Session Item

MR-guided online adaptive radiotherapy (MRgOART) has become more frequent in radiotherapy clinics with the inclusion of MR-linacs (MRL). In the workflows for adaptive fractions, an MR scan is first acquired using a dedicated MRL sequence, the structure set re-contoured, and the plan re-optimized and re-calculated. For plan re-calculation, an electron densities (ED) map is assigned from either the deformed or rigid CT simulated before treatment. The ED map extracted from the deformed CT (dCT) is not necessarily consistent with the MR image and thus subject to QA prior the treatment delivery. To assess the dosimetric impact of incorrect ED alignment with the MRI, bulk EDs were manually assigned to the MR images and the plan was re-calculated using those 2 conditions. The bulk ED assignment is typically called MR-only

The 0.35T MRL from ViewRay was used for calculating the plans and acquiring MR images. In this system, a Monte Carlo based treatment planning system uses the ED relative to water (RED) for material assignment and calculating dose-to-medium. 2 bulk ED were manually assigned: “water-like” for all the structures inside the patient (RED=1), and “lung-like” for the lungs (RED=0.12). 15 patients were included with lesions mainly located in the upper-abdominal/thoracic regions such as pancreas and liver. 1 online-adaptive fraction per patient was selected as a representation of the full treatment. The 15 patients exhibit 32 PTV targets and 216 organs at risk (OARs) in total. Several points of the dose volume histogram (DVH) were compared; the 98%, 50% and 2% (near, median and max coverage) for the targets, and the 2% and mean dose (serial and parallel type indicators) for the OARs. The discrepancies were normalized relative to the prescribed dose

Figures 1 and 2 show the relative discrepancies between the doses for the MR-only and CT approach. The dash red lines and bands indicate mean and ±1SD of the data. For all target DVH points in figure 1, all discrepancies are <2% with a systematic shift in mean value of -0.16% For the 2 OARs DVH points, the 2 the highest discrepancy is <1.4% and mean value <-0.08%. In general, the deviations for the OARs show less spread as compared to the targets which suggests less dosimetric uncertainty (more robustness for ED assignment) of the OARs than for the targets

This study evaluates the possibility of using MR-only for MRgOART in the upper/abdominal-thoracic region. With MR-only using assigned ED of water and lung, the discrepancies with the CT is less than 2% and 1.4% for the targets and OARs, respectively. Therefore, variations of less than 2% are expected when the CT-based ED map does not follow the MR image in adaptive radiotherapy workflows for the upper abdominal/thoracic region. We believe that the way forward is to assign an RED for all the structures based on e.g. the ICRU 46 report or synthetic CT generation directly from the MRI. This will allow for a more thorough comparison between ED methods