Bohong Huang's research focuses on developing deep learning–based solutions for markerless tumor motion monitoring in radiation. Her work centers on a target decomposition technique that uses patient-specific deep learning models to enhance tumor visibility on KV projection images, enabling real-time markerless lung tumor tracking during SBRT. She has validated this framework on a dynamic chest motion phantom, achieving sub-millimeter tracking accuracy with clinically feasible latency, and applied it to quantify internal tumor stability during DIBH lung SBRT, revealing that external surrogates alone may be insufficient for high-precision motion management. In parallel, she is developing diffusion model–based 4D CBCT reconstruction from sparse-view acquisitions to extend markerless tracking capability to additional treatment sites. By integrating AI-driven imaging techniques into radiotherapy workflows, her work aims to improve treatment precision and patient outcomes.
Disclosures:
- Employment: none
- Compensation: none
- Ownership: none
- Leadership: none