< Back to previous page

Project

Accurate patient-individual prediction of proton therapy benefit in thoracic cancers using 3D dose maps and radiomics

 State-of-the-art curative radiotherapy for thoracic cancers comes with a considerable risk of radiation-induced toxicity, mostly related to incidental radiation dose to the lungs and the heart. Proton therapy (PT) is an innovative treatment modality that exploits the physical properties of protons to more accurately deposit dose. Promising PT results have been published, showing low extraneous dose and limited toxicity. The high cost of PT requires a cost-effective implementation. Modelbased patient selection was therefore described based on normal tissue complication probability (NTCP) models, predicting toxicity risk for the individual patient. However, NTCP models currently have limited discrimination. Our main objective is to build reliable NTCP models based on large datasets from UZ Leuven and international collaborations. We propose 3 major improvements to the NTCP modeling paradigm. Firstly, 3D dose maps will consider regional organ dose information. A second step will introduce Monte Carlo simulations providing a better low dose assessment. Finally, radiomics analyses will explore image biomarker signatures of radiosensitivity. We hypothesize that an in-silico planning study will now result in guidelines for a more accurate patient selection for PT. The Belgian PT center ParTICLe in UZ Leuven will treat its first patient in 2020. This will provide crucial feedback to test our hypothesis and refine our models, e.g. by a European clinical trial in esophageal cancer.
 

Date:1 Oct 2020 →  30 Sep 2023
Keywords:prediction models of radiation-induced toxicity, proton therapy, radiomics image features
Disciplines:Medical imaging and therapy not elsewhere classified, Cancer therapy, Applied and interdisciplinary physics, Radiation therapy