Deep learning based MLC aperture and monitor unit prediction as a warm start for breast VMAT optimisation.

Objective. Automated treatment planning today is focussed on non-exact, two-step procedures. Firstly, dose-volume histograms (DVHs) or 3D dose distributions are predicted from the patient anatomy. Secondly, these are converted in multi-leaf collimator (MLC) apertures and monitor units (MUs) using a generic optimisation to obtain the final treatment plan. In contrast, we present a method to predict volumetric modulated arc therapy (VMAT) MLC apertures and MUs directly from patient anatomy using deep learning. The predicted plan is then provided as initialisation to the optimiser for fine-tuning.Approach. 148 patients (training: 101; validation: 23; test: 24), treated for right breast cancer, are replanned to obtain a homogeneous database of 3-arc VMAT plans (PTVBreast: 45.57 Gy; PTVBoost: 55.86 Gy) according to the clinical protocol, using RapidPlanTMwith automatic optimisation and extended convergence mode (clinical workflow). Projections of the CT and contours are created along the beam's eye view of all control points and given as input to a U-net type convolutional neural networks (CNN). The output are the MLC aperture and MU for all control points, from which a DICOM RTplan is built. This is imported and further optimised in the treatment planning system using automatic optimisation without convergence mode, with clinical PTV objectives and organs-at-risk (OAR) objectives based on the DVHs calculated from the imported plan (CNN workflow).Main results. Mean dose differences between the clinical and CNN workflow over the test set are 0.2 ± 0.5 Gy atD95%and 0.6 ± 0.4 Gy atD0.035ccof PTVBreastand -0.4 ± 0.3 Gy atD95%and 0.7 ± 0.3 Gy atD0.035ccof PTVBoost. For the OAR, they are -0.2 ± 0.2 Gy forDmean,heartand 0.04 ± 0.8 Gy forDmean,ipsilateral lung. The mean computation time is 60 and 25 min respectively.Significance. VMAT optimisation can be initialised by MLC apertures and MUs, directly predicted from patient anatomy using a CNN, reducing planning time with more than half while maintaining clinically acceptable plans. This procedure puts the planner in a supervising role over an AI-based treatment planning workflow.

The NCS code of practice for the quality assurance of treatment planning systems (NCS-35).

A subcommittee of the Netherlands Commission on Radiation Dosimetry (NCS) was initiated in 2018 with the task to update and extend a previous publication (NCS-15) on the quality assurance of treatment planning systems (TPS) (Bruinviset al2005). The field of treatment planning has changed considerably since 2005. Whereas the focus of the previous report was more on the technical aspects of the TPS, the scope of this report is broader with a focus on a department wide implementation of the TPS. New sections about education, automated planning, information technology (IT) and updates are therefore added. Although the scope is photon therapy, large parts of this report will also apply to all other treatment modalities. This paper is a condensed version of these guidelines; the full version of the report in English is freely available from the NCS website (http://radiationdosimetry.org/ncs/publications). The paper starts with the scope of this report in relation to earlier reports on this subject. Next, general aspects of the commissioning process are addressed, like e.g. project management, education, and safety. It then focusses more on technical aspects such as beam commissioning and patient modeling, dose representation, dose calculation and (automated) plan optimisation. The final chapters deal with IT-related subjects and scripting, and the process of updating or upgrading the TPS.

Clinical evaluation of a deep learning model for segmentation of target volumes in breast cancer radiotherapy.

PURPOSE/OBJECTIVE(S): Precise segmentation of clinical target volumes (CTV) in breast cancer is indispensable for state-of-the art radiotherapy. Despite international guidelines, significant intra- and interobserver variability exists, negatively impacting treatment outcomes. The aim of this study is to evaluate the performance and efficiency of segmentation of CTVs in planning CT images of breast cancer patients using a 3D convolutional neural network (CNN) compared to the manual process. MATERIALS/METHODS: An expert radiation oncologist (RO) segmented all CTVs separately according to international guidelines in 150 breast cancer patients. This data was used to create, train and validate a 3D CNN. The network's performance was additionally evaluated in a test set of 20 patients. Primary endpoints are quantitative and qualitative analysis of the segmentation data generated by the CNN for each level specifically as well as for the total PTV to be irradiated. The secondary endpoint is the evaluation of time efficiency. RESULTS: In the test set, segmentation performance was best for the contralateral breast and the breast CTV and worst for Rotter's space and the internal mammary nodal (IMN) level. Analysis of impact on PTV resulted in non-significant over-segmentation of the primary PTV and significant under-segmentation of the nodal PTV, resulting in slight variations of overlap with OARs. Guideline consistency improved from 77.14% to 90.71% in favor of CNN segmentation while saving on average 24 minutes per patient with a median time of 35 minutes for pure manual segmentation. CONCLUSION: 3D CNN based delineation for breast cancer radiotherapy is feasible and performant, as scored by quantitative and qualitative metrics.

Expanding the medical physicist curricular and professional programme to include Artificial Intelligence.

PURPOSE: To provide a guideline curriculum related to Artificial Intelligence (AI), for the education and training of European Medical Physicists (MPs). MATERIALS AND METHODS: The proposed curriculum consists of two levels: Basic (introducing MPs to the pillars of knowledge, development and applications of AI, in the context of medical imaging and radiation therapy) and Advanced. Both are common to the subspecialties (diagnostic and interventional radiology, nuclear medicine, and radiation oncology). The learning outcomes of the training are presented as knowledge, skills and competences (KSC approach). RESULTS: For the Basic section, KSCs were stratified in four subsections: (1) Medical imaging analysis and AI Basics; (2) Implementation of AI applications in clinical practice; (3) Big data and enterprise imaging, and (4) Quality, Regulatory and Ethical Issues of AI processes. For the Advanced section instead, a common block was proposed to be further elaborated by each subspecialty core curriculum. The learning outcomes were also translated into a syllabus of a more traditional format, including practical applications. CONCLUSIONS: This AI curriculum is the first attempt to create a guideline expanding the current educational framework for Medical Physicists in Europe. It should be considered as a document to top the sub-specialties' curriculums and adapted by national training and regulatory bodies. The proposed educational program can be implemented via the European School of Medical Physics Expert (ESMPE) course modules and - to some extent - also by the national competent EFOMP organizations, to reach widely the medical physicist community in Europe.

Benefits of deep learning for delineation of organs at risk in head and neck cancer.

PURPOSE/OBJECTIVE: Precise delineation of organs at risk (OARs) in head and neck cancer (HNC) is necessary for accurate radiotherapy. Although guidelines exist, significant interobserver variability (IOV) remains. The aim was to validate a 3D convolutional neural network (CNN) for semi-automated delineation of OARs with respect to delineation accuracy, efficiency and consistency compared to manual delineation. MATERIAL/METHODS: 16 OARs were manually delineated in 15 new HNC patients by two trained radiation oncologists (RO) independently, using international consensus guidelines. OARs were also automatically delineated by applying the CNN and corrected as needed by both ROs separately. Both delineations were performed two weeks apart and blinded to each other. IOV between both ROs was quantified using Dice similarity coefficient (DSC) and average symmetric surface distance (ASSD). To objectify network accuracy, differences between automated and corrected delineations were calculated using the same similarity measures. RESULTS: Average correction time of the automated delineation was 33% shorter than manual delineation (23 vs 34 minutes) (p < 10-6). IOV improved significantly with network initialisation for nearly all OARs (p < 0.05), resulting in decreased ASSD averaged over all OARs from 1.9 to 1.2 mm. The network achieved an accuracy of 90% and 84% DSC averaged over all OARs for RO1 and RO2 respectively, with an ASSD of 0.7 and 1.5 mm, which was in 93% and 73% of the cases lower than the IOV. CONCLUSION: The CNN developed for automated OAR delineation in HNC was shown to be more efficient and consistent compared to manual delineation, which justify its implementation in clinical practice.

A spectroscopic study of the chromatic properties of GafChromicEBT3 films.

PURPOSE: This work provides an interpretation of the chromatic properties of GafChromicEBT3 films based on the chemical nature of the polydiacetylene (PDA) molecules formed upon interaction with ionizing radiation. The EBT3 films become optically less transparent with increasing radiation dose as a result of the radiation-induced polymerization of diacetylene monomers. In contrast to empirical quantification of the chromatic properties, less attention has been given to the underlying molecular mechanism that induces the strong decrease in transparency. METHODS: Unlaminated GafChromicEBT3 films were irradiated with a 6 MV photon beam to dose levels up to 20 Gy. The optical absorption properties of the films were investigated using visible (vis) spectroscopy. The presence of PDA molecules in the active layer of the EBT3 films was investigated using Raman spectroscopy, which probes the vibrational modes of the molecules in the layer. The vibrational modes assigned to PDA's were used in a theoretical vis-absorption model to fit our experimental vis-absorption spectra. From the fit parameters, one can assess the relative contribution of different PDA conformations and the length distribution of PDA's in the film. RESULTS: Vis-spectroscopy shows that the optical density increases with dose in the full region of the visible spectrum. The Raman spectrum is dominated by two vibrational modes, most notably by the ν(C≡C) and the ν(C=C) stretching modes of the PDA backbone. By fitting the vis-absorption model to experimental spectra, it is found that the active layer contains two distinct PDA conformations with different absorption properties and reaction kinetics. Furthermore, the mean PDA conjugation length is found to be 2-3 orders of magnitude smaller than the crystals PDA's are embedded in. CONCLUSIONS: Vis- and Raman spectroscopy provided more insight into the molecular nature of the radiochromic properties of EBT3 films through the identification of the excited states of PDA and the presence of two PDA conformations. The improved knowledge on the molecular composition of EBT3's active layer provides a framework for future fundamental modeling of the dose-response.

Comparison of measured and estimated maximum skin doses during CT fluoroscopy lung biopsies.

PURPOSE: To measure patient-specific maximum skin dose (MSD) associated with CT fluoroscopy (CTF) lung biopsies and to compare measured MSD with the MSD estimated from phantom measurements, as well as with the CTDIvol of patient examinations. METHODS: Data from 50 patients with lung lesions who underwent a CT fluoroscopy-guided biopsy were collected. The CT protocol consisted of a low-kilovoltage (80 kV) protocol used in combination with an algorithm for dose reduction to the radiology staff during the interventional procedure, HandCare (HC). MSD was assessed during each intervention using EBT2 gafchromic films positioned on patient skin. Lesion size, position, total fluoroscopy time, and patient-effective diameter were registered for each patient. Dose rates were also estimated at the surface of a normal-size anthropomorphic thorax phantom using a 10 cm pencil ionization chamber placed at every 30°, for a full rotation, with and without HC. Measured MSD was compared with MSD values estimated from the phantom measurements and with the cumulative CTDIvol of the procedure. RESULTS: The median measured MSD was 141 mGy (range 38-410 mGy) while the median cumulative CTDIvol was 72 mGy (range 24-262 mGy). The ratio between the MSD estimated from phantom measurements and the measured MSD was 0.87 (range 0.12-4.1) on average. In 72% of cases the estimated MSD underestimated the measured MSD, while in 28% of the cases it overestimated it. The same trend was observed for the ratio of cumulative CTDIvol and measured MSD. No trend was observed as a function of patient size. CONCLUSIONS: On average, estimated MSD from dose rate measurements on phantom as well as from CTDIvol of patient examinations underestimates the measured value of MSD. This can be attributed to deviations of the patient's body habitus from the standard phantom size and to patient positioning in the gantry during the procedure.

Calibrating page sized Gafchromic EBT3 films.

PURPOSE: The purpose is the development of a novel calibration method for dosimetry with Gafchromic EBT3 films. The method should be applicable for pretreatment verification of volumetric modulated arc, and intensity modulated radiotherapy. Because the exposed area on film can be large for such treatments, lateral scan errors must be taken into account. The correction for the lateral scan effect is obtained from the calibration data itself. METHODS: In this work, the film measurements were modeled using their relative scan values (Transmittance, T). Inside the transmittance domain a linear combination and a parabolic lateral scan correction described the observed transmittance values. The linear combination model, combined a monomer transmittance state (T(0)) and a polymer transmittance state (T(∞)) of the film. The dose domain was associated with the observed effects in the transmittance domain through a rational calibration function. On the calibration film only simple static fields were applied and page sized films were used for calibration and measurements (treatment verification). Four different calibration setups were considered and compared with respect to dose estimation accuracy. The first (I) used a calibration table from 32 regions of interest (ROIs) spread on 4 calibration films, the second (II) used 16 ROIs spread on 2 calibration films, the third (III), and fourth (IV) used 8 ROIs spread on a single calibration film. The calibration tables of the setups I, II, and IV contained eight dose levels delivered to different positions on the films, while for setup III only four dose levels were applied. Validation was performed by irradiating film strips with known doses at two different time points over the course of a week. Accuracy of the dose response and the lateral effect correction was estimated using the dose difference and the root mean squared error (RMSE), respectively. RESULTS: A calibration based on two films was the optimal balance between cost effectiveness and dosimetric accuracy. The validation resulted in dose errors of 1%-2% for the two different time points, with a maximal absolute dose error around 0.05 Gy. The lateral correction reduced the RMSE values on the sides of the film to the RMSE values at the center of the film. CONCLUSIONS: EBT3 Gafchromic films were calibrated for large field dosimetry with a limited number of page sized films and simple static calibration fields. The transmittance was modeled as a linear combination of two transmittance states, and associated with dose using a rational calibration function. Additionally, the lateral scan effect was resolved in the calibration function itself. This allows the use of page sized films. Only two calibration films were required to estimate both the dose and the lateral response. The calibration films were used over the course of a week, with residual dose errors ≤2% or ≤0.05 Gy.

SU-E-J-138: Fast 2-D Fiducial Marker Detection on Sequential MV Projections in Arc Therapy.

PURPOSE: To automatically detect intrafraction motion during arc radiotherapy for prostate cancer patients by tracking fiducial markers in two-dimensional MV images acquired using the treatment beam, in order to adjust radiation dose accordingly. METHODS: Four fiducial gold markers are implanted in a patient's prostate. Patients are irradiated using a Varian Linac 2100 C/D with RapidArc upgrade (Varian Medical Systems, Palo Alto, CA). MV images (1024 × 768 pixels, 0.392 × 0.392 mm2 pixel size) acquired during a 360 degree gantry rotation at a one second interval (5 degrees) are preprocessed by subtracting a smoothed version of the image to retain only high image frequencies. Edge detection is then applied, followed by a one pixel wide dilation and erosion to transform the edges into contiguous regions. Next, our method searches the centers of visible markers (i.e. not covered by the MLC), constrained by marker estimates from the planning CT. This is done by finding all contiguous regions and maximizing a marker-region distance criterion for every visible marker. A two-dimensional estimate correction over consecutive projections is also implemented to improve marker estimates during gantry rotation. RESULTS: We applied our method on four treatment fractions of the same patient. As such, a total of 191 projections with manually indicated marker ends as ground truth were used as validation. Markers were indicated twice on all images, to include observer errors. Results show a mean detection error of less than 0.5 mm in the projection image (standard deviation 0.6 mm), with an execution time of less than one second per image in matlab. Undetected markers and false positives mostly occurred at moving leaf boundaries, where marker visibility was determined by the observer. CONCLUSIONS: Preliminary findings demonstrate that this method can be used to detect intrafraction motion during arc radiotherapy by only using projected MV images. Research sponsored by Varian Medical Systems, Palo Alto, CA.