A review on 3D deformable image registration and its application in dose warping
Haonan Xiao, Ge Ren, Jing Cai
Abstract
Deformable image registration (DIR) has been well explored in recent decades, and it is widely utilized in clinical tasks, especially dose warping. Nowadays, as deep learning (DL) develops rapidly, many DL-based methods were also applied in DIR. This paper reviews DL-based DIR methods in recent years and the application of DIR in dose warping. We collected and categorized the latest DL-based DIR studies. A thorough review of each category was presented, in which studies were discussed based on their supervision, advantage, and challenges. Then, we reviewed DIR-based dose warping and discussed its rationale, feasibility, successes, and difficulties. Lastly, we summarized the review on both parts and discussed their future development trend.