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

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

Dynamic string-averaging CQ-methods for the split feasibility problem with percentage violation constraints arising in radiation therapy treatment planning.

We study a feasibility-seeking problem with percentage violation constraints (PVCs). These are additional constraints that are appended to an existing family of constraints, which single out certain subsets of the existing constraints and declare that up to a specified fraction of the number of constraints in each subset is allowed to be violated by up to a specified percentage of the existing bounds. Our motivation to investigate problems with PVCs comes from the field of radiation therapy treatment planning (RTTP) wherein the fully discretized inverse planning problem is formulated as a split feasibility problem and the PVCs give rise to nonconvex constraints. Following the CQ algorithm of Byrne (2002, Inverse Problems, Vol. 18, pp. 441-53), we develop a string-averaging CQ-method that uses only projections onto the individual sets that are half-spaces represented by linear inequalities. The question of extending our theoretical results to the nonconvex sets case is still open. We describe how our results apply to RTTP and provide a numerical example.

Development of an Australia and New Zealand Lung Cancer Clinical Quality Registry: a protocol paper.

INTRODUCTION: Lung cancer is the leading cause of cancer mortality, comprising the largest national cancer disease burden in Australia and New Zealand. Regional reports identify substantial evidence-practice gaps, unwarranted variation from best practice, and variation in processes and outcomes of care between treating centres. The Australia and New Zealand Lung Cancer Registry (ANZLCR) will be developed as a Clinical Quality Registry to monitor the safety, quality and effectiveness of lung cancer care in Australia and New Zealand. METHODS AND ANALYSIS: Patient participants will include all adults >18 years of age with a new diagnosis of non-small-cell lung cancer (NSCLC), SCLC, thymoma or mesothelioma. The ANZLCR will register confirmed diagnoses using opt-out consent. Data will address key patient, disease, management processes and outcomes reported as clinical quality indicators. Electronic data collection facilitated by local data collectors and local, state and federal data linkage will enhance completeness and accuracy. Data will be stored and maintained in a secure web-based data platform overseen by registry management. Central governance with binational representation from consumers, patients and carers, governance, administration, health department, health policy bodies, university research and healthcare workers will provide project oversight. ETHICS AND DISSEMINATION: The ANZLCR has received national ethics approval under the National Mutual Acceptance scheme. Data will be routinely reported to participating sites describing performance against measures of agreed best practice and nationally to stakeholders including federal, state and territory departments of health. Local, regional and (bi)national benchmarks, augmented with online dashboard indicator reporting will enable local targeting of quality improvement efforts.

Using oxygen dose histograms to quantify voxelised ultra-high dose rate (FLASH) effects in multiple radiation modalities.

Purpose.To introduce a methodology to predict tissue sparing effects in pulsed ultra-high dose rate radiation exposures which could be included in a dose-effect prediction system or treatment planning system and to illustrate it by using three published experiments.Methods and materials.The proposed system formalises the variability of oxygen levels as an oxygen dose histogram (ODH), which provides an instantaneous oxygen level at a delivered dose. The histogram concept alleviates the need for a mechanistic approach. At each given oxygen level the oxygen fixation concept is used to calculate the change in DNA-damage induction compared to the fully hypoxic case. Using the ODH concept it is possible to estimate the effect even in the case of multiple pulses, partial oxygen depletion, and spatial oxygen depletion. The system is illustrated by applying it to the seminal results by Town (Nat. 1967) on cell cultures and the pre-clinical experiment on cognitive effects by Montay-Gruelet al(2017Radiother. Oncol.124365-9).Results.The proposed system predicts that a possible FLASH-effect depends on the initial oxygenation level in tissue, the total dose delivered, pulse length and pulse repetition rate. The magnitude of the FLASH-effect is the result of a redundant system, in that it will have the same specific value for a different combination of these dependencies. The cell culture data are well represented, while a correlation between the pre-clinical experiments and the calculated values is highly significant (p < 0.01).Conclusions. A system based only on oxygen related effects is able to quantify most of the effects currently observed in FLASH-radiation.

Incorporating oxygenation levels in analytical DNA-damage models-quantifying the oxygen fixation mechanism.

Purpose.To develop a framework to include oxygenation effects in radiation therapy treatment planning which is valid for all modalities, energy spectra and oxygen levels. The framework is based on predicting the difference in DNA-damage resulting from ionising radiation at variable oxygenation levels.Methods.Oxygen fixation is treated as a statistical process in a simplified model of complex and simple damage. We show that a linear transformation of the microscopic oxygen fixation process allows to extend this to all energies and modalities, resulting in a relatively simple rational polynomial expression. The model is expanded such that it can be applied for polyenergetic beams. The methodology is validated using Microdosimetric Monte Carlo Damage Simulation code (MCDS). This serves as a bootstrap to determine relevant parameters in the analytical expression, as MCDS is shown to be extensively verified with published empirical data. Double-strand break induction as calculated by this methodology is compared to published proton experiments. Finally, an example is worked out where the oxygen enhancement ratio (OER) is calculated at different positions in a clinically relevant spread out Bragg peak (SOBP) dose deposition in water. This dose deposition is obtained using a general Monte Carlo code (FLUKA) to determine dose deposition and locate fluence spectra.Results.For all modalities (electrons, protons), the damage categorised as complex could be parameterised to within 0.3% of the value calculated using microdosimetric Monte Carlo. The proton beam implementation showed some variation in OERs which differed slightly depending on where the assessment was made; before the SOBP, mid-SOBP or at the distal edge. Environment oxygenation was seen to be the more important variable.Conclusions.An analytic expression calculating complex damage depending on modality, energy spectrum, and oxygenation levels was shown to be effective and can be readily incorporated in treatment planning software, to take into account the impact of variable oxygenation, forming a first step to an optimised treatment based on biological factors.

Sparsity constrained split feasibility for dose-volume constraints in inverse planning of intensity-modulated photon or proton therapy.

A split feasibility formulation for the inverse problem of intensity-modulated radiation therapy treatment planning with dose-volume constraints included in the planning algorithm is presented. It involves a new type of sparsity constraint that enables the inclusion of a percentage-violation constraint in the model problem and its handling by continuous (as opposed to integer) methods. We propose an iterative algorithmic framework for solving such a problem by applying the feasibility-seeking CQ-algorithm of Byrne combined with the automatic relaxation method that uses cyclic projections. Detailed implementation instructions are furnished. Functionality of the algorithm was demonstrated through the creation of an intensity-modulated proton therapy plan for a simple 2D C-shaped geometry and also for a realistic base-of-skull chordoma treatment site. Monte Carlo simulations of proton pencil beams of varying energy were conducted to obtain dose distributions for the 2D test case. A research release of the Pinnacle 3 proton treatment planning system was used to extract pencil beam doses for a clinical base-of-skull chordoma case. In both cases the beamlet doses were calculated to satisfy dose-volume constraints according to our new algorithm. Examination of the dose-volume histograms following inverse planning with our algorithm demonstrated that it performed as intended. The application of our proposed algorithm to dose-volume constraint inverse planning was successfully demonstrated. Comparison with optimized dose distributions from the research release of the Pinnacle 3 treatment planning system showed the algorithm could achieve equivalent or superior results.