Real-time 4D-MRI-based internal target volume definition for moving lung tumors.

Purpose: In photon radiotherapy, respiratory-induced target motion can be accounted for by internal target volumes (ITV) or mid-ventilation target volumes (midV) defined on the basis of four-dimensional computed tomography (4D-CT). Intrinsic limitations of these approaches can result in target volumes that are not representative for the gross tumor volume (GTV) motion over the course of treatment. To address these limitations, we propose a novel patient-specific ITV definition method based on real-time 4D magnetic resonance imaging (rt-4DMRI). Methods: Three lung cancer patients underwent weekly rt-4DMRI scans. A total of 24 datasets were included in this retrospective study. The GTV was contoured on breath-hold MR images and propagated to all rt-4DMRI images by deformable image registration. Different targets were created for the first (reference) imaging sessions: ITVs encompassing all GTV positions over the complete (ITV80s) or partial acquisition time (ITV10s), ITVs including only voxels with a GTV probability-of-presence (POP) of at least 5% (ITV5%) or 10% (ITV10%), and the mid-ventilation GTV position. Reference planning target volumes (PTVr) were created by adding margins around the ITVs and midV target volumes. The geometrical overlap of the PTVr with ITV5% n from the six to eight subsequent imaging sessions on days n was quantified in terms of the Dice similarity coefficient (DSC), sensitivity [SE: (PTVr \ ITV5% n)/ITV5% n] and precision [PRE: (PTVr \ ITV5% n)/PTVr] as surrogates for target coverage and normal tissue sparing. Results: Patient-specific analysis yielded a high variance of the overlap values of PTV10s r, when different periods within the reference imaging session were sampled. The mid-ventilation-based PTVs were smaller than the ITV-based PTVs. While the SE was high for patients with small breathing pattern variations, changes of the median breathing amplitudes in different imaging sessions led to inferior SE values for the mid-ventilation PTV for one patient. In contrast, PTV5% r and PTV10% r showed higher SE values with a higher robustness against interfractional changes, at the cost of larger target volumes. Conclusions: The results indicate that rt-4DMRI could be valuable for the definition of target volumes based on the GTV POP to achieve a higher robustness against interfractional changes than feasible with today's 4D-CT-based target definition concepts. (C) 2020 The Authors. Medical Physics published by Wiley Periodicals, Inc. on behalf of American Association of Physicists in Medicine.