Project Details
Projekt
Development and validation of artificial intelligence models using CT data and spatial dose distributions for the prediction of radiation pneumonitis in lung cancer patients treated within the multicenter prospective REQUITE trial
Subject Area
Radiology
Term
since 2023
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 515279324
In this project we aim to develop an artificial intelligence (AI)-based system for individualized toxicity prediction in lung cancer patients using CT imaging and spatial radiotherapy (RT) dose maps. All AI models will be externally validated using prospective data from the multicentric REQUITE trial. We have defined five work packages addressing key research questions necessary to achieve this goal: (i) Initially, we will integrate, preprocess, and segment CT and RT dose data from the prospective REQUITE multicentric trial and an existing in-house cohort providing comprehensive and high-quality toxicity and outcome data. (ii) We will develop AI prediction models using hand-crafted radiomic features and deep learning techniques for the prediction of radiation pneumonitis (RP) using CT and RT dose data. The incremental value of multitask outcome prediction of overall survival. By being able to better predict the occurrence of RP, radiotherapy plans could be better individualized. Preliminary work has shown that certain lung regions may be more susceptible to the development of radiation pneumonitis. (iii) We will identify common intra-pulmonal spatial high-risk regions based on voxel-based dose analysis and deployment of neural network-based techniques. (iv) Based on CT data will predict individual high-risk region using spatially resolved radiomic features maps and interpretable deep learning approaches. (v) Finally, we will integrate the best AI models into a clinical decision support system that will enable individualized RP risk-stratification, outcome prediction and identify individual high-risk regions. All developed AI models will be made available to the scientific community online.
DFG Programme
Priority Programmes
