Radiation pneumonitis prediction model with integrating multiple dose-function features on 4DCT ventilation images

Area of Science:

• Radiation Oncology
• Medical Physics
• Computational Biology

Background:

• Radiation pneumonitis (RP) is a significant dose-limiting toxicity in non-small-cell lung cancer (NSCLC) radiotherapy.
• Accurate prediction of RP is crucial for optimizing treatment plans and minimizing patient toxicity.

Purpose of the Study:

• To develop a predictive model for radiation pneumonitis (RP) in non-small-cell lung cancer (NSCLC) patients.
• To integrate dose-function features derived from four-dimensional computed tomography (4DCT) ventilation imaging using the least absolute shrinkage and selection operator (LASSO) method.

Main Methods:

• A cohort of 126 NSCLC patients undergoing 4DCT scans between 2013 and 2020 was analyzed.
• Dose-function features were computed from ventilation images, considering different functional lung zones and dose thresholds (e.g., >20 Gy).
• A LASSO regression model was employed for feature selection and prediction, validated through fivefold cross-validation.

Main Results:

• 39.3% of patients experienced grade ≥2 RP.
• The developed model incorporating dose-function features achieved a mean Area Under the Curve (AUC) of 0.814.
• Relative regression coefficients indicated significant impact of dose on specific functioning lung zones, with the top 20% most functioning zone showing the highest impact.

Conclusions:

• Integrating dose-function features from 4DCT ventilation imaging into a machine learning framework enhances RP prediction accuracy.
• The LASSO-derived relative regression coefficients quantify the impact of dose on functioning lung zones, supporting functional image-guided radiotherapy.
• This approach holds potential for improving treatment planning in NSCLC patients to reduce radiation pneumonitis.