A closer look at the conventional Winston-Lutz test: Analysis in terms of dose

Integrating cine EPID, dynamic delivery, and the off-axis Winston-Lutz test to enhance quality control in multiple brain metastasis stereotactic radiotherapy

PURPOSE

Stereotactic radiotherapy (SRT) has transformed cancer treatment, especially for brain metastases. Ensuring accurate SRT delivery is crucial, with the Winston-Lutz test being an important quality control tool. Off-axis Winston-Lutz (OAWL) tests are designed for accuracy assessment, but most are limited to fixed angles and hampered by local-field shifts caused by suboptimal Multi-Leaf Collimator (MLC) positioning. This study introduces a new OAWL approach for quality control in multi-brain-metastasis SRT. Utilizing cine Electronic Portal Imaging Device (EPID) images, it can be used with dynamic conformal arc (DCA) therapy. However, dynamic OAWL (DOAWL) is prone to more local-field shifts due to dynamic MLC movements. A two-step DOAWL is proposed: step 1 calculates local-field shifts using dynamic MLC movements in the beam-eye view data from the Treatment Planning System (TPS), while step 2 processes cine EPID images with an OAWL algorithm to isolate true deviations.

METHODS

Validation involved an anthropomorphic head phantom with metallic ball-bearings, Varian TrueBeam STx accelerator delivering six coplanar/non-coplanar DCA beams, cine EPID, and ImageJ's OAWL analysis algorithm.

RESULTS

Inherent local-field shifts ranged from 0.11 to 0.49 mm; corrected mean/max EPID-measured displacement was 0.34/1.03 mm. Few points exceeded 0.75/1.0-mm thresholds.

CONCLUSIONS

This two-step DOAWL test merges cine-EPID acquisitions, DCA, OAWL, and advanced analysis and offers effective quality control for multi-brain-metastasis SRT. Its routine implementation may also improve physicist knowledge of the treatment precision of their machines.

Evaluating the Efficacy of Machine Performance Checks as an Alternative to Winston-Lutz Quality Assurance Testing in the TrueBeam Linear Accelerator with HyperArc

HyperArc is a preferred technique for treating brain metastases, employing a single isocenter for multiple lesions. Geometrical isocentricity in the TrueBeam linear accelerator with HyperArc is crucial. We evaluated machine performance checks (MPCs) as an alternative to the Winston-Lutz (WL) test to verify the treatment isocenter. Between January and July 2023, we assessed 53 data points using MPC and Winston-Lutz tests. The isocenter size obtained from the MPC and its sum, including the rotation-induced couch shift, were compared with the maximum total delta value from the Winston-Lutz test. The maximum total delta was 0.68 ± 0.10 mm, while the isocenter size was 0.28 ± 0.02 mm. The sum of the isocenter size and rotation-induced couch shift measured by MPC was 0.61 ± 0.03 mm. During the Winston-Lutz test (without couch rotation), the maximum total delta value was 0.56 ± 0.13 mm. A t-test analysis revealed a significant difference in the isocenter size averages between the Winston-Lutz and MPC outcomes, whereas the Pearson's correlation coefficient yielded no correlation. Our study highlights the necessity for separate MPC and Winston-Lutz tests for isocenter verification. Therefore, the Winston-Lutz test should precede stereotactic radiosurgery for isocenter verification.

AAPM Medical Physics Practice Guideline 8.b: Linear accelerator performance tests

The purpose of this guideline is to provide a list of critical performance tests to assist the Qualified Medical Physicist (QMP) in establishing and maintaining a safe and effective quality assurance (QA) program. The performance tests on a linear accelerator (linac) should be selected to fit the clinical patterns of use of the accelerator and care should be given to perform tests which are relevant to detecting errors related to the specific use of the accelerator. Current recommendations for linac QA were reviewed to determine any changes required to those tests highlighted by the original report as well as considering new components of the treatment process that have become common since its publication. Recommendations are made on the acquisition of reference data, routine establishment of machine isocenter, basing performance tests on clinical use of the linac, working with vendors to establish QA tests and performing tests after maintenance and upgrades. The recommended tests proposed in this guideline were chosen based on consensus of the guideline's committee after assessing necessary changes from the previous report. The tests are grouped together by class of test (e.g., dosimetry, mechanical, etc.) and clinical parameter tested. Implementation notes are included for each test so that the QMP can understand the overall goal of each test. This guideline will assist the QMP in developing a comprehensive QA program for linacs in the external beam radiation therapy setting. The committee sought to prioritize tests by their implication on quality and patient safety. The QMP is ultimately responsible for implementing appropriate tests. In the spirit of the report from American Association of Physicists in Medicine Task Group 100, individual institutions are encouraged to analyze the risks involved in their own clinical practice and determine which performance tests are relevant in their own radiotherapy clinics.

Evaluation of PTV margins and plan robustness for single isocentre multiple target stereotactic radiosurgery

PURPOSE

Robustness to residual setup errors and linac delivery errors of BrainLab Elements single-isocentre-multiple-target stereotactic radiosurgery was evaluated.

METHODS

Residual setup errors of 13 patients were evaluated. Linac delivery error was quantified through multi-metastases-Winston-Lutz measurements. PTV margins were calculated using the van Herk recipe. Patient scans were translated and rotated by the median and 95th percentile of the combined uncertainties, and plans were recalculated subsequently. Previous patients' plans were then replanned with the derived margins, effects on GTV coverage and normal brain doses were assessed.

RESULTS

Mean (±stdev) coverage of all targets in the original plans were 99.4% (±0.9%) and 98.9% (±1.0%) for 1 and 3-fraction patients respectively. Median geometrical errors did not result in significant differences. A statistically significant reduction in coverage to 91.4% (±10.4%) and 93.0% (±9.6%) was seen under 95th percentile errors. Applying the derived optimal margin of 0.5 mm resulted in 78% of the GTVs retaining a coverage of 98% or above even in the presence of 95th percentile errors, compared to only 30% if no margins were applied. Replanning with margins also caused no significant increase to local normal brain doses, however global dose increases varied according to the number of metastases.

CONCLUSIONS

Plans were shown to be robust to average geometrical uncertainties despite targets having no margins, however occurrence of GTV under-coverage increased under 95th percentile scenarios. The margin was proven to substantially improve the target dose coverage with limited change to local normal brain doses, although not all sources of geometrical uncertainty were considered.

An improved method for evaluating LINAC isocenter

BACKGROUND

The traditional approach to medical linear accelerator (LINAC) isocenter quality assurance is to compare the radius of the LINAC isocenter, determined via analysis of Winston-Lutz (WL) dataset, to a threshold in order to approve the LINAC for clinical use. This scalar metric provides little insight into beam-to-target accuracy with gantry and couch motion.

PURPOSE

To develop a method for verifying isocenter with increased sensitivity over traditional WL techniques by accounting for geometric errors that could normally be overlooked.

METHODS

From WL images, we construct radiation beam axis and marker shift locations in a 3D coordinate system. These axes and shift positions are used to construct a new isocenter performance metric, the Z $\mathcal {Z}$ -matrix, which predicts direction and magnitude of beam-to-target errors across combinations of couch and gantry position. We introduce clinical isocenter as the location optimizing a cost function derived from the Z $\mathcal {Z}$ -matrix, which serves as an optimal target for tumor positioning. We demonstrated these techniques on a clinical LINAC with an initial randomly positioned marker which was subsequently repositioned based on the optimized clinical isocenter location. The marker shifts, radiation isocenter, and Z $\mathcal {Z}$ -matrix were compared before and after repositioning. We compared the new technique against typically used WL techniques using a monte carlo simulation modeling variations in LINAC geometry, marker position, and measurement noise.

RESULTS

The technique was successfully demonstrated on a Varian LINAC. Marker repositioning to clinical isocenter yielding an error matrix with magnitudes all below 0.81 mm. As expected, marker position had little impact on the radiation isocenter location and radius, and also had little impact on clinical isocenter location. The verification results show the accuracy of the Z $\mathcal {Z}$ -matrix to predict beam to tumor geometric inaccuracies. The Monte Carlo simulations demonstrate that the Z $\mathcal {Z}$ -matrix is more sensitive and specific for detecting potential treatment errors compared to the traditional WL techniques.

CONCLUSIONS

We have developed and demonstrated the usefulness of a framework for verifying isocenter based on a 3D model of the radiation beam axes and tumor movement from couch rotations, derived from 2D WL transmission images. The Z $\mathcal {Z}$ -matrix replaces the scalar isocenter radius as a metric of isocenter quality, providing insight into contribution of couch and radiation beams to isocenter quality, and exposing treatment errors ignored by the traditional method. The clinical isocenter provides an alternate to physical isocenter as a target for tumor positioning in cases of suboptimal LINAC geometry.

Evaluation of HyperArc™ using film and portal dosimetry quality assurance

HyperArc™ is a stereotactic radiotherapy modality designed for targeting multiple brain metastases using a single isocenter with multiple non-coplanar arcs. This study aimed to assess the efficacy of two patient-specific quality assurance methods, film and the Varian Portal Dosimetry System with Varian's HyperArc™ technique and raise important considerations in the customisation of patient-specific quality assurance to accommodate HyperArc™ delivery. Assessment criteria included gamma analysis and mean dose at full width half maximum. The minimum metastasis size, maximum off-axis distance and suitable energy were identified and validated. Patient-specific quality assurance procedures were applied to a range of clinically relevant brain metastasis plans. Initial investigation into energy selection showed no significant differences in gamma pass rates using 6MV, 6MV FFF, or 10MV FFF for metastasis sizes greater than 15 mm diameter at the isocenter. Gamma pass rates (2%/2mm) for 15 mm metastases at the isocenter for all energies were greater than 96.0% for portal dosimetry and greater than 98.7% for film. Fields of size 15 mm placed at various distances (10-70 mm) from the isocenter resulted in a maximum mean dose difference of 1.5% between film and planned. Clinically relevant plans resulted in a maximum mean dose difference for selected metastases of 1.0% between film and plan and a maximum point dose difference of 2.9% between portal dose and plan. Portal dose image prediction was a quick and convenient quality assurance tool for metastases larger than 15 mm near the isocenter but provided diminished geometrical relevance for off-axis metastases. Film QA required exacting procedures but offered the ability to assess the accuracy of geometrical targeting for off-axis metastases and provided dosimetric accuracy for metastases to well below 15 mm diameter.

Simple quality assurance based on filtered back projection for geometrical/irradiation accuracy in single-isocenter multiple-target stereotactic radiotherapy

In single-isocenter multiple-target stereotactic radiotherapy (SIMT-SRT), it is difficult to evaluate both the geometrical accuracy and absorbed dose measurement when irradiating off-isocenter targets. This study aimed to develop a simple quality assurance (QA) method to evaluate off-isocenter irradiation position accuracy in SIMT-SRT and compare its feasibility with that of a commercial device. First, we created two types of inserts and metallic balls with a diameter of 5 mm to be inserted into a commercially available phantom (SIMT phantom). Second, we developed a dedicated analysis software using Python for the Winston-Lutz test (WLT). Third, an image processing software, including the filtered back-projection algorithm, was developed to analyze the images obtained using an electronic portal imaging device (EPID). Fourth, the feasibility of our method was evaluated by comparing it with the results of WLT using two commercially available phantoms: WL-QA and MultiMet-WL cubes. Notably, 92% of the results in one-dimensional deviations were within 0.26 mm (EPID pixel width). The correlation coefficients were 0.52, 0.92, and 0.96 in the left-right, superior-inferior, and anterior-posterior directions, respectively. In the WLT, a maximum two-dimensional deviation of 0.70 mm was detected in our method, while the deviation in the other method was within 0.5 mm. The advantage of our method is that it can evaluate the geometrical accuracy at any gantry angle during dynamic rotation irradiation using a filtered back-projection algorithm, even if the target is located off the isocenter. Our method can perform WLT at arbitrary positions and is suitable for the QA of dynamic rotation irradiation using an EPID.

Design and automation of specific geometric quality controls for cone-radiosurgery treatments

PURPOSE

The Varian TrueBeam STx linac can be equipped with BrainLAB stereotactic cones and ExacTrac imaging system for SRS treatments. However, these two third-party systems lack integration in a self-performance diagnosis tool dedicated to the SRS platform. The aim of this work was to design and automate essential geometric tests considering the complete set of cones with diameter range from 4 to 15 mm.

METHODS

EPID-based tests were focused on the cone alignment, the radiation isocentricity and the isocenters congruence. Images acquired with or without the BrainLAB pointer were analysed using the Hough transform and morphological filtering operations, after assessment of the algorithm accuracy using simulated images. The new Machine QA program was experimented over one year.

RESULTS

A subpixel resolution of 0.02 mm was found for the circular-field center detection algorithm. The tests results did not depend on the pointer location. The maximum deviations reported were in accordance with the AAPM SRS guidelines. The accurate localization of the linac radiation isocenter allowed for guidance of the ExacTrac calibration. A misalignment reaching 0.2 mm was measured for all cones but one, highlighting the benefit of systematizing this control before each patient SRS treatment.

CONCLUSION

An effective in-house QA program dedicated to SRS cones was developed to supplement the standard machine performance check on our mixed SRS platform. Specific geometric tests even include the smallest 4-mm cone, which could be of great interest for future clinical indications such as the radiosurgery of functional disorders.

DeepWL: Robust EPID based Winston-Lutz analysis using deep learning, synthetic image generation and optical path-tracing

Radiation therapy requires clinical linear accelerators to be mechanically and dosimetrically calibrated to a high standard. One important quality assurance test is the Winston-Lutz test which localises the radiation isocentre of the linac. In the current work we demonstrate a novel method of analysing EPID based Winston-Lutz QA images using a deep learning model trained only on synthetic image data. In addition, we propose a novel method of generating the synthetic WL images and associated 'ground-truth' masks using an optical path-tracing engine to 'fake' mega-voltage EPID images. The model called DeepWL was trained on 1500 synthetic WL images using data augmentation techniques for 180 epochs. The model was built using Keras with a TensorFlow backend on an Intel Core i5-6500T CPU and trained in approximately 15 h. DeepWL was shown to produce ball bearing and multi-leaf collimator field segmentations with a mean dice coefficient of 0.964 and 0.994 respectively on previously unseen synthetic testing data. When DeepWL was applied to WL data measured on an EPID, the predicted mean displacements were shown to be statistically similar to the Canny Edge detection method. However, the DeepWL predictions for the ball bearing locations were shown to correlate better with manual annotations compared with the Canny edge detection algorithm. DeepWL was demonstrated to analyse Winston-Lutz images with an accuracy suitable for routine linac quality assurance with some statistical evidence that it may outperform Canny Edge detection methods in terms of segmentation robustness and the resultant displacement predictions.

Intra-Fraction Motion Management for Radiosurgical Treatments of Trigeminal Neuralgia: Clinical Experience, Imaging Frequency, and Motion Analysis

Purpose

The aim of this study is to evaluate the patient positioning and intra-fraction motion management performance of an image-guidance protocol established for radiosurgical treatments of trigeminal neuralgia patients. Specifically, it also aims to analyze patient motion data for the evaluation of current motion tolerance levels and imaging frequency utilized for repositioning patients.

Methods

A linear accelerator equipped with ExacTrac is used for patient positioning with stereoscopic imaging and treatments. Treatments are delivered with 4-mm conical collimators using seven equally spaced arcs. Arcs are 20 degrees apart and span 100 arc degrees each. Following initial ExacTrac positioning, cone beam computed tomography (CBCT) is obtained for independent confirmation of patient position. Patients are then stereoscopically imaged prior to the delivery of each arc and repositioned when 0.5-mm translational tolerance in any direction is exceeded. After the patient has been repositioned, verification stereoscopic images are obtained. Data from 48 patients with 607 image pairs were analyzed for this study.

Results

Over the course of 48 patient treatments, the mean magnitude of mean 3D deviations was 0.64 mm ± 0.12 mm (range: 0.07-2.74 mm). With the current 0.50-mm tolerance level for repositioning, patients exceeded the tolerance 51.4% of the time considering only images following an arc segment. For those instances, patients were repositioned with a mean magnitude of 0.85 mm ± 0.15 mm (1 SD). For a 0.25-mm tolerance level, 86.1% of arc segments would have required repositioning following the delivery of an arc segment, with a mean magnitude of 0.68 mm ± 0.12 mm. Conversely, for 0.75-mm and 1.00-mm tolerance levels, the tolerance would have been exceeded only 21.5% and 6.6% of instances following the delivery of an arc segment, with a mean magnitude of 1.08 mm ± 0.21 mm and 1.34 mm ± 0.24 mm, respectively. Each repositioning adds approximately 2 minutes to treatment time, which accounts for parts of the variability in patient treatment times. Following the initial ExacTrac and CBCT, the mean treatment time from first arc to treatment end was 57 minutes (range: 33-63 minutes).

Discussions

The current 0.50-mm tolerance level results in a clinically manageable but significant number of patient repositions during trigeminal neuralgia treatments. Frequent patient repositioning can result from actual patient motion convolved with the accuracy and precision limitations of the image analysis. Increasing the repositioning tolerance could more selectively correct for actual patient motion and shorten the treatment time at the expense of more variations in patient position. A more lenient tolerance level of 0.75 mm would decrease the repositioning rate by approximately a factor of 2; however, the permissible magnitude of motion will increase, leading to possible dosimetric consequences. Once treatment begins, there was no trend as to when patients exceeded the tolerance.

Conclusions

Current imaging protocol for patient positioning and intra-fraction motion management fits the clinical workflow with clinically acceptable residual patient motion. The next important step would be to assess how the number of repositions and magnitude of residual movements affect treatment outcomes.

Comparison of three different phantoms used for Winston-Lutz test with Artiscan software

BACKGROUND

One of the most important test in every quality assurances process of medical linear accelerators is the Winston-Lutz test, allowing an evaluation of the treatment isocentre in the light of uncertainty of the position of the collimator, the gantry and the couch.

AIM

The purpose of this work was analysis of the results of the Winston-Lutz test performed with three different phantoms for two different accelerators.

MATERIALS AND METHODS

Measurements were performed on two Varian machines: TrueBeam equipped with aS1200 EPID and TrueBeam equipped with aS1000 EPID. During the study three different phantoms dedicated for verification of the radiation isocentre were used: PTW Isoball, AQUILAB Isocentre Phantom and Varian Isocentre Cube. Analysis of the DICOM images was performed in Artiscan software.

RESULTS

For TrueBeam with as1200 EPID, gantry MV isocentre was about 0.18 mm larger for Varian Isocentre Cube than for two other phantoms used in this study. The largest variability of this parameter was observed for the couch. The results differed to 1.16 mm. For TrueBeam with as1000 EPID, results for collimator isocentre with PTW Isoball phantom were about 0.10 mm larger than for two other phantoms. For the gantry, results obtained with Varian Isocentre Cube were 0.21 mm larger.

CONCLUSION

The obtained results for all three phantoms are within the accepted tolerance range. The largest differences were observed for treatment couch, which may be related to the phantom mobility during couch movement.