Radiomics-based histological grading of pancreatic ductal adenocarcinoma using ¹⁸F-FDG PET/CT: A two-center study.

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Pancreatic ductal adenocarcinoma (PDAC) grading is crucial for treatment planning.
• Accurate preoperative histological grading of PDAC remains challenging.
• ¹⁸F-FDG PET/CT offers functional and morphological imaging insights.

Purpose of the Study:

• To evaluate the efficacy of radiomics features from ¹⁸F-FDG PET/CT in predicting PDAC histological grade.
• To develop and validate a radiomics model for preoperative PDAC grading.
• To assess the potential of radiomics in guiding neoadjuvant therapy and personalized treatment.

Main Methods:

• Retrospective analysis of 111 PDAC patients undergoing ¹⁸F-FDG PET/CT.
• Layer-by-layer tumor segmentation of PET and CT images.
• Feature extraction, selection (least absolute shrinkage and selection), and machine learning models (SVM, RF, LR) for grade prediction.
• 5-fold cross-validation and ROC curve analysis for model performance evaluation.

Main Results:

• PET/CT-based radiomics models achieved high predictive performance, with mean AUCs up to 0.844 (SVM) and 0.840 (LR).
• PET-based models showed AUCs around 0.76-0.77, while CT-based models ranged from 0.576 to 0.770.
• The combined PET/CT radiomics approach demonstrated superior accuracy in distinguishing high-grade from low-grade PDAC.

Conclusions:

• ¹⁸F-FDG PET/CT-derived radiomics models show significant potential for accurate preoperative histological grading of PDAC.
• This non-invasive imaging approach can serve as a valuable tool for neoadjuvant therapy stratification.
• Radiomics facilitates personalized medical decision-making in PDAC management.