Author: Liyuan Chen, Steve Jiang, Chenyang Shen 👨🔬
Affiliation: Department of Radiation Oncology, UT Southwestern Medical Center, UT Southwestern Medical Center 🌍
Purpose: Delays in radiation therapy (RT) initiation caused by conventional CT simulation processes can hinder timely treatment delivery and patient outcomes. This study proposes a Virtual Treatment Simulation (VTS) framework to transform diagnostic CT or MRI scans into synthetic CT (sCT) images. These sCT images replicate in-treatment patient positioning and anatomical configurations with precise Hounsfield Units (HU), enabling accurate dose calculation and RT planning while bypassing the need for separate CT simulation.
Methods: The VTS framework is built on BioSynCT-Net, a biomechanically informed diagnostic-to-synthetic CT transformation network. BioSynCT-Net operates in two steps: (1) A neural network predicts a deformation field, ∅ , modeling anatomical transitions from diagnostic images (curved diagnostic couch) to simulation configurations (flat treatment couch), creating transformed images, xtrans. (2) A second network applies HU conversion/correction to xtran, producing sCT images, xsCT, suitable for clinical planning. The framework training integrates three loss functions: smoothness loss for deformation regularization, reconstruction loss for alignment between sCT and real simulation CT, and contour consistency loss, which ensures structural coherence by leveraging OAR auto-segmentation models.
Results: A proof-of-principle study was performed using a simplified 2D version of BioSynCT-Net on a dataset of 40 gynecological (GYN) patients (20 training, 10 validation, 10 testing). This simplified model focused on body, muscle, and bony contour deformations without HU correction. Results showed that the generated sCT images closely matched patient anatomy on a flat treatment couch, validating the feasibility of the proposed framework.
Conclusion: By streamlining RT planning, BioSynCT-Net eliminates the need for separate CT simulations, significantly reducing treatment delays while maintaining accuracy. Integrating diagnostic imaging into RT workflows, the VTS framework enhances precision, efficiency, and alignment across disciplines, offering a transformative approach to improve patient care in radiation oncology.