Novel Harmonization Method for Multi-Centric Radiomic Studies in Non-Small Cell Lung Cancer

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Stereotactic body radiation therapy (SBRT) is a standard treatment for early-stage non-small cell lung cancer (NSCLC).
• Predicting treatment response and progression-free survival (PFS) is crucial for optimizing SBRT outcomes.

Purpose of the Study:

• To investigate the association between radiomic features from pre-treatment CT and PET scans and clinical outcomes in early-stage NSCLC patients receiving SBRT.
• To develop and validate predictive models for 24-month PFS using radiomic data.

Main Methods:

• Retrospective analysis of 117 early-stage NSCLC patients from seven Italian centers treated with SBRT.
• Extraction of 3004 radiomic features from pre-treatment CT and PET images.
• Development of a novel CT feature harmonization technique using LASSO for feature selection.

Main Results:

• Harmonized CT radiomic models (B models) outperformed models using original CT features (C models).
• A linear support vector machine (SVM) model (A1) incorporating harmonized CT and PET features achieved an AUC of 0.77 (0.63-0.85) for predicting 24-month PFS in an external validation cohort.
• Clinical features did not significantly improve the predictive performance of the radiomic models.

Conclusions:

• A novel CT data harmonization strategy, termed delta radiomics, shows promise for improving radiomic model performance.
• Harmonized radiomic models, particularly those combining CT and PET features, can effectively predict patient prognosis in early-stage NSCLC treated with SBRT.