A novel methodology for exact targeting of human and non-human primate brain structures and skull implants using atlas-based 3D reconstruction

Abstract

BackgroundAccurate targeting of brain areas for stimulation and/or electrophysiological recording is key in many therapeutic applications and basic neuroscience research. Nevertheless, there are currently no end-to-end packages that accommodate all steps required for exact localization, visualization, and targeting regions of interest (ROIs) using standard atlases and for designing skull implants. New MethodWe have implemented a new processing pipeline that addresses this issue in macaques and humans including various preprocessing, registration, warping procedures, and 3D reconstructions, and provide a noncommercial open-source graphical software which we refer to as the MATLAB-based reconstruction for recording and stimulation (MATres). ResultsThe results of skull stripping were shown to work seamlessly in humans and monkeys. Linear and nonlinear warping of the standard atlas to the native space outperformed state-of-the-art using AFNI with improvements being more prominent in humans which had a more complex gyration geometry. The skull surface extracted by MATres using MRI images had more than 90% match with CT ground truth and could be used to design skull implants that conformed well to the skull’s local curvature. Comparison with Existing Method(s)The accuracy of the various steps including skull stripping, standard atlas registration, and skull reconstruction in MATres was compared with and shown to outperform the AFNI. The localization accuracy of the recording chambers designed with MATres and implanted in two macaque monkeys was further confirmed using MRI imaging. ConclusionsPrecise localization of ROIs offered by MATres can be used to plan electrode penetrations for recording and shallow or deep brain stimulation (DBS).