Estimation of Patient Brain Connectome for Individualized the Virtual Brain Model to Predict Neurosurgical Treatment Outcomes in Parkinson’s Disease

Abstract

Background and aim: Parkinson’s disease (PD) is a neurodegenerative disorder characterized by involuntary, uncontrolled movements. Currently one of the best available treatment options is subthalamic nucleus (STN) deep brain stimulation (DBS). However, this invasive procedure may not lead to the substantial improvement in PD symptoms. The goal of the study is to build an individualized brain connectome to later embed it into The Virtual Brain (TVB) platform for human brain activity simulation. The long-term aim is to predict the outcomes of DBS in PD patients to plan successful treatment. Materials and methods: Ten PD patients underwent STN DBS implantation surgery. Preoperative T1 and diffusion MRI images and a postoperative CT image were collected. A structural T1 MRI scan was processed with Freesufers recon-all pipeline to obtain tissue segmentation and reconstruction of the cortical surfaces. Diffusion weighted images were processed using MRtrix3, performing artefact corrections, constrained spherical deconvolution and tractography. The structural T1 image was rigidly registered with the diffusion image in order to project the cortical and subcortical parcellation of the brain onto the reconstructed tracts. The Desikan and DISTAL atlas for cortical and subcortical brain areas were used, respectively, to obtain a structural connectome. The DBS electrode was identified on the postoperative CT image and registered with the T1 image. Results: The structure of the virtual brain model from the patient T1 MRI neuroimaging data was built. The connectome including subcortical areas of globus pallidus internal, globus pallidus external and STN was calculated. The DBS electrode was projected into the virtual brain model to simulate the volume of tissue activated by electric stimulation. Conclusions: The connectivity extracted will be embedded in the TVB platform to model the activity of the patient brain in deep brain stimulation conditions. A dynamical neural mass model will be equipped to every node of the connectome to simulate neural activity. Perturbations of the dynamics will be modelled by a realistic stimulus through the virtualized DBS electrodes. Patient specific virtual brain modelling will improve our understanding in inter-individual outcomes of DBS treatment.