Optimization of fluoroscopy parameters using pattern matching prediction in the real-time tumor-tracking radiotherapy...

Area of Science:

• Medical Physics
• Radiotherapy Technology
• Image Processing

Background:

• Real-time tumor tracking radiotherapy relies on fluoroscopy for internal fiducial marker positioning.
• Current manual selection of fluoroscopy parameters risks unnecessary radiation dose or marker signal loss.
• Pattern Recognition Score (PRS) quality is crucial for accurate marker tracking.

Purpose of the Study:

• To develop an optimization method for minimizing fluoroscopic dose in real-time tumor tracking.
• To predict the PRS to ensure usable image quality for marker tracking.
• To enable accurate assessment of marker tracking accuracy.

Main Methods:

• Proposed a novel optimization method to predict PRS in potential marker movement regions.
• Validated the PRS prediction method through experiments under various conditions.
• Conducted experiments using a motion controller and anthropomorphic chest phantom to simulate clinical scenarios.

Main Results:

• The proposed method demonstrated good agreement between predicted and measured PRS.
• The root mean square error between predicted and measured PRS was within 1.44.
• The prediction method showed feasibility for clinical application in simulated scenarios.

Conclusions:

• The novel PRS prediction method effectively minimizes fluoroscopic dose while maintaining image quality for marker tracking.
• The method is expected to be applicable in real clinical settings for improved radiotherapy.
• Accurate PRS prediction enhances the reliability of real-time tumor tracking systems.