Patient-specific predictive modeling that integrates advanced cancer imaging

Project: Research project

Project Details


Abstract We are integrating the mathematical modeling of tumor proliferation and invasion with advanced cancer imaging methods. We are applying this approach to gliomas, which are aggressive and highly invasive primary brain tumors associated with dismal prognoses. Because of the relative inaccessibility of tissue, the clinical management of gliomas are strongly directed by imaging, thus tools integrating changes on imaging with a dynamic understanding of the cancer system are sorely needed. The goals of our project are twofold: To impact current clinical challenges with treatment of gliomas, and to provide tools for the development of new therapies for these challenging cancers. Our first goal is to develop image-based response metrics based on the growth kinetics of each patient's tumor, as seen on both anatomical imaging (MR) and functional imaging (PET and advanced MR). We will use mathematical modeling to develop a patient-specific Untreated Virtual Imaging Control (UVIC) that quantifies the dynamics of each patient's tumor system. We will then test the UVIC model against a novel set of paired PET and MR images at multiple time-points (five on average) for each of 20 glioblastoma patients. The paired images will be acquired throughout the course of therapy and compared with the UVIC predicted images of hypoxia (FMISO-PET), necrosis (T1-Gd MR) and cellularity (DWI MR). The second, and overall, goal of this project is to extend the UVIC model to the early response assessment of individual patients in clinical trials. This will provide a tool for the development of much-needed therapies that are more effective for gliomas. The methodologies developed in the project could be extended by refining the biological modeling, and could also be applied to other cancers by the use of appropriate growth kinetic models.
StatusNot started


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.