Inverse Problem Regularization for 3D Multi-Species Tumor Growth Models.

Area of Science:

• Computational Biology
• Medical Imaging
• Oncology

Background:

• Glioblastoma multiforme (GBM) is an aggressive brain cancer characterized by rapid growth and infiltration.
• Accurate modeling of GBM tumor dynamics is crucial for effective treatment planning.
• Current methods often struggle to quantify infiltrative tumor cell populations from clinical imaging.

Purpose of the Study:

• To present a multi-species partial differential equation (PDE) model for glioblastoma growth.
• To develop an algorithm for calibrating this model using magnetic resonance imaging (MRI) data.
• To enable stable estimation and quantification of non-observable tumor species, particularly infiltrative cells.

Main Methods:

• A multi-species PDE model was formulated, representing proliferative, infiltrative, and necrotic tumor cells.
• Model calibration was treated as an inverse problem, solved via PDE-constrained optimization using multi-parametric MRI data.
• A two-stage inversion process, incorporating compressed sensing and weighted regularization, addressed the ill-posed nature of the problem.

Main Results:

• The proposed scheme enabled stable estimation of non-observable tumor species and quantification of infiltrative cells.
• Tumor segmentation accuracy, measured by the Dice score, improved by 5%-10% compared to single-species models.
• Model parameter reconstruction errors were reduced by 4%-80% with regularization, demonstrating improved model calibration.

Conclusions:

• The developed multi-species PDE model and calibration algorithm provide a robust framework for analyzing GBM.
• The method successfully estimates infiltrative tumor cell populations, offering valuable clinical insights.
• This approach enhances the quantitative analysis of tumor dynamics from standard clinical MRI scans.