Applicability and usage of dose mapping/accumulation in radiotherapy

Abstract

Dose mapping/accumulation (DMA) is a topic in radiotherapy (RT) for years, but has not yet found its widespread way into clinical RT routine. During the ESTRO Physics workshop 2021 on "commissioning and quality assurance of deformable image registration (DIR) for current and future RT applications", we built a working group on DMA from which we present the results of our discussions in this article. Our aim in this manuscript is to shed light on the current situation of DMA in RT and to highlight the issues that hinder consciously integrating it into clinical RT routine. As a first outcome of our discussions, we present a scheme where representative RT use cases are positioned, considering expected anatomical variations and the impact of dose mapping uncertainties on patient safety, which we have named the DMA landscape (DMAL). This tool is useful for future reference when DMA applications get closer to clinical day-to-day use. Secondly, we discussed current challenges, lightly touching on first-order effects (related to the impact of DIR uncertainties in dose mapping), and focusing in detail on second-order effects often dismissed in the current literature (as resampling and interpolation, quality assurance considerations, and radiobiological issues). Finally, we developed recommendations, and guidelines for vendors and users. Our main point include: Strive for context-driven DIR (by considering their impact on clinical decisions/judgements) rather than perfect DIR; be conscious of the limitations of the implemented DIR algorithm; and consider when dose mapping (with properly quantified uncertainties) is a better alternative than no mapping.

Keywords: Anatomical changes; DIR uncertainties; Deformable image registration (DIR); Dose mapping/accumulation; Dose mapping/accumulation landscape (DMAL); Impact of dose mapping uncertainties.

Fig. 1.

Schematic presenting the use of image registration to map dose distributions. a) Registration set-up. Registration is performed between the source and destination images, here presented in black and white drawings, aiming at mapping the source dose distribution to the destination grid. b) Dose mapping using transformation T, showing the underlying assumption that the registration aligning the pair of images is valid to map any spatially correlated image, such as dose distributions. c) Application of the transformation, depending on the used mapping strategy. Note we distinguish between source/destination and fixed/floating images. Source image is the image that is associated with the dose to be mapped. Destination image is where we want to map the dose. Fixed and floating images are the roles these images take in the registration process.

Fig. 2.

An illustrative case demonstrating the effect of registration uncertainties on dose mapping, and its interplay with dose gradients. For each voxel, A and B, two arrows are shown: 1) a red arrow representing an “erroneous” vector resulting after image registration, 2) a green arrow representing the “correct” vector. Even though there is a large distance between the end-points of two arrows for voxel A, its mapped dose differs slightly. On the other hand, the distance between the end-points for voxel B is small (below “accepted thresholds”), but the mapped dose differs considerably.

Fig. 3.

Illustration of the DMAL axes. a) Expected anatomical variations. b) Impact of dose mapping uncertainties on patient safety.

Fig. 4.

DMAL, presenting the current landscape of use cases. The span of each box represents the typical ranges in the anatomical variation and expected impact of dose mapping uncertainties for a given use case.

Fig. 5.

1D scheme highlighting the main problems of using ‘energy/mass transfer’ resulting from naive use of push interpolation. a) The DVF used to push the data representing a constant expansion over 1/3 of the image and a constant compression over the other 2/3, with the 2nd pixel having no mapped pixels. b) The input data. c) The resampled data using the EMT method described in [53]. For each voxel in the dose distributions shown in b) and c), the value corresponds to energy divided by mass. Note the 2nd pixel of the resampled energy and mass distributions corresponds to a ‘hole’ (value of 0), which results in an undefined mapped dose value (hatched pixel). Additionally, the energy and mass distributions contain undulations that would not be expected from the DVF (as it represents constant expansion/compression) and the dose has an undesirable ‘step-like’ appearance (pixel 3 vs 4 having the same value, same for 5 vs 6 and 7 vs 8).