Fast and precise dose estimation for very high energy electron radiotherapy with graph neural networks

Introduction External beam radiotherapy (RT) is one of the most common treatments against cancer, with photon-based RT and particle therapy being commonly employed modalities. Very high energy electrons (VHEE) have emerged as promising candidates for novel treatments, particularly in exploiting the FLASH effect, offering potential advantages over traditional modalities. Methods This paper introduces a Deep Learning model based on graph convolutional networks to determine dose distributions of therapeutic VHEE beams in patient tissues. The model emulates Monte Carlo (MC) simulated doses within a cylindrical volume around the beam, enabling high spatial resolution dose calculation along the beamline while managing memory constraints. Results Trained on diverse beam orientations and energies, the model exhibits strong generalization to unseen configurations, achieving high accuracy metrics, including a <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="m1"><mml:mrow><mml:mi>δ</mml:mi></mml:mrow></mml:math> -index 3% passing rate of 99.8% and average relative error <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="m2"><mml:mrow><mml:mo>&lt;</mml:mo></mml:mrow></mml:math> 1% in integrated dose profiles compared to MC simulations. Discussion Notably, the model offers three to six orders of magnitude increased speed over full MC simulations and fast MC codes, generating dose distributions in milliseconds on a single GPU. This speed could enable direct integration into treatment planning optimization algorithms and leverage the model’s differentiability for exact gradient computation.