High-density microfibers as a deep brain bidirectional optical interface

• Main Author: Perkins, Lewis Nathan
• Other Authors: Boas, David A.
• Language: en_US
• Published: 2019
• Subjects: Neurosciences
• Online Access: https://hdl.handle.net/2144/32998

Optical interrogation and manipulation of neural dynamics is a cornerstone of systems neuroscience. Genetic targeting enable delivering fluorescent indicators and opsins to specific neural subpopulations. Optic probes can fluorescently sense and convey calcium, voltage, and neurotransmitter dynamics. This optical toolkit enables recording and perturbing cellular-resolution activity in thousands of neurons across a field of view. Yet these techniques are limited by the light scattering properties of tissues. The cutting edge of microscopy, three-photon imaging, can record from intact tissues at depths up to 1 mm, but requires head-fixed experimental paradigms. To access deeper layers and non-cortical structures, researchers rely on optical implants, such as GRIN lenses or prisms, or the removal of superficial tissue. In this thesis, we introduce a novel implant for interfacing with deep brain regions constructed from bundles of hundreds or thousands of dissociated, small diameter (<8 µm) optical fibers. During insertion into the tissue, the fibers move independently, splaying through the target region. Each fiber achieves near total internal reflection, acting as a bidirectional optical interface with a small region of tissue near the fiber aperture. The small diameter and flexibility of the fibers minimize tissue response, preserving local connectivity and circuit dynamics. Histology and immunohistochemistry from implants into zebra finch basal ganglia (depth 2.9 mm) show the splaying of the fibers and the presence of NeuN-stained cells in close proximity to the fiber tips. By modeling the optical properties of the fibers and tissue, we simulate the interface properties of a bundle of fibers. Overlap in the sensitivity between nearby fibers allows application of blind source separation to extract individual neural traces. We describe a nonnegative independent component analysis algorithm especially suited to the interface. Finally, experimental data from implants in transgenic mice yield proof of principle recordings during both cortical spreading depolarization and forepaw stimulation. Collectively, the data presented here paint a compelling picture of splaying microfibers as a deep brain interface capable of sampling or perturbing neural activity at hundreds or thousands of points throughout a 3D volume of tissue while eliciting less response than existing optical implants.