Comparison of dynamic tumor tracking error measurement methods for robotic radiosurgery

Area of Science:

• Robotic radiosurgery
• Medical physics
• Oncology

Background:

• Dynamic tumor motion tracking is crucial for robotic radiosurgery of cancers affected by respiratory motion, such as lung and liver tumors.
• Existing methods for measuring tracking error lack comparative analysis, leaving the optimal approach undetermined.

Purpose of the Study:

• To evaluate and compare tracking errors in individual patients using various methods.
• To identify the most effective method for optimizing dynamic tumor tracking in robotic radiosurgery.

Main Methods:

• Comparison of four tracking error evaluation methods: beam's eye view (BEV), machine learning (ML), log (addition error: AE), and log (root sum square: RSS).
• Log (AE) and log (RSS) were derived from system log files.
• Statistical analysis using a t-test with a significance level of 5% to determine differences.

Main Results:

• Mean tracking errors were: BEV (2.87 mm), log (AE) (3.91 mm), log (RSS) (2.91 mm), and ML (3.74 mm).
• Log (AE) and ML methods showed significantly higher errors than BEV (p < 0.001).
• Log (RSS) error was comparable to BEV, indicating its potential as a substitute. Log (RSS) calculation is simpler, potentially improving clinical efficiency.

Conclusions:

• The study identified key differences between tracking error evaluation methods in dynamic tumor tracking radiotherapy.
• Log (RSS) calculated from log files is a viable and simpler alternative to the BEV method for robotic radiosurgery.