SU-D-BRA-02: An Extended Time-Variant Seasonal Autoregressive Model-Based Prediction for Irregular Breathing Motion...

Area of Science:

• Medical Physics
• Radiotherapy Technology
• Computational Biology

Background:

• Dynamic tumor tracking in radiotherapy faces challenges due to observation-irradiation delays.
• Existing Time-Variant Seasonal Autoregressive (TVSAR) models predict regular breathing motion with high accuracy.
• Limitations exist in predicting non-regular breathing patterns, impacting treatment precision.

Purpose of the Study:

• To develop an extended prediction method enhancing the TVSAR model for diverse respiratory motion patterns.
• To improve the accuracy of tumor motion prediction by incorporating residual components.
• To compensate for delays in dynamic irradiation systems for moving tumors.

Main Methods:

• An autoregressive (AR) model was employed to predict the residual component not captured by the TVSAR model.
• Adaptive determination of AR model order and parameters for each residual component using an information criterion.
• Evaluation using eleven 3D lung tumor motion datasets from the Cyberknife Synchrony system.

Main Results:

• The proposed method demonstrated superior performance compared to conventional and state-of-the-art methods for 0-1 second ahead prediction.
• Achieved an average prediction error of 0.920 ± 0.348 mm for 0.5-second forward prediction.
• Successfully predicted various breathing patterns, including irregular ones.

Conclusions:

• A novel prediction method integrating TVSAR with adaptive residual prediction has been developed.
• This method enhances dynamic radiotherapy by compensating for system delays across various breathing patterns.
• The improved prediction accuracy supports more effective treatment of moving tumors.