A Bayesian framework for the detection of physiological pulmonary ventilation changes

Area of Science:

• Medical imaging analysis
• Radiotherapy research
• Pulmonary function testing

Background:

• Radiation pneumonitis is a common side effect of lung cancer radiation therapy, impacting treatment.
• Assessing regional ventilation changes is crucial for understanding treatment response but is hindered by noisy data.
• Current methods struggle with artifacts in ventilation maps, complicating analysis.

Purpose of the Study:

• To develop a robust framework for analyzing longitudinal functional ventilation changes.
• To accurately identify physiological changes from noisy ventilation scan data.
• To quantify significant increases and declines in lung function during radiotherapy.

Main Methods:

• A novel framework was created to estimate physiological changes from longitudinal ventilation scans.
• The algorithm prioritizes monotonic trends and down-weights fluctuating regions.
• The model was validated on synthetic data and applied to 11 lung cancer patients undergoing radiotherapy, using CT-derived ventilation maps.

Main Results:

• The framework identified significant functional decline in 3/11 patients and functional increase in 4/11, linked to tumor regression.
• A control dataset showed only 1.6% significant changes, compared to 32% in the original patient data.
• This demonstrates the framework's ability to distinguish true changes from noise and artifacts.

Conclusions:

• A new framework effectively analyzes functional ventilation changes from longitudinal data.
• Significant functional shifts occur during lung cancer radiotherapy, impacting treatment.
• This framework can potentially guide adaptive radiotherapy strategies by assessing ventilation changes.