Machine Learning-Based Range Verification in Proton Therapy with Pet

Project: National Science and Technology CouncilNational Science and Technology Council Academic Grants

Project Details

Abstract

Proton therapy is a promising modality in clinical radiation oncology due to their well-defined range and favorable depth dose characteristics. However, because of the steep dose gradient at the distal edge of the Bragg peak, uncertainties in the determination of this range have a profound impact on the actually applied dose distribution. Therefore, tools to monitor and control these uncertainties are therefore highly desirable to optimize the proton therapy on an individual level for fully exploiting the superior characteristic of proton beams. During the nuclear interaction between the proton beam and tissues, the elements within the patient were activated and transformed to the positron emitters. Positron emission tomography (PET) has been suggested as a promising tool for verifying the delivery location of the planned dose by imaging the proton-activated β+ isotope. The central goal of the project is to develop a PET-based range verification tool for proton therapy by leveraging the artificial intelligence and deep learning method. Three sets of studies are going to be performed. First, to establish a GATE/GEANT4 Monte Carlo simulation platform towards imaging in proton therapy. Based on this platform, the relationships between proton dose and positron emitters are modeled using the deep learning method. Second, a novel proton dose reconstruction algorithm will be developed and verified with physical phantom experiments. Third, to construct a visualizable verification tool for proton range by integrating the developed deep learning model and dose reconstruction algorithm. In order to mimic a realistic scenario in patients receive proton therapy and PET imaging, an anthropomorphic phantom will be used to validate and evaluate the developed tool. Overall, the new physical experiment data and methodological outputs of the project will provide with a rigorous frame and efficient tools that are like to have a strong impact on particle therapy as a new proton range verification approach.

Project IDs

Project ID:PC10701-1157
External Project ID:MOST106-2314-B182-062-MY2
StatusFinished
Effective start/end date01/08/1831/07/19

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.