Monte Carlo modeling of the Elekta Versa HD and patient dose calculation with EGSnrc/BEAMnrc.

INTRODUCTION: Numerous studies have proven the Monte Carlo method to be an accurate means of dose calculation. Although there are several commercial Monte Carlo treatment planning systems (TPSs), some clinics may not have access to these resources. We present a method for routine, independent patient dose calculations from treatment plans generated in a commercial TPS with our own Monte Carlo model using free, open-source software. MATERIALS AND METHODS: A model of the Elekta Versa HD linear accelerator was developed using the EGSnrc codes. A MATLAB script was created to take clinical patient plans and convert the DICOM RTP files into a format usable by EGSnrc. Ten patients' treatment plans were exported from the Monaco TPS to be recalculated using EGSnrc. Treatment simulations were done in BEAMnrc, and doses were calculated using Source 21 in DOSXYZnrc. Results were compared to patient plans calculated in the Monaco TPS and evaluated in Verisoft with a gamma criterion of 3%/2 mm. RESULTS: Our Monte Carlo model was validated within 1%/1-mm accuracy of measured percent depth doses and profiles. Gamma passing rates ranged from 82.1% to 99.8%, with 7 out of 10 plans having a gamma pass rate over 95%. Lung and prostate patients showed the best agreement with doses calculated in Monaco. All statistical uncertainties in DOSXYZnrc were less than 3.0%. CONCLUSION: A Monte Carlo model for routine patient dose calculation was successfully developed and tested. This model allows users to directly recalculate DICOM RP files containing patients' plans that have been exported from a commercial TPS.

Clinical assessment of deep learning-based uncertainty maps in lung cancer segmentation.

Objective. Prior to radiation therapy planning, accurate delineation of gross tumour volume (GTVs) and organs at risk (OARs) is crucial. In the current clinical practice, tumour delineation is performed manually by radiation oncologists, which is time-consuming and prone to large inter-observer variability. With the advent of deep learning (DL) models, automated contouring has become possible, speeding up procedures and assisting clinicians. However, these tools are currently used in the clinic mostly for contouring OARs, since these systems are not reliable yet for contouring GTVs. To improve the reliability of these systems, researchers have started exploring the topic of probabilistic neural networks. However, there is still limited knowledge of the practical implementation of such networks in real clinical settings.Approach. In this work, we developed a 3D probabilistic system that generates DL-based uncertainty maps for lung cancer CT segmentations. We employed the Monte Carlo (MC) dropout technique to generate probabilistic and uncertainty maps, while the model calibration was evaluated by using reliability diagrams. A clinical validation was conducted in collaboration with a radiation oncologist to qualitatively assess the value of the uncertainty estimates. We also proposed two novel metrics, namely mean uncertainty (MU) and relative uncertainty volume (RUV), as potential indicators for clinicians to assess the need for independent visual checks of the DL-based segmentation. Main results. Our study showed that uncertainty mapping effectively identified cases of under or over-contouring. Although the overconfidence of the model, a strong correlation was observed between the clinical opinion and MU metric. Moreover, both MU and RUV revealed high AUC values in discretising between low and high uncertainty cases.Significance. Our study is one of the first attempts to clinically validate uncertainty estimates in DL-based contouring. The two proposed metrics exhibited promising potential as indicators for clinicians to independently assess the quality of tumour delineation.

An anomaly detector as a clinical decision support system for parotid gland delineations.

Purpose. Auto-contouring (AC) is rapidly becoming standard practice for OAR contouring. However, in clinical practice, clinicians still need to manually check and correct contours. Anomaly detection systems (ADS) can aid the clinical decision process by suggesting which structures require corrections or not, greatly enhancing the value of AC. The purpose of this work is to develop and evaluate a decision support system for detecting anomalies in the case of parotid gland delineations. METHODS: Head and neck parotid gland delineations (1037 right, 1038 left), were retrieved from the Netherlands Cancer Institute (NKI) database. Morphological and image-based features were extracted from each patient's CT and structure set. An isolation forest model was initially trained on 70% of the data, of which 10% had synthetically generated anomalies and validated on the remaining 30% of clinical data. The ADS was tested on an independent set of 250 patients (Normal: 174, Anomalies: 76) and on a clinical autocontouring software. RESULTS: Applied to the validation set, the ADS system resulted in area under the curve (AUC) values of 0.93 and 0.94 for the parotid left and right respectively. Image features appeared more important than morphological, but using all features resulted marginally in the best model. Applied to the test set the ADS system reached an accuracy level of 0.83 and 0.81 for the parotid left and right respectively. The ADS was particularly sensitive to uniform expansions/contractions, misplacements, extra/missing slices and anisotropic over-contouring. CONCLUSION: Anomaly detection can serve as a powerful contour quality assurance tool, especially for cases of organ misplacement and over-contouring.

Positional and angular tracking of HDR 192 Ir source for brachytherapy quality assurance using radiochromic film dosimetry.

PURPOSE: To quantify and verify the dosimetric impact of high-dose rate (HDR) source positional uncertainty in brachytherapy, and to introduce a model for three-dimensional (3D) position tracking of the HDR source based on a two-dimensional (2D) measurement. This model has been utilized for the development of a comprehensive source quality assurance (QA) method using radiochromic film (RCF) dosimetry including assessment of different digitization uncertainties. METHODS: An algorithm was developed and verified to generate 2D dose maps of the mHDR-V2 192 Ir source (Elekta, Veenendaal, Netherlands) based on the AAPM TG-43 formalism. The limits of the dosimetric error associated with source (0.9 mm diameter) positional uncertainty were evaluated and experimentally verified with EBT3 film measurements for 6F (2.0 mm diameter) and 4F (1.3 mm diameter) size catheters at the surface (4F, 6F) and 10 mm further (4F only). To quantify this uncertainty, a source tracking model was developed to incorporate the unique geometric features of all isodose lines (IDLs) within any given 2D dose map away from the source. The tracking model normalized the dose map to its maximum, then quantified the IDLs using blob analysis based on features such as area, perimeter, weighted centroid, elliptic orientation, and circularity. The Pearson correlation coefficients (PCCs) between these features and source coordinates (x, y, z, θy , θz ) were calculated. To experimentally verify the accuracy of the tracking model, EBT3 film pieces were positioned within a Solid Water® (SW) phantom above and below the source and they were exposed simultaneously. RESULTS: The maximum measured dosimetric variations on the 6F and 4F catheter surfaces were 39.8% and 36.1%, respectively. At 10 mm further, the variation reduced to 2.6% for the 4F catheter which is in agreement with the calculations. The source center (x, y) was strongly correlated with the low IDL-weighted centroid (PCC = 0.99), while the distance to source (z) was correlated with the IDL areas (PCC = 0.96) and perimeters (PCC = 0.99). The source orientation θy was correlated with the difference between high and low IDL-weighted centroids (PCC = 0.98), while θz was correlated with the elliptic orientation of the 60-90% IDLs (PCC = 0.97) for a maximum distance of z = 5 mm. Beyond 5 mm, IDL circularity was significant, therefore limiting the determination of θz (PCC ≤ 0.48). The measured positional errors from the film sets above and below the source indicated a source position at the bottom of the catheter (-0.24 ± 0.07 mm). CONCLUSIONS: Isodose line features of a 2D dose map away from the HDR source can reveal its spatial coordinates. RCF was shown to be a suitable dosimeter for source tracking and dosimetry. This technique offers a novel source QA method and has the potential to be used for QA of commercial and customized applicators.

Size-specific dose estimations for pediatric chest, abdomen/pelvis and head CT scans with the use of GATE.

PURPOSE: The purpose of this study is to create an organ dose database for pediatric individuals undergoing chest, abdomen/pelvis, and head computed tomography (CT) examinations, and to report the differences in absorbed organ doses, when anatomical differences exist for pediatric patients. METHODS: The GATE Monte Carlo (MC) toolkit was used to model the GE BrightSpeed Elite CT model. The simulated scanner model was validated with the standard Computed Tomography Dose Index (CTDI) head phantom. Twelve computational models (2.1-14 years old) were used. First, contributions to effective dose and absorbed doses per CTDIvol and per 100 mAs were estimated for all organs. Then, doses per CTDIvol were correlated with patient model weight for the organs inside the scan range for chest and abdomen/pelvis protocols. Finally, effective doses per dose-length product (DLP) were estimated and compared with the conventional conversion k-factors. RESULTS: The system was validated against experimental CTDIw measurements. The doses per CTDIvol and per 100 mAs for selected organs were estimated. The magnitude of the dependency between the dose and the anatomical characteristics was calculated with the coefficient of determination at 0.5-0.7 for the internal scan organs for chest and abdomen/pelvis protocols. Finally, effective doses per DLP were compared with already published data, showing discrepancies between 13 and 29% and were correlated strongly with the total weight (R2 > 0.8) for the chest and abdomen protocols. CONCLUSIONS: Big differences in absorbed doses are reported even for patients of similar age or same gender, when anatomical differences exist on internal organs of the body.

Modeling the primary source intensity distribution: reconstruction and inter-comparison of six Varian TrueBeam sources.

The primary source size is one of the most important beam model parameters in small photon fields. In this work we apply a recently suggested reconstruction technique to characterize the primary source of 6 Varian TrueBeam (TB) linacs. A series of photon fluence profile measurements were performed on 6 Varian TB linacs in the crossplane and inplane orientation using radiochromic film in air and a 2 mm Pb foil as a build-up layer. An image reconstruction algorithm was then applied, based on the maximum likelihood expectation-maximization (MLEM) algorithm, to estimate the source distribution. The method iteratively ray-traces photons from the source plane to the measurement plane to extract source profile corrections. The technique was first benchmarked using a Monte Carlo (MC) model of a Varian TrueBeam with known input Gaussian source sizes. The robustness of the suggested technique was also tested by randomly sampling different combinations of source and field size values and repeating the reconstruction. At the MC benchmarking stage the MLEM reconstruction algorithm was capable of reproducing the Gaussian shape with a RMSE less than 4.0%, while the reconstructed source size (FWHM) and field size were determined with an accuracy level of 0.14 mm and 0.10 mm respectively. Experimentally, the reconstructed TB sources presented FWHM values between 1.02-1.5 mm ([Formula: see text]-0.18 mm) and 1.08-1.42 mm ([Formula: see text]-0.13 mm) in the crossplane and inplane orientations respectively. All TB sources studied in this work can be considered symmetric within uncertainties with the exception of one. The source distribution presented systematic deviations from a Gaussian distribution mostly in the lower tail region. Multi-parameter functional forms, such as Pearson VII or double Gaussian presented improvements in modeling the source in this region, but increase the model complexity. The reconstructed sources measured in this work can serve as reference values for commissioning beam models in small fields and set upper and lower thresholds values of the expected source size for a TB linac.

Monte Carlo simulations of different CT X-ray energy spectra within CTDI phantom and the influence of its changes on radiochromic film measurements.

PURPOSE: In this work we use Monte Carlo simulations to investigate change in Computed tomography (CT) X-ray energy spectra between exposures in air and within CT dose index (CTDI) phantom. While the results of these simulations will be relevant when measuring CTDI with any dosimeter, we apply the appropriate beam quality change correction for CTDI measurements using XR-QA2 model GafChromic™ film. METHODS: Dose profiles were measured with film strips, sandwiched between acrylic rods cut in half, placed within CTDI phantoms and scanned before and after irradiation with document scanner in reflective mode. Reference dosimetry system was calibrated in terms of air kerma in air, which was converted into absorbed dose using ratio of mass-energy absorption coefficients water-to-air for a given beam quality, following the AAPM TG-61 protocol. RESULTS: Beam qualities for all film positions within CTDI phantom show beam softening for HVLs above 6 mm Al and beam hardening for HVLs bellow 6 mm Al. Calculated CTDI values using HVL in air for all CTDI positions, and those calculated using the appropriate calibration curves based on beam quality correction show for Head CTDI phantom differences ranging from 0.3% to 2.1% and for Body CTDI phantom from 2.5% to 5.7%. CONCLUSIONS: We describe method for CTDI measurements using radiochromic film dosimetry protocol corrected by the beam quality change within the phantom. Our results show differences in CTDI measurements of up to 5.7% when compared to using film calibration curves for beam quality in air.

Dose-response linearization in radiochromic film dosimetry based on multichannel normalized pixel value with an integrated spectral correction for scanner response variations.

PURPOSE: To introduce a model that reproducibly linearizes the response from radiochromic film (RCF) dosimetry systems at extended dose range. To introduce a correction method, generated from the same scanned images, which corrects for scanner temporal response variation and scanner bed inhomogeneity. METHODS: Six calibration curves were established for different lot numbers of EBT3 GAFCHROMIC™ film model based on four EPSON scanners [10000XL (2 units), 11000XL, 12000XL] at three different centers. These films were calibrated in terms of absorbed dose to water based on TG51 protocol or TRS398 with dose ranges up to 40 Gy. The film response was defined in terms of a proposed normalized pixel value ( n P V RGB ) as a summation of first-order equations based on information from red, green, and blue channels. The fitting parameters of these equations are chosen in a way that makes the film response equal to dose at the time of calibration. An integrated set of correction factors (one per color channel) was also introduced. These factors account for the spatial and temporal changes in scanning states during calibration and measurements. The combination of n P V RGB and this "fingerprint" correction formed the basis of this new protocol and it was tested against net optical density ( n e t O D X = R , G , B ) single-channel dosimetry in terms of accuracy, precision, scanner response variability, scanner bed inhomogeneity, noise, and long-term stability. RESULTS: Incorporating multichannel features (RGB) into the normalized pixel value produced linear response to absorbed dose (slope of 1) in all six RCF dosimetry systems considered in this study. The "fingerprint" correction factors of each of these six systems displayed unique patterns at the time of calibration. The application of n P V RGB to all of these six systems could achieve a level of accuracy of ± 2.0% in the dose range of interest within modeled uncertainty level of 2.0%-3.0% depending on the dose level. Consistent positioning of control and measurement film pieces and integrating the multichannel correction into the response function formalism mitigated possible scanner response variations of as much as ± 10% at lower doses and scanner bed inhomogeneity of ± 8% to the established level of uncertainty at the time of calibration. The system was also able to maintain the same level of accuracy after 3 and 6 months post calibration. CONCLUSIONS: Combining response linearity with the integrated correction for scanner response variation lead to a sustainable and practical RCF dosimetry system that mitigated systematic response shifts and it has the potential to reduce errors in reporting relative information from the film response.

Image quality for radiotherapy CT simulators with different scanner bore size.

PURPOSE: We compare image quality parameters derived from phantom images taken on three commercially available radiotherapy CT simulators. To make an unbiased evaluation, we assured images were obtained with the same surface dose measured using XR-QA2 model GafChromic™ film placed at the imaging phantom surface for all three CT-simulators. METHODS: Radiotherapy CT simulators GE LS 16, Philips Brilliance Big Bore, and Toshiba Aquilion LB were compared in terms of spatial resolution, low contrast detectability, image uniformity, and contrast to noise ratio using CATPHAN-504 phantom, scanned with Head and Pelvis protocols. Dose was measured at phantom surface, with CT scans repeated until doses on all scanners were within 2%. RESULTS: In terms of spatial resolution, the GE simulator appears slightly better, while Philips CT images are superior in terms of SNR for both scanning protocols. The CNR results show that Philips CT images appear to be better, except for high Z material, while Toshiba appears to fit in between the two simulators. CONCLUSIONS: While the image quality parameters for three RT CT simulators show comparable results, the scanner bore size is of vital importance in various radiotherapy applications. Since the image quality is a function of a large number of confounding parameters, any loss in image quality due to scanner bore size could be compensated by the appropriate choice of scanning parameters, including the exposure and by balancing between the additional imaging dose to the patient and high image quality required in highly conformal RT techniques.

An investigation into the INTRABEAM miniature x-ray source dosimetry using ionization chamber and radiochromic film measurements.

PURPOSE: Intraoperative radiotherapy using The INTRABEAM System (Carl Zeiss Meditec AG, Jena, Germany), a miniature low-energy x-ray source, has proven to be an effective modality in the treatment of breast cancer. However, some uncertainties remain in its dosimetry. In this work, we investigated the INTRABEAM system dosimetry by performing ionization chamber and radiochromic film measurements of absorbed dose in a water phantom. METHODS: Ionization chamber measurements were performed with a PTW 34013 parallel-plate soft x-ray chamber at source to detector distances of 5 to 30 mm calculated using (a) the dose formula consistent with the TARGIT breast protocol (TARGIT), (b) the formula recommended by the manufacturer (Zeiss), and (c) the recently proposed CQ formalism of Watson et al. (Physics in Medicine & Biology, 2018;63:015016) EBT3 Gafchromic film measurements were made at the same depths in water. To account for the energy dependence of EBT3 film, multiple dose response calibration curves were employed across a range of photon beam qualities relevant to the INTRABEAM spectrum in water. RESULTS: At all depths investigated, the TARGIT dose was significantly lower than that measured by the Zeiss and CQ methods, as well as film. These dose differences ranged from 14% to as large as 80%. In general, the doses measured by film, and the Zeiss and CQ methods were in good agreement to within measurement uncertainties (5-6%). CONCLUSIONS: These results suggest that the TARGIT dose underestimates the physical dose to water from the INTRABEAM source. Understanding the correlation between the TARGIT and physical dose is important for any studies wishing to make dosimetric comparisons between the INTRABEAM and other radiation emitting devices.

Technical Note: Response time evolution of XR-QA2 GafChromic™ film models.

PURPOSE: To evaluate the response of the newest XR-QA2 GafChromic™ film model in terms of postexposure signal growth and energy response in comparison with the older XR-QA (Version 2) model. METHODS: Pieces of film were irradiated to air kerma in air values up to 12 cGy with several beam qualities (5.3-8.25 mm Al) commonly used for CT scanning. Film response was scored in terms of net reflectance from scanned film images at various points in time postirradiation ranging from 1 to 7 days and 5 months postexposure. To reconstruct the measurement signal changes with postirradiation delay, we irradiated one film piece and then scanned it at different point times starting from 2" min and up to 3 days postexposure. RESULTS: For all beam qualities and dose range investigated, it appears that the XR-QA2 film signal completely saturated after 15 h. Compared to 15 h postirradiation scanning time, the observed variation in net reflectance were 3%, 2%, and 1% for film scanned 2" min, 20 min, and 3 h after exposure, respectively, which is well within the measurement uncertainty of the XR-QA2 based reference radiochromic film dosimetry system. A comparison between the XR-QA (Version 2) and the XR-QA2 film response after several months (relative to their responses after 24 h) show differences in up to 8% and 1% for each film model respectively. CONCLUSIONS: The replacement of cesium bromide in the older XR-QA (Version 2) film model with bismuth oxide in the newer XR-QA2 film, while keeping the same single sensitive layer structure, lead to a significantly more stable postexposure response.

Experimental investigation on the accuracy of plastic scintillators and of the spectrum discrimination method in small photon fields.

PURPOSE: To investigate the accuracy of output factor measurements using a commercial (Exradin W1, SI) and a prototype, "in-house" developed, plastic scintillation dosimeter (PSD) in small photon fields. METHODS: Repetitive detector-specific output factor OFdet measurements were performed in water (parallel to the CAX) using two W1 PSDs (SI), a PTW microLion, a PTW microDiamond and an unshielded diode D1V (SI) to which Monte Carlo calculated corrections factors were applied. Four sets of repetitive measurements were performed with the W1 PSD positioned parallel and perpendicular to the CAX, each set on a different day, and with analytically calculated volume averaging corrections applied. The W1 OFdet measurements were compared to measurements using an "in-house" developed PSD in water (CHUQ) and both were validated against a previously commissioned Monte Carlo beam model in small photon fields. The performance of the spectrum discrimination calibration procedure was evaluated under different fiber orientations and wavelength threshold choices and the impact on the respective OFdet was reported. RESULTS: For all detectors in the study an excellent agreement was observed down to a field size of 1 × 1 cm2 . For the smallest field size of 0.5 × 0.5 cm2 , the W1 PSDs presented OFdet readings higher by 3.8 to 5.0% relative to the mean corrected OFdet of the rest of the detectors and by 5.8 to 6.1% relative to the CHUQ PSD. The repetitive W1 OFdet measurements in water (parallel CAX) were higher by 3.9% relative to the OFdet measurements in Solid WaterTM (perpendicular CAX) even after volume averaging corrections were applied, indicating a potential fiber orientation dependency in small fields. Uncertainties in jaw and detector repositioning as well as source variations with time were estimated to be less than 0.9% (1 σ) for the W1 under both orientations. The CHUQ PSD agreed with the MC dose calculations in water, for the smallest field size, within 1.1-1.7% before any corrections and within 0.3-0.8% after volume averaging corrections. The spectrum discrimination method provided reproducible Cherenkov spectra under the different calibration set-ups with noisier spectra extracted if the calibration is performed in water and parallel to the CAX. The impact of fiber orientation and wavelength threshold during calibration on OFdet was in general minimal. CONCLUSIONS: Clinically relevant differences were observed between similar scintillator dosimeters in photon fields smaller than 1 ×  1 cm2 . Further research on PSDs is needed that can explain the origin of these differences especially related to the Cherenkov spectrum dependencies on the optical fiber technical characteristics.

A protocol for EBT3 radiochromic film dosimetry using reflection scanning.

PURPOSE: To evaluate the performance of the EBT3 radiochromic film dosimetry system using reflection measurements and to suggest a calibration protocol for precise and accurate reflection film dosimetry. METHODS: A set of 14 Gafchromic EBT3 film pieces were irradiated to various doses ranging from 0 to 8 Gy and subsequently scanned using both the reflection and transmission mode. Scanning resolution varied from 50 to 508 dpi (0.5-0.05 mm/pixel). Both the red and green color channels of scanned images were used to relate the film response to the dose. A sensitivity, uncertainty, and accuracy analysis was performed for all scanning modes and color channels. The total uncertainty, along with the fitting and experimental uncertainty components, was identified and analyzed. A microscope resolution target was used to evaluate possible resolution losses under reflection scanning. The calibration range was optimized for reflection scanning in the low (<2 Gy) and high (>2 Gy) dose regions based on the reported results. RESULTS: Reflection scanning using the red channel exhibited the highest sensitivity among all modes, being up to 150% higher than transmission mode in the red channel for the lowest dose level. Furthermore, there was no apparent loss in resolution between the two modes. However, higher uncertainties and reduced accuracy were observed for the red channel under reflection mode, especially at dose levels higher than 2 Gy. These uncertainties were mainly attributed to saturation effects which were translated in poor fitting results. By restricting the calibration to the 0-2 Gy dose range, the situation is reversed and the red reflection mode was superior to the transmission mode. For higher doses, the green channel in reflection mode presented comparable results to the red transmission. CONCLUSIONS: A two-color reflection scanning protocol can be suggested for EBT3 radiochromic film dosimetry using the red channel for doses less than 2 Gy and the green channel for higher doses. The precision and accuracy are significantly improved in the low dose region following such a protocol.

A source model for modulated electron radiation therapy using dynamic jaw movements.

PURPOSE: The development of fast and accurate source models (SMs) might be of crucial importance for the future clinical implementation of modulated electron radiation therapy (MERT). In this study, a SM is presented for reconstructing phase-space information of modulated electron beams using a few-leaf electron collimator (FLEC) and the photon jaws. METHODS: During a FLEC-based delivery, two collimation devices (jaws and FLEC) modulate the electron beam characteristics dynamically. The SM separates the beam into a primary and a scattered component. The primary component is derived by a fast Monte Carlo (MC) transport calculation in air using the EGSnrc/BEAMnrc code. The scattered beam is modeled analytically. The accelerator was decomposed into its individual leaf components and the scattered beam was characterized at various levels of the accelerator. Scattered particles are assigned an energy and position by sampling pre-calculated probability distributions. The direction is estimated by geometrical arguments. Particles were assumed to emerge from tunable virtual sources on the side of each collimator leaf. A leaf-hit algorithm was developed to dynamically reject particles that are incident on any collimating leaf. Electron transport in air between the two collimation levels was calculated based on a MC-modified version of the Fermi-Eyges scattering theory. Correlations between direction and position were observed and taken into account at the final collimation level. RESULTS: To validate the model, reconstructed phase-space data were compared with the full accelerator MC phase-space data. The model accurately reproduced the beam characteristics and preserved important correlations. Depth and profile dose distributions in water were derived for square, rectangular, and off-axis field sizes and for a range of clinical energies. Discrepancies in the dose distributions and dose output were within 3% in all cases. CONCLUSIONS: Fast and accurate SMs open the possibility for fast treatment planning in MERT, based on an inverse optimization MC treatment planning scheme.