Related Experiment Videos
Using an artificial neural network to define the planning target volume in radiotherapy
N Kaspari1, B Michaelis, G Gademann
1Otto von Guericke University Magdeburg, Clinic for Radiotherapy, Germany.
Journal of Medical Systems
|April 29, 1998
Summary
This study introduces a novel neural network to predict radiotherapy planning target volumes from tumor shape. The AI model generalizes expert knowledge for accurate predictions, showing promise for brain tumor treatment.
Area of Science:
- Medical Physics
- Artificial Intelligence
- Radiotherapy
Background:
- Accurate delineation of the planning target volume (PTV) is crucial for effective radiotherapy.
- Manual PTV contouring requires significant expertise and time.
- Automating PTV prediction can improve consistency and efficiency in radiation oncology.
Purpose of the Study:
- To design and test a neural network for predicting the PTV from detected tumor shape.
- To evaluate the network's ability to generalize expert medical knowledge.
- To assess the feasibility of AI-driven PTV prediction in radiotherapy.
Main Methods:
- Development of a novel neural network architecture.
- Input: Three-dimensional image of the detected tumor.
- Output: Predicted planning target volume.
Main Results:
- The neural network successfully predicts PTV from tumor shape.
- The model demonstrates generalization of expert medical knowledge.
- Initial results show promising performance for simple-shaped brain tumors.
Conclusions:
- A neural network can effectively predict PTV in radiotherapy.
- AI holds potential to assist radiation oncologists in treatment planning.
- Further research is warranted for complex tumor shapes and diverse patient populations.