Dose Calculation for Adaptive Radiation Therapy Replan Decisions
Dose Calculation for Adaptive Radiation Therapy Replan Decisions Context
The NRG Oncology Adaptive Radiation Therapy (ART) review¹ determined that ART is justified if any one of three conditions are met: (1) the maximum dose of any dose limiting organ at risk is exceeded, (2) the prescription isodose volume coverage of any clinical target volume (CTV) is less than 85%, or (3) a significant improvement in CTV coverage can be achieved as the result of a favourable shift in anatomy. Following these guidelines requires the need to include an evaluation of dose in the ART replan decision process.
Therefore, the dose calculation requirement for if and when to make the ART decision to trigger a replan can be focused on achieving the above three criteria. This ART decision dose calculation requirement is a lower burden than the treatment planning system dose calculation requirement. Naturally, a desirable characteristic for any dose calculation algorithm is to best estimate the delivered patient dose to the full 3D (or 4D) anatomy.
With this context, the options, needs, and future directions of dose calculation for ART replan decisions are briefly reviewed below, including the role of superposition dose calculation, SeeTreat’s algorithm of choice, to perform the ART replan decision task.
Dose calculation options
In the 1980s, the radiation therapy community shifted from correcting measurements based on anatomy to modelling patient-specific radiation transport and interactions within each individual patient. Our knowledge of the interactions of megavoltage photons and electrons with human tissue, (we are primarily made of relatively low atomic number elements), is accurate and precise.
A major advance in dose calculation was the superposition method. With remarkable foresight, pioneers developed physics models of beam generation and radiation transport that explicitly account for primary and scattered photon and electron transport, interactions, and energy deposition in media and patients.₂ A particularly important innovation, used by several vendors today, was to discretise the superposition approach into cones while preserving its ability to account for primary and scattered radiation transport, interactions, and energy deposition. This approach was coined collapsed cone convolution by Ahnesjö (1989)²⁻⁵ and remains a widely implemented and validated method.
Other dose calculation algorithms popular today include Monte Carlo particle transport and solvers for the Boltzmann transport equation that model radiation transport on a particle-by-particle basis and transporting particle distributions through finite element solvers, respectively. Each algorithm has strengths and weaknesses. Superposition makes the assumption of an average density between interaction and deposition points, and transporting through water-like rather than material-like media. Solutions to these limitations have been proposed, such as a combined superposition-Monte Carlo approach,⁶ and material-specific kernel generation.⁷ Monte Carlo algorithms are limited by statistical uncertainty, where variance is inversely proportional to the number of simulated particles and therefore computation time. Boltzmann transport algorithms are constrained by the resolution and discretisation of their models. All dose calculation algorithms are limited by the source models used to approximate the complex beams produced during modulated dynamic treatments.
ART replan decision dose calculation needs
A natural question for the radiation oncology community is: which dose calculation algorithms are appropriate for adaptive radiation therapy (ART) replan decisions? The answer requires balancing two requirements, accuracy and speed. On both counts, superposition is a well-justified choice, noting that advances in computation speed for Monte Carlo and Boltzmann transport approaches make these appropriate choices also.
Regarding accuracy, it is important to consider where dose calculation sits within the hierarchy of uncertainties in ART. The dominant sources of uncertainty are contour delineation,⁸ deformable image registration,⁹ and patient motion.¹⁰ The dosimetric uncertainty of a well-commissioned superposition algorithm is consistently below these dominant uncertainties. The same is true for Monte Carlo and Boltzmann transport equation methods. In other words, SeeTreat’s choice of superposition is sufficiently accurate for ART: investing in more computationally demanding algorithms does not meaningfully reduce the overall uncertainty budget in a clinical ART workflow.
The radiation oncology community can therefore have confidence that a well-commissioned superposition algorithm is an appropriate choice for dose calculation in ART. It meets the accuracy requirements set by the clinical context and supports the turnaround times that adaptive workflows demand.
Future directions
It is worth reflecting that the earliest dose measurements were individual patient outcome observations. Personalized volumetric optimal dose distributions are ultimately still what we seek: to understand and deliver what is best for each patient throughout the treatment course, including ART. Today's dose calculation, performed for each individual, remains a surrogate for treatment response, treading the fine balance between treatment success and toxicity by applying population-derived models to individual patients. As the field moves toward more functional imaging and personalised biological information being incorporated into treatment planning, the core function of accurately modelling radiation physics within the patient will remain indispensable. The superposition framework implemented by SeeTreat, computationally efficient and sufficiently accurate, is well-positioned to serve as a cornerstone of dose calculation as adaptive radiation therapy matures and becomes an integral part of every patient’s treatment.
Sources
¹ Glide-Hurst, C. K. et al. Adaptive radiation therapy (ART) strategies and technical considerations: a state of the ART review from NRG oncology. International Journal of Radiation Oncology* Biology* Physics 109, 1054–1075 (2021).
² Ahnesjö, A. Collapsed cone convolution of radiant energy for photon dose calculation in heterogeneous media. MedicalPhysics 16, 577–592 (1989).
³ Mohan, R., Chui, C. & Lidofsky, L. Energy and angular distributions of photons from medical linear accelerators. MedicalPhysics 12, 592–597 (1985).
⁴ Mackie, T. R., Scrimger, J. W. & Battista, J. J. A convolution method of calculating dose for 15‐MV x rays. MedicalPhysics 12, 188–196 (1985).
⁵ Mackie, T. R., Bielajew, A. F., Rogers, D. W. O. & Battista, J. J. Generation of photon energy deposition kernels using the EGS Monte Carlo code. Physics in Medicine & Biology 33, 1–20 (1988).
⁶ Keall, P. J. & Hoban, P. W. Superposition dose calculation incorporating Monte Carlo generated electron track kernels. MedicalPhysics 23, 479–485 (1996).
⁷ Huang, J. Y. et al. Investigation of various energy deposition kernel refinements for the convolution/superposition method. MedicalPhysics 40, 121721 (2013).
⁸ Jameson, M. G., Holloway, L. C., Vial, P. J., Vinod, S. K. & Metcalfe, P. E. A review of methods of analysis in contouring studies for radiation oncology. Journal ofMedicalImaging andRadiationOncology 54, 401–410 (2010).
⁹ Brock, K. K., Mutic, S., McNutt, T. R., Li, H. & Kessler, M. L. Use of image registration and fusion algorithms and techniques in radiotherapy: Report of the AAPM Radiation Therapy Committee Task Group No. 132. MedicalPhysics 44, e43–e76 (2017).
¹⁰ Keall, P. J. et al. The management of respiratory motion in radiation oncology report of AAPM Task Group 76 a. MedicalPhysics 33, 3874–3900 (2006).