Dosimetric impact of real-time re-optimization of proton pencil-beam scanning for moving targets

TL;DR

The paper addresses the interplay between respiratory motion and proton PBS delivery, which can degrade target coverage. It introduces a real-time adaptive framework based on receding-horizon control (IPAO) that re-optimizes upcoming PBS spots after each energy layer using observed and predicted motion, with horizon length $h_{\hat{k}}$. Across three lung patients and 500 breathing-pattern realizations, IPAO markedly improves near-worst-case CTV coverage, increasing CTV D98 from about 96.4% to 98.9% of the prescription and reducing dose-variability (interquartile range) by more than 80% relative to non-adaptive IPO/IPRO, while generally preserving target homogeneity and OAR limits. The findings suggest motion-adaptive PBS delivery can substantially mitigate the interplay effect, contingent on accurate and fast motion modeling and feasible real-time computation.

Abstract

When treating moving tumors, the precise delivery of proton therapy by pencil beam scanning (PBS) is challenged by the interplay effect. Although several 4D-optimization methods have been proposed, what is the most beneficial motion management technique is still an open question. In this study, we wish to investigate the dosimetric impact of re-optimizing the PBS spot weights during the treatment delivery in response to, and in anticipation of, variations in the patient's breathing pattern. We simulate for PBS the implementation of a real-time adaptive framework based on principles from receding horizon control. We consider the patient motion as characterized by a one-dimensional amplitude signal and a 4DCT, to simulate breathing of variable frequency. The framework tracks the signal and predicts the future motion with uncertainty increasing with the length of the prediction horizon. After each delivered energy layer, the framework re-optimizes the spot weights of the next layer based on the delivered dose and the predicted motion. For three lung patients, we generate 500 variable breathing patterns to evaluate the dosimetric results of the framework and compare them to those of implementations of previously proposed non-adaptive methods. Compared to the best non-adaptive method, the adaptive framework improves the CTV D98 in the near-worst breathing scenario (5th percentile), from 96.4 to 98.9 % of the prescribed dose and considerably reduces the variation as measured by a mean decrease in the inter-quartile range by more than 80 %. The target coverage improvements are achieved without generally compromising target dose homogeneity or OAR dose. The study indicates that a motion-adaptive approach based on re-optimization of spot weights during delivery has the potential to substantially improve the dosimetric performance of PBS given fast and accurate models of patient motion.

Figures (6)

• Figure 1: A representation of three consecutive breathing cycles sampled from the motion model and the resulting spot-to-phase assignment. The horizontal lines illustrate the bins used to convert the continuous signal $a(t)$ (blue) into the discrete state $s(t)$ (orange). The horizontal placement of each vertical bar (gray) represents the delivery time of a single spot, and nearby bars constitute an energy layer. The height and the marker (black) at the top of each bar coincide with the state during which the spot is delivered.
• Figure 2: Examples of the motion prediction model in the adaptive approach. The solid lines represent the actual motion signal $a(t)$ (blue) and the actual discrete state $s(t)$ (orange). The dashed lines represent the corresponding predictions. The horizontal placements of the vertical bars indicate the start times of the spots in the next energy layer, and the markers at the end of each line coincide with the state at the spot's delivery time.
• Figure 3: CTV DVH bands for each of the evaluated methods. At each dose value $\hat{d}$, a DVH band represents a 95% confidence interval for the value of $V_{\hat{d}}$. For each method, the dashed line represents the DVH in the nominal motion scenario.
• Figure 4: Box plots showcasing the spread of the CTV D98 and the CTV D2. The box edges indicate quartiles, while the whiskers indicate 5th and 95th percentiles. The mean and median are indicated by the plus and the solid line, respectively.
• Figure 5: Box plots showcasing the spread of the OAR D2s for each patient. The box edges indicate quartiles, while the whiskers indicate 5th and 95th percentiles. The mean and median are indicated by the plus and the solid line, respectively.