Title
Personalized Radiotherapy Planning for Glioma Using Multimodal Bayesian Model Calibration.
Abstract
Existing radiotherapy (RT) plans for brain tumors derive from population studies and scarcely account for patient-specific conditions. We propose a personalized RT design through the combination of patient multimodal medical scans and state-of-the-art computational tumor models. Our method integrates complementary information from high-resolution MRI scans and highly specific FET-PET metabolic maps to infer tumor cell density in glioma patients. The present methodology relies on data from a single time point and thus is applicable to standard clinical settings. The Bayesian framework integrates information across multiple imaging modalities, quantifies imaging and modelling uncertainties, and predicts patient-specific tumor cell density with confidence intervals. The computational cost of Bayesian inference is reduced by advanced sampling algorithms and parallel software. An initial clinical population study shows that the RT plans generated from the inferred tumor cell infiltration maps spare more healthy tissue thereby reducing radiation toxicity while yielding comparable accuracy with standard RT protocols. Moreover, the inferred regions of high tumor cell density coincide with the tumor radio-resistant areas, providing guidance for personalized dose escalation. The proposed integration of data and models provides a robust, non-invasive tool to assist personalized RT planning.
Year
Venue
Field
2018
arXiv: Computational Engineering, Finance, and Science
Population,Time point,Bayesian inference,Computer science,Medical imaging,Glioma,Radiation therapy,Artificial intelligence,Machine learning,Gibbs sampling,Bayesian probability
DocType
Volume
Citations 
Journal
abs/1807.00499
0
PageRank 
References 
Authors
0.34
0
14