RPI ID: 2020-059-201

Innovation Summary:
A recurrent neural network (RNN) architecture is developed to predict tumor movement during radiation therapy. The model uses time-series data from respiratory and anatomical signals to forecast tumor position in real time. It supports adaptive radiation delivery and minimizes exposure to surrounding healthy tissue. The system is designed for integration with treatment planning and delivery platforms.

Challenges / Opportunities:
Tumor motion due to respiration complicates precise radiation targeting. This invention provides a predictive solution that adjusts treatment dynamically. It opens opportunities for improved outcomes in lung and abdominal cancers. The method supports real-time feedback and closed-loop control in radiotherapy systems.

Key Benefits / Advantages:
✔ Real-time tumor motion prediction
✔ Reduces radiation exposure to healthy tissue
✔ Enhances treatment accuracy
✔ Integrates with radiotherapy platforms
✔ Supports adaptive therapy

Applications:
• Radiation oncology
• Cancer treatment planning
• Motion-adaptive therapy

Keywords:
#tumorprediction #radiationtherapy #RNN #medicalAI #oncology #motiontracking

Intellectual Property:
US Patent Application, US20210267487A1, 17/188312, filed 01-Mar-2021