Personalized Radiotherapy Design for Glioblastoma: Integrating Mathematical Tumor Models, Multimodal Scans, and...

Area of Science:

• Neuro-oncology
• Medical imaging analysis
• Computational biology

Background:

• Glioblastoma (GBM) is a highly invasive brain tumor with infiltrating cells often missed by standard imaging.
• Current radiotherapy planning for GBM lacks patient-specific personalization, relying on population-based data.
• Infiltrative GBM cells require precise targeting to improve treatment efficacy and reduce toxicity.

Purpose of the Study:

• To develop a Bayesian machine learning framework for personalized radiotherapy planning in glioblastoma.
• To integrate multimodal medical scans (MRI and FET-PET) for accurate tumor cell density inference.
• To improve radiotherapy by sparing healthy tissue and guiding dose escalation for GBM patients.

Main Methods:

• Utilized a Bayesian machine learning framework incorporating mathematical modeling.
• Integrated high-resolution MRI and FET-PET metabolic maps to infer patient-specific GBM cell density.
• Quantified imaging and modeling uncertainties to predict tumor cell density with credible intervals.

Main Results:

• Radiotherapy plans based on inferred tumor infiltration maps spared more healthy tissue, reducing toxicity.
• Treatment accuracy remained comparable to standard radiotherapy protocols.
• Inferred high tumor cell density regions correlated with radioresistant areas, aiding personalized dose escalation.

Conclusions:

• The proposed framework offers a robust, non-invasive tool for personalized radiotherapy design in GBM.
• Integration of multimodal scans and mathematical modeling enhances precision in targeting invasive tumor cells.
• This approach facilitates reduced radiation toxicity and optimized dose delivery for individual GBM patients.