MR-linac MLC positioning QA by digitally stitching dual double-exposed films.

PURPOSE: The picket fence (PF) test is highly recommended for multi-leaf collimator (MLC) quality assurance. However, since the electronic portal imaging device (EPID) on the Elekta Unity only covers a small area, it is not feasible to perform the PF test for the entire MLC. Here, we propose a technique for the PF test by stitching two double-exposed films. METHODS: Two EBT3 films were used to encompass the entire MLC, with each one covering one half of the area. Two fields were employed to apply double exposure: a PF pattern consisting of 11 2 mm wide pickets and a 2.84 cm x 22 cm open field. The edges of the open field defined by the diaphragms were used to correct film rotation as well as align them horizontally. The PF pattern was also measured with the EPID where the pickets were used to align the films vertically. Individual leaf positions were detected on the merged film for quantitative analysis. Various MLC positioning errors were introduced to evaluate the technique's sensitivity. RESULTS: The merged films covered 72 leaf pairs properly (four leaf pairs on both sides were outside the treatment couch). With the EPID, the leaf positioning accuracy was -0.02 ± 0.07 mm (maximum: 0.29 mm) and the picket width variation was 0.00 ± 0.03 mm (maximum: 0.11 mm); with the films, the position accuracy and width variation were -0.03 ± 0.13 mm (maximum: 0.80 mm) and 0.00 ± 0.13 mm (maximum: 0.74 mm), respectively. The EPID was able to detect errors of 0.5 mm or above with submillimeter accuracy; the films were only able to detect errors > 1.0 mm. CONCLUSION: We developed a quantitative technique for the PF test on the Elekta Unity. The merged films covered nearly the entire MLC leaf banks. The technique exhibited clinically acceptable accuracy and sensitivity to MLC positioning errors.

Performance assessment of variant UNet-based deep-learning dose engines for MR-Linac-based prostate IMRT plans.

Objective. UNet-based deep-learning (DL) architectures are promising dose engines for traditional linear accelerator (Linac) models. Current UNet-based engines, however, were designed differently with various strategies, making it challenging to fairly compare the results from different studies. The objective of this study is to thoroughly evaluate the performance of UNet-based models on magnetic-resonance (MR)-Linac-based intensity-modulated radiation therapy (IMRT) dose calculations.Approach. The UNet-based models, including the standard-UNet, cascaded-UNet, dense-dilated-UNet, residual-UNet, HD-UNet, and attention-aware-UNet, were implemented. The model input is patient CT and IMRT field dose in water, and the output is patient dose calculated by DL model. The reference dose was calculated by the Monaco Monte Carlo module. Twenty training and ten test cases of prostate patients were included. The accuracy of the DL-calculated doses was measured using gamma analysis, and the calculation efficiency was evaluated by inference time.Results. All the studied models effectively corrected low-accuracy doses in water to high-accuracy patient doses in a magnetic field. The gamma passing rates between reference and DL-calculated doses were over 86% (1%/1 mm), 98% (2%/2 mm), and 99% (3%/3 mm) for all the models. The inference times ranged from 0.03 (graphics processing unit) to 7.5 (central processing unit) seconds. Each model demonstrated different strengths in calculation accuracy and efficiency; Res-UNet achieved the highest accuracy, HD-UNet offered high accuracy with the fewest parameters but the longest inference, dense-dilated-UNet was consistently accurate regardless of model levels, standard-UNet had the shortest inference but relatively lower accuracy, and the others showed average performance. Therefore, the best-performing model would depend on the specific clinical needs and available computational resources.Significance. The feasibility of using common UNet-based models for MR-Linac-based dose calculations has been explored in this study. By using the same model input type, patient training data, and computing environment, a fair assessment of the models' performance was present.

An ultra-fast deep-learning-based dose engine for prostate VMAT via knowledge distillation framework with limited patient data.

Objective. Deep-learning (DL)-based dose engines have been developed to alleviate the intrinsic compromise between the calculation accuracy and efficiency of the traditional dose calculation algorithms. However, current DL-based engines typically possess high computational complexity and require powerful computing devices. Therefore, to mitigate their computational burdens and broaden their applicability to a clinical setting where resource-limited devices are available, we proposed a compact dose engine via knowledge distillation (KD) framework that offers an ultra-fast calculation speed with high accuracy for prostate Volumetric Modulated Arc Therapy (VMAT).Approach. The KD framework contains two sub-models: a large pre-trained teacher and a small to-be-trained student. The student receives knowledge transferred from the teacher for better generalization. The trained student serves as the final engine for dose calculation. The model input is patient computed tomography and VMAT dose in water, and the output is DL-calculated patient dose. The ground-truth \dose was computed by the Monte Carlo module of the Monaco treatment planning system. Twenty and ten prostate cases were included for model training and assessment, respectively. The model's performance (teacher/student/student-only) was evaluated by Gamma analysis and inference efficiency.Main results. The dosimetric comparisons (input/DL-calculated/ground-truth doses) suggest that the proposed engine can effectively convert low-accuracy doses in water to high-accuracy patient doses. The Gamma passing rate (2%/2 mm, 10% threshold) between the DL-calculated and ground-truth doses was 98.64 ± 0.62% (teacher), 98.13 ± 0.76% (student), and 96.95 ± 1.02% (student-only). The inference time was 16 milliseconds (teacher) and 11 milliseconds (student/student-only) using a graphics processing unit device, while it was 936 milliseconds (teacher) and 374 milliseconds (student/student-only) using a central processing unit device.Significance. With the KD framework, a compact dose engine can achieve comparable accuracy to that of a larger one. Its compact size reduces the computational burdens and computing device requirements, and thus such an engine can be more clinically applicable.

A polar-coordinate-based pencil beam algorithm for VMAT dose computation with high-resolution gantry angle sampling.

PURPOSE: Most commercially available treatment planning systems (TPSs) approximate the continuous delivery of volumetric modulated arc therapy (VMAT) plans with a series of discretized static beams for treatment planning, which can make VMAT dose computation extremely inefficient. In this study, we developed a polar-coordinate-based pencil beam (PB) algorithm for efficient VMAT dose computation with high-resolution gantry angle sampling that can improve the computational efficiency and reduce the dose discrepancy due to the angular under-sampling effect. METHODS AND MATERIALS: 6 MV 1 × 1 m m 2 $1 \times 1{\rm{\;m}}{{\rm{m}}^2}$ pencil beams were simulated on a uniform cylindrical phantom under an EGSnrc Monte Carlo (MC) environment. The MC-generated PB kernels were collected in the polar coordinate system for each bixel on a 40 × 40 c m 2 $40 \times 40{\rm{\;c}}{{\rm{m}}^2}$ fluence map and subsequently fitted via a series of Gaussians. The fluence was calculated using a detectors' eye view with off-axis and MLC transmission factors corrected. Doses of VMAT arc on the phantom were computed by summing the convolution results between the corresponding PB kernels and fluence for each bixel in the polar coordinate system. The convolution was performed using fast Fourier transform to expedite the computing speed. The calculated doses were converted to the Cartesian coordinate system and compared with the reference dose computed by a collapsed cone convolution (CCC) algorithm of the TPS. A heterogeneous phantom was created to study the heterogeneity corrections using the proposed algorithm. Ten VMAT arcs were included to evaluate the algorithm performance. Gamma analysis and computation complexity theory were used to measure the dosimetric accuracy and computational efficiency, respectively. RESULTS: The dosimetric comparisons on the homogeneous phantom between the proposed PB algorithm and the CCC algorithm for 10 VMAT arcs demonstrate that the proposed algorithm can achieve a dosimetric accuracy comparable to that of the CCC algorithm with average gamma passing rates of 96% (2%/2mm) and 98% (3%/3mm). In addition, the proposed algorithm can provide better computational efficiency for VMAT dose computation using a PC equipped with a 4-core processor, compared to the CCC algorithm utilizing a dual 10-core server. Moreover, the computation complexity theory reveals that the proposed algorithm has a great advantage with regard to computational efficiency for VMAT dose computation on homogeneous medium, especially when a fine angular sampling rate is applied. This can support a reduction in dose errors from the angular under-sampling effect by using a finer angular sampling rate, while still preserving a practical computing speed. For dose calculation on the heterogeneous phantom, the proposed algorithm with heterogeneity corrections can still offer a reasonable dosimetric accuracy with comparable computational efficiency to that of the CCC algorithm. CONCLUSIONS: We proposed a novel polar-coordinate-based pencil beam algorithm for VMAT dose computation that enables a better computational efficiency while maintaining clinically acceptable dosimetric accuracy and reducing dose error caused by the angular under-sampling effect. It also provides a flexible VMAT dose computation structure that allows adjustable sampling rates and direct dose computation in regions of interest, which makes the algorithm potentially useful for clinical applications such as independent dose verification for VMAT patient-specific QA.

Reconstruction of volume averaging effect-free continuous photon beam profiles from discrete ionization chamber array measurements using a machine learning technique.

PURPOSE: The use of the ionization chamber array ICProfiler (ICP) is limited by its relatively poor detector spatial resolution and the inherent volume averaging effect (VAE). The purpose of this work is to study the feasibility of reconstructing VAE-free continuous photon beam profiles from ICP measurements with a machine learning technique. METHODS: In- and cross-plane photon beam profiles of a 6 MV beam from an Elekta linear accelerator, ranging from 2 × 2 to 10 × 10 cm2 at 1.5 cm, 5 cm, and 10 cm depth, were measured with an ICP. The discrete measurements were interpolated with a Makima method to obtain continuous beam profiles. Artificial neural networks (ANNs) were trained to restore the penumbra of the beam profiles. Plane-specific (in- and cr-plane) ANNs and a combined ANN were separately trained. The performance of the ANNs was evaluated using the penumbra width difference (PWD, the difference between the penumbra widths of the reconstructed and the reference profile). The plane-specific and the combined ANNs were compared to study the feasibility of using a single ANN for both in- and cross-plane. RESULTS: The profiles reconstructed with all the ANNs had excellent agreement with the reference. For in-plane, the ANNs reduced the PWD from 1.6 ± 0.7 mm at 1.5 cm depth to 0.1 ± 0.1 mm, from 1.8 ± 0.6 mm at 5.0 cm depth to 0.1 ± 0.1 mm, and from 2.4 ± 0.1 mm at 10.0 cm depth to 0.0 ± 0.0 mm; for cross-plane, the ANNs reduced the PWD from 1.2 ± 0.4 mm at 1.5 cm depth, 1.2 ± 0.3 mm at 5.0 cm depth, and 1.6 ± 0.1 mm at 10.0 cm depth, to 0.1 ± 0.1 mm. CONCLUSIONS: This study demonstrated the feasibility of using simple ANNs to reconstruct VAE-free continuous photon beam profiles from discrete ICP measurements. A combined ANN can restore the penumbra of in- and cross-plane beam profiles of various fields at different depths.

AAPM Task Group 198 Report: An implementation guide for TG 142 quality assurance of medical accelerators.

The charges on this task group (TG) were as follows: (a) provide specific procedural guidelines for performing the tests recommended in TG 142; (b) provide estimate of the range of time, appropriate personnel, and qualifications necessary to complete the tests in TG 142; and (c) provide sample daily, weekly, monthly, or annual quality assurance (QA) forms. Many of the guidelines in this report are drawn from the literature and are included in the references. When literature was not available, specific test methods reflect the experiences of the TG members (e.g., a test method for door interlock is self-evident with no literature necessary). In other cases, the technology is so new that no literature for test methods was available. Given broad clinical adaptation of volumetric modulated arc therapy (VMAT), which is not a specific topic of TG 142, several tests and criteria specific to VMAT were added.

Tumor phase recognition using cone-beam computed tomography projections and external surrogate information.

PURPOSE: Directly extracting the respiratory phase pattern of the tumor using cone-beam computed tomography (CBCT) projections is challenging due to the poor tumor visibility caused by the obstruction of multiple anatomic structures on the beam's eye view. Predicting tumor phase information using external surrogate also has intrinsic difficulties as the phase patterns between surrogates and tumors are not necessary to be congruent. In this work, we developed an algorithm to accurately recover the primary oscillation components of tumor motion using the combined information from both CBCT projections and external surrogates. METHODS: The algorithm involved two steps. First, a preliminary tumor phase pattern was acquired by applying local principal component analysis (LPCA) on the cropped Amsterdam Shroud (AS) images. In this step, only the cropped image of the tumor region was used to extract the tumor phase pattern in order to minimize the impact of pattern recognition from other anatomic structures. Second, by performing multivariate singular spectrum analysis (MSSA) on the combined information containing both external surrogate signal and the original waveform acquired in the first step, the primary component of the tumor phase oscillation was recovered. For the phantom study, a QUASAR respiratory motion phantom with a removable tumor-simulator insert was employed to acquire CBCT projection images. A comparison between LPCA only and our method was assessed by power spectrum analysis. Also, the motion pattern was simulated under the phase shift or various amplitude conditions to examine the robustness of our method. Finally, anatomic obstruction scenarios were simulated by attaching a heart model, PVC tubes, and RANDO® phantom slabs to the phantom, respectively. Each scenario was tested with five real-patient breathing patterns to mimic real clinical situations. For the patient study, eight patients with various tumor locations were selected. The performance of our method was then evaluated by comparing the reference waveform with the extracted signal for overall phase discrepancy, expiration phase discrepancy, peak, and valley accuracy. RESULTS: In tests of phase shifts and amplitude variations, the overall peak and valley accuracy was -0.009 ± 0.18 sec, and no time delay was found compared to the reference. In anatomical obstruction tests, the extracted signal had 1.6 ± 1.2 % expiration phase discrepancy, -0.12 ± 0.28 sec peak accuracy, and 0.01 ± 0.15 sec valley accuracy. For patient studies, the extracted signal using our method had -1.05 ± 3.0 % overall phase discrepancy, -1.55 ± 1.45% expiration phase discrepancy, 0.04 ± 0.13 sec peak accuracy, and -0.01 ± 0.15 sec valley accuracy, compared to the reference waveforms. CONCLUSIONS: An innovative method capable of accurately recognizing tumor phase information was developed. With the aid of extra information from the external surrogate, an improvement in prediction accuracy, as compared with traditional statistical methods, was obtained. It enables us to employ it as the ground truth for 4D-CBCT reconstruction, gating treatment, and other clinic implementations that require accurate tumor phase information.

Evaluation of a neural network-based photon beam profile deconvolution method.

PURPOSE: The authors have previously shown the feasibility of using an artificial neural network (ANN) to eliminate the volume average effect (VAE) of scanning ionization chambers (ICs). The purpose of this work was to evaluate the method when applied to beams of different energies (6 and 10 MV) and modalities [flattened (FF) vs unflattened (FFF)], measured with ICs of various sizes. METHODS: The three-layer ANN extracted data from transverse photon beam profiles using a sliding window, and output deconvolved value corresponding to the location at the center of the window. Beam profiles of seven fields ranging from 2 × 2 to 10 × 10 cm2 at four depths (1.5, 5, 10 and 20 cm) were measured with three ICs (CC04, CC13, and FC65-P) and an EDGE diode detector for 6 MV FF and FFF. Similar data for the 10 MV FF beam was also collected with CC13 and EDGE. The EDGE-measured profiles were used as reference data to train and test the ANNs. Separate ANNs were trained by using the data of each beam energy and modality. Combined ANNs were also trained by combining data of different beam energies and/or modalities. The ANN's performance was quantified and compared by evaluating the penumbra width difference (PWD) between the deconvolved and reference profiles. RESULTS: Excellent agreement between the deconvolved and reference profiles was achieved with both separate and combined ANNs for all studied ICs, beam energies, beam modalities, and geometries. After deconvolution, the average PWD decreased from 1-3 mm to under 0.15 mm with separate ANNs and to under 0.20 mm with combined ANN. CONCLUSIONS: The ANN-based deconvolution method can be effectively applied to beams of different energies and modalities measured with ICs of various sizes. Separate ANNs yielded marginally better results than combined ANNs. An IC-specific, combined ANN can provide clinically acceptable results as long as the training data includes data of each beam energy and modality.

Beam flatness modulation for a flattening filter free photon beam utilizing a novel direct leaf trajectory optimization model.

Flattening filter free (FFF) linear accelerators produce a fluence distribution that is forward peaked. Various dosimetric benefits, such as increased dose rate, reduced leakage and out of field dose has led to the growth of FFF technology in the clinic. The literature has suggested the idea of vendors offering dedicated FFF units where the flattening filter (FF) is removed completely and manipulating the beam to deliver conventional flat radiotherapy treatments. This work aims to develop an effective way to deliver modulated flat beam treatments, rather than utilizing a physical FF. This novel optimization model is an extension of the direct leaf trajectory optimization (DLTO) previously developed for volumetric modulated radiation therapy (VMAT) and is capable of accounting for all machine and multileaf collimator (MLC) dynamic delivery constraints, using a combination of linear constraints and a convex objective function. Furthermore, the tongue and groove (T&G) effect was also incorporated directly into our model without introducing nonlinearity to the constraints, nor nonconvexity to the objective function. The overall beam flatness, machine deliverability, and treatment time efficiency were assessed. Regular square fields, including field sizes of 10 × 10 cm2 to 40 × 40 cm2 were analyzed, as well as three clinical fields, and three arbitrary contours with "concave" features. Quantitative flatness was measured for all modulated FFF fields, and the results were comparable or better than their open FF counterparts, with the majority having a quantitative flatness of less than 3.0%. The modulated FFF beams, due to the included efficiency constraint, were able to achieve acceptable delivery time compared to their open FF counterpart. The results indicated that the dose uniformity and flatness for the modulated FFF beams optimized with the DLTO model can successfully match the uniformity and flatness of their conventional FF counterparts, and may even provide further benefit by taking advantage of the unique FFF beam characteristics.

A sliding-window approach for improved VMAT dose calculation accuracy.

The continuous delivery of volumetric modulated arc therapy (VMAT) plans is usually approximated by discrete apertures at evenly-spaced gantry angles for dose calculation purposes. This approximation can potentially lead to large dose calculation errors if the gantry angle spacings are large and/or there are large changes in the MLC apertures from one control point (CP) to the next. In this work, we developed a sliding-window (SW) method to improve VMAT dose calculation accuracy. For any 2 adjacent VMAT CPs ni and ni + 1, the dose distribution was approximated by a 2-CP SW IMRT beam with the starting MLC positions at CP ni and ending MLC positions at CP ni + 1, with the gantry angle fixed in the middle of the 2 VMAT CPs. Therefore, a VMAT beam with N CPs was approximated by a SW plan with N-1 SW beams. To validate the method, VMAT plans were generated for 10 patients in Pinnacle using 4° gantry spacing. Each plan was converted to a SW plan and dose was recalculated. Another VMAT plan, with 1° gantry spacing, was created by interpolating the original VMAT beam. The original plans were delivered on an Elekta Versa HD and measured with ArcCHECK. For both the isodose distribution and DVH, there were significant differences between the original VMAT plan and either the SW or the interpolated plan. However, they were indistinguishable between the SW and the interpolated plans. When compared with measurement, the average passing rates of the original VMAT plans were 87.3 ± 2.8% and 93.1 ± 1.0% for the 5 HN and 5 spine SBRT cases, respectively. On the other hand, the passing rates for both the VMAT1 and SW plans were above 95% for all the 10 cases studied. The dose calculation times of the original VMAT plans and the SW plans were very similar. We conclude that the proposed SW approach improves VMAT dose calculation accuracy without increase in dose calculation time.

Error detection and classification in patient-specific IMRT QA with dual neural networks.

PURPOSE: Despite being the standard metric in patient-specific quality assurance (QA) for intensity-modulated radiotherapy (IMRT), gamma analysis has two shortcomings: (a) it lacks sensitivity to small but clinically relevant errors (b) it does not provide efficient means to classify the error sources. The purpose of this work is to propose a dual neural network method to achieve simultaneous error detection and classification in patient-specific IMRT QA. METHODS: For a pair of dose distributions, we extracted the dose difference histogram (DDH) for the low dose gradient region and two signed distance-to-agreement (sDTA) maps (one in x direction and one in y direction) for the high dose gradient region. An artificial neural network (ANN) and a convolutional neural network (CNN) were designed to analyze the DDH and the two sDTA maps, respectively. The ANN was trained to detect and classify six classes of dosimetric errors: incorrect multileaf collimator (MLC) transmission (±1%) and four types of monitor unit (MU) scaling errors (±1% and ±2%). The CNN was trained to detect and classify seven classes of spatial errors: incorrect effective source size, 1 mm MLC leaf bank overtravel or undertravel, 2 mm single MLC leaf overtravel or undertravel, and device misalignment errors (1 mm in x- or y direction). An in-house planar dose calculation software was used to simulate measurements with errors and noise introduced. Both networks were trained and validated with 13 IMRT plans (totaling 88 fields). A fivefold cross-validation technique was used to evaluate their accuracy. RESULTS: Distinct features were found in the DDH and the sDTA maps. The ANN perfectly identified all four types of MU scaling errors and the specific accuracies for the classes of no error, MLC transmission increase, MLC transmission decrease were 98.9%, 96.6%, and 94.3%, respectively. For the CNN, the largest confusion occurred between the 1-mm-MLC bank overtravel class and the 1-mm-device alignment error in x-direction class, which brought the specific accuracies down to 90.9% and 92.0%, respectively. The specific accuracy for the 2-mm-single MLC leaf undertravel class was 93.2% as it misclassified 5.7% of the class as being error free (false negative). Otherwise, the specific accuracy was above 95%. The overall accuracies across the fivefold were 98.3 ± 0.7% and 95.6% ± 1.5% for the ANN and the CNN, respectively. CONCLUSIONS: Both the DDH and the sDTA maps are suitable features for error classification in IMRT QA. The proposed dual neural network method achieved simultaneous error detection and classification with excellent accuracy. It could be used in complement with the gamma analysis to potentially shift the IMRT QA paradigm from passive pass/fail analysis to active error detection and root cause identification.

A self-checking treatment couch coordinate calculation system in radiotherapy.

PURPOSE: Traditionally, the treatment couch coordinates (TCCs) for patients undergoing radiotherapy can only be determined at the time of treatment, placing pressure on the treating therapists and leaving several pathways for errors such as wrong-site treatment or wrong treatment table shift from a reference point. The purpose of this work is to propose an accurate, robust, and streamlined system that calculates TCC in advance. METHODS: The proposed system combines the advantages of two different calculation methods that use an indexed immobilization device. The first method uses an array of reference ball bearings (BBs) embedded in the CT scanner's couch-top. To obtain the patient-specific TCC, the spatial offset of the treatment planning isocenter from the reference BB is used. The second method performs a calculation using the one-to-one mapping relationship between the CT scanner's DICOM (Digital Imaging and Communications in Medicine) coordinate system and the TCC system. Both methods use a reference point in the CT coordinate system to correlate a point in the TCC system to perform the coordinate transfer between the two systems. Both methods were used to calculate the TCC and the results were checked against each other, creating an integrated workflow via automated self-checking. The accuracy of the calculation system was retrospectively evaluated with 275 patients, where the actual treatment position determined with cone-beam CT was used as a reference. RESULTS: An efficient workflow transparent to the therapists at both CT simulation and treatment was created. It works with any indexed immobilization device and can be universally applied to all treatment sites. The two methods had comparable accuracy, with 95% of the calculations within 3 mm. The inter-fraction variation was within ± 1.0 cm for 95% of the coordinates across all the treatment sites. CONCLUSIONS: A robust, accurate, and streamlined system was implemented to calculate TCCs in advance. It eases the pressure on the treating therapists, reduces patient setup time, and enhances the patient safety by preventing setup errors.

Feasibility study of using flattening-filter-free photon beams to deliver conventional flat beams.

Various dosimetric benefits such as increased dose rate, and reduced leakage and out of field dose have led to the growth of flattening-filter-free (FFF) technology in the clinic. In this study, we concentrate on investigating the feasibility of using FFF beams to deliver conventional flat beams, since completely getting rid of the flattening-filter module from the gantry head can not only simplify the gantry design but also decrease the workload on machine maintenance and quality assurance. Two intensity modulated radiotherapy techniques, step-and-shoot (S&S) and sliding window (SW), were used to generate flat beam profiles for 6 regular-shaped beams and 3 clinical beams while operating in FFF mode. The inverse plans were generated based on uniform dose optimization. Degree of flatness, MU efficiency, and beam delivery time for both methods were assessed. S&S technique is able to achieve a degree of flatness less than 2.5% for most field configurations. While SW technique was able to generate relatively flat beams for field sizes less than 18 × 18 cm2. For all field configurations, S&S beams resulted in a longer delivery time compared to reference flat beams and SW beams. For field sizes less than 18 × 18 cm2, SW modulated FFF beams resulted in a faster delivery time compared to reference flat beams. The ability to deliver conventional flat beams is not absent when operating in FFF mode. Utilizing beam modulation, FFF mode can achieve reasonable flat profiles and comparable efficiency to conventional flat beams. The ability to deliver most clinical treatments from the same treatment unit will allow for less quality assurance as well as maintenance, and completely eliminate the need for the flattening filter on modern linacs.

Feasibility of photon beam profile deconvolution using a neural network.

PURPOSE: Ionization chambers are the detectors of choice for photon beam profile scanning. However, they introduce significant volume averaging effect (VAE) that can artificially broaden the penumbra width by 2-3 mm. The purpose of this study was to examine the feasibility of photon beam profile deconvolution (the elimination of VAE from ionization chamber-measured beam profiles) using a three-layer feedforward neural network. METHODS: Transverse beam profiles of photon fields between 2 × 2 and 10 × 10 cm2 were collected with both a CC13 ionization chamber and an EDGE diode detector on an Elekta Versa HD accelerator. These profiles were divided into three datasets (training, validation and test) to train and test a three-layer feedforward neural network. A sliding window was used to extract input data from the CC13-measured profiles. The neural network produced the deconvolved value at the center of the sliding window. The full deconvolved profile was obtained after the sliding window was moved over the measured profile from end to end. The EDGE-measured beam profiles were used as reference for the training, validation, and test. The number of input neurons, which equals the sliding window width, and the number of hidden neurons were optimized with a parametric sweeping method. A total of 135 neural networks were fully trained with the Levenberg-Marquardt backpropagation algorithm. The one with the best overall performance on the training and validation dataset was selected to test its generalization ability on the test dataset. The agreement between the neural network-deconvolved profiles and the EDGE-measured profiles was evaluated with two metrics: mean squared error (MSE) and penumbra width difference (PWD). RESULTS: Based on the two-dimensional MSE plots, the optimal combination of sliding window width of 15 and 5 hidden neurons was selected for the final neural network. Excellent agreement was achieved between the neural network-deconvolved profiles and the reference profiles in all three datasets. After deconvolution, the mean PWD reduced from 2.43 ± 0.26, 2.44 ± 0.36, and 2.46 ± 0.29 mm to 0.15 ± 0.15, 0.04 ± 0.03, and 0.14 ± 0.09 mm for the training, validation, and test dataset, respectively. CONCLUSIONS: We demonstrated the feasibility of photon beam profile deconvolution with a feedforward neural network in this work. The beam profiles deconvolved with a three-layer neural network had excellent agreement with diode-measured profiles.

Routine Adaptive Replanning of p16-Positive Stage N2b Oropharyngeal Cancer: Quality Improvement or Waste of Time?

PURPOSE/OBJECTIVE(S): To determine if routinely replanning patients treated for oropharyngeal cancer that is p16-positive and clinical neck stage N2b (AJCC 7th edition) is likely to result in dose changes that will improve patient outcomes to a meaningful degree. METHODS: In 10 consecutive patients treated with primary radiotherapy (RT) and concurrent weekly chemotherapy for p16-positive N2b oropharyngeal carcinoma, we prospectively evaluated dose changes from replanning for the final 4 or 2 weeks of RT of a 7-week RT program. RESULTS: Replanning for the final 4 or 2 weeks improved planning target volume coverage by an average of 4 and 2 percentage points, respectively. For all normal structures, the dose change was small (<1 Gy) with replanning. CONCLUSIONS: In patients with p16-positive N2b oropharynx cancer, the value of replanning RT is a small improvement in target coverage with minimal improvement in normal tissue sparing. In response to our study, some of the physicians in our group replan most node-positive oropharyngeal cancer cases while others think routine replanning is not valuable.

A universal parameterized gradient-based method for photon beam field size determination.

PURPOSE: To propose a universal, parameterized gradient-based method (PGM) for radiation field size determination. METHODS: The PGM locates the beam profile's edge by parameterizing its penumbra region with a modified sigmoid function where the inflection point can be determined in a closed form. The parametrization was validated with filter-flattened (FF), flattening-filter-free (FFF) and wedged profiles measured on two Elekta linac models (Synergy and Versa HD). Gamma analysis with the delta dose function set to zero was used to quantitatively assess the parameterization accuracy. Field sizes of FF beams were calculated with the PGM and the full width at half maximum (FWHM) methods for comparison. To assess the consistency of the PGM and the FWHM method with geometric scaling across different depths, the calculated field size at a reference depth was scaled to other depths and compared with the field sizes calculated from the measured profiles. The method was also validated against a maximum-slope method (MSM) with wedge and FFF profiles. We also evaluated the robustness of the three methods with respect to measurement noise, varying scanning step sizes, detector characteristics, and beam energy/modality. RESULTS: Small distance-to-agreement (0.02 ± 0.02 mm) between the measured and parameterized penumbra region was observed for all profiles. The differences between the field sizes calculated with the FWHM method and the PGM were consistent (0.9 ± 0.3 mm), with the FWHM method yielding larger values. With geometrical scaling, the PGM and the FWHM method produced maximum differences of 0.26 and 1.16 mm, respectively. For wedge and FFF beams, the mean differences relative to FF fields were 0.15 ± 0.09 mm and 0.57 ± 0.91 mm for the PGM and the MSM, respectively. The PGM was also found to produce more consistent results than the FWHM method and the MSM when measurement noise, scanning step size, detector characteristics, and beam energy/modality changed. CONCLUSION: The proposed PGM is universally applicable to all beam modalities (FF, wedge and FFF) for accurate field size determination. Compared to the FWHM and the MSM, it is more robust to variations in measurement condition and detection system.

Dose-volume toxicity modeling for de-intensified chemo-radiation therapy for HPV-positive oropharynx cancer.

BACKGROUND AND PURPOSE: The aim is to determine the radiobiological parameters of four popular normal tissue complication probability (NTCP) models that describe the dose-response relations of salivary glands and pharyngeal constrictors to the severity of patient reported xerostomia and dysphagia, respectively 6 and 12months post chemo-radiotherapy, furthermore, to evaluate the goodness-of-fit of the NTCP models for different combinations of glands and constrictors. MATERIAL AND METHODS: Forty-three patients were treated on a prospective multi-institutional phase II study (ClinicalTrials.gov, NCT01530997) assessing the efficacy of de-intensified chemoradiotherapy in patients with favorable risk, HPV-associated oropharyngeal squamous cell carcinoma. All patients received 60Gy intensity modulated radiotherapy with concurrent weekly intravenous cisplatinum. All patients reported severity of their xerostomia and dysphagia (pre- and post-treatment) using the patient reported outcome version of the CTCAE (PRO-CTCAE). A change in severity (from baseline) of ≥2 was considered clinically meaningful. The Lyman-Kutcher-Burman (LKB), Relative Seriality (RS), Logit, and Relative Seriality Logit (RSL) NTCP models were used to fit the patients' dose/volume data to changes in PRO-CTCAE severity of xerostomia and dysphagia (from baseline to 6 and 12months post-treatment). The correlation of the models with the patient outcomes was performed for different combinations of salivary glands and different sections of pharyngeal constrictors. The goodness-of-fit of the different models was assessed through the area under the receiver operating characteristic curve (AUC), maximum of the log-likelihood function, normal error distribution and Akaike information criterion (AIC). RESULTS: The dose/volume metrics of the combined contralateral (parotid+submandibular) glands appear to correlate best with xerostomia, at both 6- and 12-months. Among the different sections of pharyngeal constrictors, the dose/volume metrics of the superior pharyngeal constrictors appear to correlate best with dysphagia at 6months. The AUC values ranged from 0.72 to 0.85 in the case of xerostomia and 0.73 to 0.74 in the case of dysphagia over the different models. The four NTCP models showed similar goodness-of-fit. The differences in AIC between the different models were less than 2 and ranged within 0.7 and 0.8 in the cases of xerostomia and dysphagia, respectively. The calculated parameters of the LKB model were D50=26.9Gy, m=0.63, n=1.0 for the combined contralateral glands at 12months and D50=62.0Gy, m=0.10, n=0.49 for the superior pharyngeal constrictors at 6months. CONCLUSIONS: The values of the parameters of four NTCP models were determined for salivary glands and pharyngeal constrictors. All four models could fit the clinical data equally well. The NTCP predictions of the combined contralateral glands and superior pharyngeal constrictors showed the best correlation with the patient reported outcomes of xerostomia and dysphagia, respectively.

A hybrid phantom Monte Carlo-based method for historical reconstruction of organ doses in patients treated with cobalt-60 for Hodgkin's lymphoma.

Historical radiotherapy treatment plans lack 3D images sets required for estimating mean organ doses to patients. Alternatively, Monte Carlo-based models of radiotherapy devices coupled with whole-body computational phantoms can permit estimates of historical in-field and out-of-field organ doses as needed for studies associating radiation exposure and late tissue toxicities. In recreating historical patient treatments with 60Co based systems, the major components to be modeled include the source capsule, surrounding shielding layers, collimators (both fixed and adjustable), and trimmers as needed to vary field size. In this study, a computational model and experimental validation of the Theratron T-1000 are presented. Model validation is based upon in-field commissioning data collected at the University of Florida, published out-of-field data from the British Journal of Radiology (BJR) Supplement 25, and out-of-field measurements performed at the University of Wisconsin's Accredited Dosimetry Calibration Laboratory (UWADCL). The computational model of the Theratron T-1000 agrees with central axis percentage depth dose data to within 2% for 6 × 6 to 30 × 30 cm2 fields. Out-of-field doses were found to vary between 0.6% to 2.4% of central axis dose at 10 cm from field edge and 0.42% to 0.97% of central axis dose at 20 cm from the field edge, all at 5 cm depth. Absolute and relative differences between computed and measured out-of-field doses varied between ±2.5% and ±100%, respectively, at distances up to 60 cm from the central axis. The source-term model was subsequently combined with patient-morphometry matched computational hybrid phantoms as a method for estimating in-field and out-of-field organ doses for patients treated for Hodgkin's Lymphoma. By changing field size and position, and adding patient-specific field shaping blocks, more complex historical treatment set-ups can be to recreated, particularly those for which 2D or 3D image sets are unavailable.

Angular under-sampling effect on VMAT dose calculation: An analysis and a solution strategy.

PURPOSE: Most VMAT algorithms compute the dose on discretized apertures with small angular separations for practical reasons. However, machines deliver the VMAT dose with a continuously moving MLC and gantry and a continuously changing dose rate. The computed dose can deviate from the delivered dose, especially if no, or loose, MLC movement constraints are applied for the VMAT optimization. The goal of this paper is to establish a simplified mathematical model to analyze the discrepancy between the VMAT plan calculation dose and the delivered dose and to provide a reasonable solution for clinical implementation. METHODS: A simplified metric is first introduced to describe the discrepancy between doses computed with discretized apertures and a continuous delivery model. The delivery fluences were formed separately for six different leaf movement scenarios. The formula was then rewritten in a more general form. The correlation between discretized and continuous fluence is summarized using this general form. The Fourier analysis for the impacts from three separate factors - dose kernel width, aperture width, aperture distance - to the dose discrepancy is also presented in order to provide insight into the dose discrepancy caused by under-sampling in the frequency domain. Finally, a weighting-based interpolation (WBI) algorithm, which can improve the aperture interpolation efficiency, is proposed. The associated evaluation methods and criteria for the proposed algorithm are also given. RESULTS: The comparisons between the WBI algorithm and the equal angular interpolation (EAI) method suggested that the proposed algorithm has a great advantage with regard to aperture number efficiency. To achieve a 90% gamma passing rate using the dose computed with apertures generated with 0.5° EAI, with the initial optimization apertures as the standard for the comparison, the WBI needs only 66% and 54% of the aperture numbers that the EAI method needs for a 2° and a 4° angular separation of the VMAT optimization, respectively. The results also suggested that the weighted dose error index value, Θ, can be used as a stopping criterion for an interpolation algorithm, e.g., WBI or EAI, or as an indicator for sampling level evaluations. The phantom results indicate that the gamma passing rate decreases with increasing depth, from the phantom surface to the iso center, for the plans computed with under-sampled apertures. No obvious variation trends were observed for the plans computed with well-sampled apertures. CONCLUSIONS: The mathematical analysis suggests that the dose discrepancies due to under-sampling are strongly correlated with the aperture width, the distance between apertures, and the width of the dose kernel. The WBI algorithm proves to be an efficient aperture interpolation strategy and is useful for dose computation of VMAT plans.

Efficient independent planar dose calculation for FFF IMRT QA with a bivariate Gaussian source model.

The aim of this study is to perform a direct comparison of the source model for photon beams with and without flattening filter (FF) and to develop an efficient independent algorithm for planar dose calculation for FF-free (FFF) intensity-modulated radiotherapy (IMRT) quality assurance (QA). The source model consisted of a point source modeling the primary photons and extrafocal bivariate Gaussian functions modeling the head scatter, monitor chamber backscatter, and collimator exchange effect. The model parameters were obtained by minimizing the difference between the calculated and measured in-air output factors (Sc ). The fluence of IMRT beams was calculated from the source model using a backprojection and integration method. The off-axis ratio in FFF beams were modeled with a fourth degree polynomial. An analytical kernel consisting of the sum of three Gaussian functions was used to describe the dose deposition process. A convolution-based method was used to account for the ionization chamber volume averaging effect when commissioning the algorithm. The algorithm was validated by comparing the calculated planar dose distributions of FFF head-and-neck IMRT plans with measurements performed with a 2D diode array. Good agreement between the measured and calculated Sc was achieved for both FF beams (<0.25%) and FFF beams (<0.10%). The relative contribution of the head-scattered photons reduced by 34.7% for 6 MV and 49.3% for 10 MV due to the removal of the FF. Superior agreement between the calculated and measured dose distribution was also achieved for FFF IMRT. In the gamma comparison with a 2%/2 mm criterion, the average passing rate was 96.2 ± 1.9% for 6 MV FFF and 95.5 ± 2.6% for 10 MV FFF. The efficient independent planar dose calculation algorithm is easy to implement and can be valuable in FFF IMRT QA.