Mean position tracking of respiratory motion

Area of Science:

• Medical Physics
• Radiotherapy
• Biomedical Engineering

Background:

• Respiration-induced tumor motion poses challenges for accurate radiation delivery.
• Current methods like gating or adaptive alignment may benefit from tracking general motion trends.
• Traditional moving average filters require reliable phase estimation, which is difficult with noisy, semi-periodic respiratory data.

Purpose of the Study:

• To develop a robust method for tracking the mean position of respiratory tumor motion.
• To avoid explicit estimation of instantaneous breathing phase.
• To provide real-time tracking algorithms suitable for adaptive radiotherapy.

Main Methods:

• Forming a state vector from current and past respiration signal values.
• Fitting an ellipse model to training data to capture motion hysteresis (eccentricity and orientation).
• Implementing two recursive online algorithms: adaptive window size and temporal discounting.

Main Results:

• The proposed ellipse model and recursive algorithms effectively track the mean position of respiratory motion.
• The method demonstrates robustness even with noisy breathing signals.
• Performance is validated against simulated data and real-time displacement (RPM) signals, showing comparable or improved results over moving averages.

Conclusions:

• The developed method offers a reliable, phase-independent approach for real-time mean tumor motion tracking.
• This technique can enhance the precision and adaptability of radiotherapy by providing accurate motion trend information.
• The recursive algorithms are suitable for online implementation in clinical settings.