A Scalable Multi-Probe System for Simultaneous Neuropixel Recordings Across Multiple Cortical Regions in Nonhuman Primates

Abstract

Understanding how distributed motor circuits coordinate complex behavior requires recording technologies capable of simultaneously capturing large-scale, depth-resolved neural activity across multiple cortical areas in clinically relevant animal models. Existing systems for nonhuman primate (NHP) electrophysiology fail to meet this need: single-probe designs preclude simultaneous multi-area sampling, multi-probe designs suffer from mechanical instability at larger probe counts, and no existing platform integrates optical stimulation as a primary design constraint. This thesis addresses these gaps through two aims: the development and validation of scalable multi-probe Neuropixels fixtures for rhesus macaques, and the development of supporting software and phantom-based characterization tools for atlas-guided probe targeting.Two complementary fixture designs were developed to span the trade-off between spatial flexibility and simultaneous channel count. The three-probe linear fixture supports session-by-session repositionability across the rostrocaudal extent of motor cortex, consolidating three Neuropixels probes into a narrow linear body that preserves the approximately 10 mm accessible cortical footprint of the single-probe predecessor. The six-probe fixture maximizes simultaneously recorded channel count and integrates co-registered optical fibers for optogenetic stimulation, enabling dense, multimodal sampling of a fixed cortical footprint. Both fixtures share a single shared drive mechanism for simultaneous probe advancement, incorporate structural isolation between probe mounts to prevent electromagnetic cross-talk, and include a locking mechanism for probe retention during recording. Electrical characterization confirmed that inter-probe coherence in both fixtures was limited to environmental 60 Hz line noise, with no stimulus-locked cross-talk detected across the full neural recording frequency range. Phantom insertion testing using ballistic gelatin established that Neuropixels probe trajectories are highly repeatable under controlled conditions, with single-probe insertions across 15 trials producing tracks of maximum width 0.111 mm and depth 6.3 mm. Multi-probe testing revealed that probe shank curvature is the primary source of inter-probe separation error, producing deviations of up to approximately 1 mm from the 4.8 mm nominal spacing in the three-probe fixture when probes of varying visual straightness are used. Visual pre-screening for probe straightness is identified as a practical and effective mitigation. Brain-geometry phantom testing demonstrated that cortical surface curvature reduces net tip displacement relative to flat-phantom predictions, suggesting that flat-surface measurements represent a conservative upper bound on in vivo trajectory error. StereoPlan, a Python-based desktop application, was developed to fill the gap in NHP-compatible probe trajectory planning tools. Built around the NIMH Macaque Template v2.0 and the CHARM cytoarchitectonic atlas, it provides interactive 3D probe placement, atlas-labeled tissue-sequence readout, bend-tolerance cylinder sampling, and a population reach heatmap for target-driven entry point selection. Together, these contributions establish the hardware, software, and methodological infrastructure required to conduct simultaneous, multimodal, laminar-resolution recordings across distributed motor cortical circuits in a clinically relevant primate model, and provide a foundation for future in vivo validation and extension to additional cortical areas.