18F-FDG PET/CT Radiomics for Predicting Therapy Response in Primary Mediastinal B-Cell Lymphoma: A Bi-Centric Pilot...

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Primary mediastinal B-cell lymphoma (PMBCL) requires accurate early assessment of therapy response.
• The Deauville score (DS) is a standard measure, but predictive biomarkers are valuable.
• Radiomics analysis of pre-treatment imaging offers potential for predicting treatment outcomes.

Purpose of the Study:

• To investigate the predictive capability of pre-treatment [18F]FDG PET/CT radiomics for therapy response in PMBCL.
• To develop and validate machine learning models for predicting the Deauville score.

Main Methods:

• A bi-centric study included PMBCL patients undergoing [18F]FDG PET/CT.
• Quantitative radiomics features were extracted and harmonized.
• Machine learning models (Random Forest, SVM) were trained and validated using cross-validation and external datasets.

Main Results:

• 27 robust radiomics features were identified for both CT and PET modalities.
• Both CT and PET radiomics models demonstrated effectiveness in predicting the Deauville score.
• External validation showed promising performance metrics, including AUCs of 0.75 for CT and 0.80 for PET.

Conclusions:

• Machine learning models utilizing [18F]FDG PET/CT radiomic features can reliably predict the Deauville score in PMBCL patients.
• Radiomics holds potential as a non-invasive tool for early therapy response assessment in PMBCL.
• These findings support the integration of radiomics into clinical decision-making for PMBCL management.