Evaluation of the Fisher-Kolmogorov Glioma Growth Model for Radiotherapy Target Delineation
J Unkelbach1*, F Dittmann1, B Menze2, E Konukoglu1, H Shih1, (1) Massachusetts General Hospital, Boston, MA, (2) Massachusetts General Hospital, Boston, MA, (2) ETH, Zurich, SwitzerlandSU-E-T-486 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: Gliomas infiltrate the adjacent brain parenchyma far beyond the tumor mass visible on current imaging modalities. In current clinical practice, an isotropic margin is applied to account for infiltrative disease. However, histopathology suggests that glioma growth is anisotropic: (1) the ventricles and the falx represent barriers for migrating tumor cells; (2) gray matter is infiltrated less than white matter; and (3) glioma cells appear to preferentially spread along white matter fibers. Developing objective and quantitative approaches to account for these growth patterns may improve and automate target delineation for gliomas.
Methods: We study the Fisher-Kolmogorov glioma growth model, which represents a partial differential equation for the tumor cell density and replicates the observed growth characteristics. Via brain segmentation into anatomical barriers, cerebrospinal fluid, and white and gray matter, the model equations are solved on the patient-specific anatomy. Preferential spread along white matter fibers is incorporated through DTI imaging. The radiotherapy target can be defined as an isoline of the derived tumor cell density. We performed a retrospective study involving 10 glioblastoma patients to (1) identify the anatomical situations for which the model suggests target volumes that differ from the manually drawn targets, and (2) fully understand the capabilities and limitations of the model for target delineation.
Results: It was found that tumors located close to the corpus callosum potentially benefited the most from model-based target delineation because it consistently predicted the contralateral tumor extension across the corpus callosum in addition to easily modeling falx and ventricles as boundaries. Furthermore, for tumors located close to major sulci (Sylvian fissure), modeling reduced gray matter infiltration can potentially reduce the amount of brain tissue targeted for irradiation.
Conclusion: The tumor growth model represents a promising tool to objectively create target volumes for radiotherapy of gliomas by consistently accounting for known growth patterns.