A generalizable dose prediction model for automatic radiotherapy planning based on physics-informed priors and
Tianyu Xiong1, Ge Ren1, Zhi Chen1
1Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong SAR, People's Republic of China.
Medical Physics
|December 31, 2025
Summary
This study developed a universal deep learning model for accurate radiotherapy dose prediction across various cancer sites and techniques. This generalizable model shows promise for improving automatic treatment planning efficiency and quality.
Area of Science:
- Medical Physics
- Artificial Intelligence in Medicine
- Radiotherapy
Background:
- Deep learning for radiotherapy planning offers improved efficiency and consistency.
- Current models are often scenario-specific, limiting clinical adoption.
- A need exists for generalizable dose prediction models.
Purpose of the Study:
- Develop a generalizable dose prediction model for diverse tumor sites, techniques, and doses.
- Achieve high accuracy in dose distribution prediction.
- Demonstrate the feasibility of universal automatic radiotherapy planning.
Main Methods:
- Utilized the GDP-HMM dataset (3234 plans) for training and evaluation.
- Employed a 3D MedNeXt architecture with physics-informed priors (density maps, beam plates).
- Incorporated large-kernel convolutions and UpKern initialization for enhanced accuracy.
Main Results:
- Achieved median prediction errors within 2 percentage points for DVH metrics.
- Demonstrated significant error reduction with physics-informed inputs and larger kernels.
- Generated deliverable plans comparable to reference plans in test cases.
Conclusions:
- Developed a generalizable, accurate dose prediction model for radiotherapy.
- Physics-informed priors and large-kernel convolutions enhance prediction.
- The model enables universal automatic planning, improving efficiency and quality.
Related Concept Videos
01:17Radiation: Applications
1.7K
The average temperature of Earth is the subject of much current discussion. Earth is in radiative contact with both the Sun and dark space; it receives almost all its energy from the radiation of the Sun and reflects some of it into outer space. Dark space is very cold, about 3 K, so Earth radiates energy into it. For instance, heat transfer occurs from soil and grasses, the rate of which can be so rapid that frost can occur on clear summer evenings, even in warm latitudes.
The average...
The average...
1.7K
01:06Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
223
Pharmacokinetic models are mathematical constructs that represent and predict the time course of drug concentrations in the body, providing meaningful pharmacokinetic parameters. These models are categorized into compartment, physiological, and distributed parameter models.
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
223