CT guidance in radiation therapy.

The clinical introduction of volumetric x-ray image-guided radiotherapy systems necessitates formal commissioning of the hardware and image-guided processes to be used and drafts quality assurance (QA) for both hardware and processes. Satisfying both requirements provides confidence on the system's ability to manage geometric variations in patient setup and internal organ motion. As these systems become a routine clinical modality, the authors present data from their QA program tracking the image quality performance of ten volumetric systems over a period of 3 years. These data are subsequently used to establish evidence-based tolerances for a QA program. The volumetric imaging systems used in this work combines a linear accelerator with conventional x-ray tube and an amorphous silicon flat-panel detector mounted orthogonally from the accelerator central beam axis, in a cone-beam computed tomography (CBCT) configuration. In the spirit of the AAPM Report No. 74, the present work presents the image quality portion of their QA program; the aspects of the QA protocol addressing imaging geometry have been presented elsewhere. Specifically, the authors are presenting data demonstrating the high linearity of CT numbers, the uniformity of axial reconstructions, and the high contrast spatial resolution of ten CBCT systems (1-2 mm) from two commercial vendors. They are also presenting data accumulated over the period of several months demonstrating the long-term stability of the flat-panel detector and of the distances measured on reconstructed volumetric images. Their tests demonstrate that each specific CBCT system has unique performance. In addition, scattered x rays are shown to influence the imaging performance in terms of spatial resolution, axial reconstruction uniformity, and the linearity of CT numbers.