Team members Fang and Alan along with collaborators from the Department of Human Oncology recently published an article in the Journal of Applied Clinical Medical Physics entitled “MR‐based treatment planning in radiation therapy using a deep learning approach.” This work describes the development of a deep learning technique to synthesize CT images from MRI inputs, and apply them to radiation therapy planning in the head. Using this technique, the absolute percent differences for dosimetric parameters compared to the gold-standard CT-based approach was 0.24% ± 0.46% for the planning target volume (PTV), 1.39% ± 1.31% for the maximum dose, and 0.27% ± 0.79% for the PTV receiving 95% of the prescribed dose (V95). These results suggest that the use of MRI only, without CT, may be a viable approach to radiation treatment planning, providing advantages from the ability of MRI to discern soft tissue details relative to CT.
More details here: https://aapm.onlinelibrary.wiley.com/doi/full/10.1002/acm2.12554