An optimized generic cerebral tumor growth modeling framework by coupling biomechanical and diffusive models with treatment effects

Mathematical modeling of cerebral tumor growth is of great importance in clinics. It can help in understanding the physiology of tumor growth, future prognosis of tumor shape and volume, quantify tumor aggressiveness, and the response to therapy. This can be achieved at macroscopic level using medical imaging techniques (particularly, magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI)) and complex mathematical models which are either diffusive or biomechanical. We propose an optimized generic modeling framework that couples tumor diffusivity and infiltration with the induced mass effect. Tumor cell diffusivity and infiltration are captured using a modified reaction-diffusion model with logistic proliferation term. On the other hand, tumor mass effect is modeled using continuum mechanics formulation. In addition, we consider the treatment effects of both radiotherapy and chemotherapy. The efficacy of chemotherapy is included via an adaptively modified log-kill method to consider tissue heterogeneity while the efficacy of radiotherapy is considered using the linear quadratic model. Moreover, our model efficiently utilizes the diffusion tensor of the diffusion tensor imaging. Furthermore, we optimize the tumor growth parameters to be patient-specific using bio-inspired particle swarm optimization (PSO) algorithm. Our model is tested on an atlas and real MRI scans of 8 low grade gliomas subjects. Experimental results show that our model efficiently incorporates both treatment effects in the growth modelingprocess. In addition, simulated growths of our model have high accuracy in terms of Dice coefficient (average 87.1%) and Jaccard index (77.14%) when compared with the follow up scans (ground truth) and other models as well.