Modeling brain activity in brain tumor patients and healthy controls: a proof-of-concept study.

Abstract

Event Abstract Back to Event Modeling brain activity in brain tumor patients and healthy controls: a proof-of-concept study. Hannelore Aerts1*, Dirk Van Roost2, Karen Caeyenberghs3, Wim Fias4, Eric Achten5 and Daniele Marinazzo1 1 Ghent University, Department of Data Analysis, Faculty of Psychology and Educational Sciences, Belgium 2 Ghent University Hospital, Department of Neurosurgery, Belgium 3 Australian Catholic University, School of Psychology, Faculty of Health Sciences, Australia 4 Ghent University, Department of Experimental Psyschology, Faculty of Psychology and Educational Sciences, Belgium 5 Ghent University Hospital, Department of Radiology and Nuclear Medicine, Belgium Recent advances in noninvasive imaging technology have allowed the creation of comprehensive whole-brain maps of the structural and functional connections of the human cerebrum [1]. In turn, this has enabled the construction of computational models of resting-state brain activity, allowing the investigation of the intricate relationship between structural and functional connectivity [2,3]. In addition, whole-brain computational models may be used as unique predictive tools for investigating the impact of structural connectivity damage on brain dynamics. In these computational models, nodes and/or edges of the structural connectivity matrix can be deleted – corresponding to the presence of necrosis or the selective removal of brain tissue – and effects on functional connectivity can be evaluated (e.g., [4]). In this study, we simulate brain activity in a brain tumor patient and a healthy control subject. In particular, functional connectivity is modeled based on the individuals’ (pre-operative) structural connectome obtained by diffusion MRI combined with tractography, using a neuroinformatics platform with a suitable dynamical model. Afterwards, predicted and empirical resting-state fMRI connectivity are compared to assess the performance of the computational model. In subsequent studies, this will be extended to the prediction of effects of invasive interventions on the functional network, by virtually lesioning brain tumor patients’ pre-operative structural connectome in order to mimic brain tumor resection. Six months after patients’ surgery, structural and functional brain connectivity will be re-evaluated, and correspondence between theoretical predictions of the effects of a virtual lesion on the network dynamics, and the actual outcome of the surgery will be evaluated. If a correspondence is found, this will represent an extremely valuable benchmark for all pre-surgical evaluations. Conversely, we will take into account the post-surgery network information and try to improve the neuroinformatics platform for future use.