Can radiomics signatures and machine learning methods reinforce the revived role of ۱۸F-NaF in metastatic bone disease?

Purpose: To evaluate whether radiomic features extracted from ۱۸F-NaF PET/CT scans, analyzed using machine learning (ML) methods, can improve the differentiation between true metastatic bone lesions (TP) and false-positive benign uptake (FP), thereby enhancing the diagnostic utility of ۱۸F-NaF PET/CT. Methods: This retrospective study included ۶۲ patients with known primary malignancies who underwent ۱۸F-NaF PET/CT. Lesions were classified as TP or FP based on consensus interpretation including follow-up. Patients were randomly split into training (n=۴۱) and validation (n=۲۱) groups. Radiomic features were extracted from PET images using LIFEx software. Feature selection (ANOVA, RFE) and ML model training (SVM, Random Forest, XGBoost) were performed. Model performance was evaluated using accuracy, specificity, sensitivity, and AUC, initially with a train/validation split and subsequently with ۵-fold cross-validation incorporating feature engineering and hyperparameter tuning. Feature importance was assessed using SHAP. Results: Significant differences in SUVmax (p=۰.۰۰۶) and SUVmean (p=۰.۰۳۴) were observed between TP and FP lesions. Initial validation showed XGBoost performed best (AUC=۰.۷۸). After optimization and ۵-fold cross-validation on the combined dataset (n=۶۲), the tuned XGBoost model achieved the highest performance (Mean Accuracy: ۸۵.۷% ±۲.۹%, Mean AUC: ۰.۸۶), outperforming Random Forest (AUC: ۰.۷۹) and SVM (AUC :۰.۷۴). SHAP analysis identified SUV max, SUV mean, Voxel Volume Num, GLRLM RLNU, and Skew. Conclusion: Radiomics-based machine learning classifiers, particularly XGBoost, demonstrated strong performance in distinguishing true metastatic from false-positive benign lesions on ۱۸F-NaF PET/CT. Integrating radiomics and ML can potentially improve the diagnostic accuracy and robustness of ۱۸F-NaF PET/CT for assessing bone metastases. Further validation in larger cohorts is warranted.

کلیدواژه ها:

NaF PET ، Radiomics ، Machine Learning ، Bone metastases