Events Calendar

Gregory Buti (MGH): Practical Solutions for fast and clinical robust optimization in proton therapy
Tuesday 02 November 2021, 12:00pm - 01:00pm

Robust optimization (RO) has proven to be a useful tool in order to generate treatment plans that do not deteriorate at the moment of delivery. Especially for proton therapy treatments of moving targets embedded in heterogeneous media, such as lung tumor cases, RO can provide major advantages over conventional safety margins. Unfortunately, RO is a computationally expensive process. This may limit its applicability in online adaptive workflows. Moreover, its inherently conservative nature can lead to suboptimal trade-offs between target coverage and normal tissue sparing. In this presentation, we present the work developed so far during my PhD which aim to tackle these issues. First, we focus on “worst-case” optimization, a RO algorithm commonly used in clinical practice. In worst-case optimization, the uncertainty scenarios are pre-defined by the treatment planner. The optimizer will subsequently evaluate the planning objectives in each scenario and use the worst-case scenario to guide the optimization solution. We present a pre-selection method for the scenarios that can yield both a reduction of optimization time as well as prevent overly conservative treatment plans. Moreover, we present an approach that uses a limited set of scenarios, evaluated on-the-fly, to accelerate the optimization process. Afterwards, we extend these ideas to biological uncertainties, in our case, target volume uncertainty. Today, the CTV is used in RO to evaluate target coverage. We show how using the CTV together with geometric uncertainties (e.g., setup errors) lead to including scenarios which are statistically too improbable. Based on this, we present a probabilistic RO framework that uses the so-called “clinical target distribution”, a continuous alternative to the CTV.  A fully probabilistic approach aims to explicitly consider target volume uncertainty, in a statistically consistent way. 

Short bio: Gregory Buti graduated from the KU Leuven (Belgium) in 2018 with a master's degree in Medical Radiation Physics. He is since pursuing a PhD at the University UCLouvain (Belgium) in the research lab MIRO (Medical Imaging, Radiation & Oncology). His main topic is the clinical application of robust optimization, mainly applied to proton therapy treatments.

Location : Virtual