4DMRI and motion management in adaptive particle therapy of cancer

In Particle Therapy (PT), precise radiation dose delivery is critical due to the steep dose gradients of charged particles, with organ motion posing a significant challenge to maintain treatment accuracy. To ensure successful treatments, effective motion management is essential during both treatment planning and delivery. Current clinical planning relies on 4D computed tomography (4DCT), which enables the estimation of electron density information and charged particle stopping power ratio (SPR) needed for dose calculations. However, 4DCT suffers from limited soft tissue contrast, susceptibility to motion artifacts, and exposure to ionizing radiation, making it suboptimal for repeated imaging. On the delivery side, respiratory gating, typically guided by a single, 1D external surrogate signal, is clinically used to synchronize beam delivery with a stable breathing phase. Yet, this simple surrogate signal often fails to capture the complexity of internal tumor motion. This PhD project aims to advance and overcome current limitations in motion management of abdominal tumors in PT, by developing novel techniques for both treatment planning and delivery, leveraging Magnetic Resonance Imaging (MRI) and Optical Motion Tracking, respectively. For treatment planning, dynamic MRI is explored as an alternative to 4DCT, due to its superior soft tissue contrast and radiation-free nature, allowing multiple and repeated imaging sessions. However, challenges arise as the lack in electron density information, needed for dose calculations, hinders MRI’s integration in the clinical routine. Over the years, there has been growing interest in synthetic CT (sCT) generation from MRI using deep learning (DL). Regarding the abdominal site challenges arise due to the anatomical complexity of the area, hindering sCT generation. Here, a DL-based approach is developed for the generation of 4D-sCT from abdominal 4DMRI data, capable to facilitate accurate dose calculations and support treatment planning and adaptation. A comparative analysis of two different DL architectures is also performed to further investigate this dynamic synthetic imaging method, demonstrating that both model models have strong potential for 4DsCT synthesis from 4DMRI. Additionally, non-periodic cine-MRI data are exploited to generate abdominal virtual 4DCT data, exploited to evaluate multiple motion mitigation strategies and identify optimal planning-delivery techniques for robust dose distributions. Regarding delivery, this work advances real-time motion management by integrating optical motion monitoring with 4D dose delivery. Unlike traditional 1D gating signals, optical monitoring enables 3D motion tracking of multiple surrogates and improved internal tumor motion – external surrogate motion correlation, which lays the foundation for more advanced motion mitigation strategies like Tumor Tracking. To exploit those features, a dedicated interface was developed to link an optical tracking system with the 4D dose delivery system, allowing for precise synchronization between tumor motion and dynamic beam delivery. Additionally, a framework for real-time motion modeling and treatment verification by means of optical motion monitoring is proposed to support adaptive workflows, with the capability to compensate for both regular and irregular respiratory motion. The combined use of 4DMRI for planning and optical motion tracking for dynamic delivery lays the foundation for a comprehensive motion-compensated workflow in PT.