Technologies to study neural activity
Recording the activity of a sufficient number of neurons at significant time scales and spatial distributions is one of the main challenges for a better understanding of how neural ensembles work.
Optical methods are increasingly used because they allow activity over a large area to be monitored from the same layer of brain tissue. In addition, they are inherently limited to recording from superficial structures of the brain or require the use of probes or surgery to provide access to deep brain regions.
A number of innovations have been made in electrical recording using flexible materials on the surface of the brain, and microelectrodes have long been the norm, but their number of channels is limited due to connection, volumetric displacement and tissue damage.
At the same time, electronics continue to evolve at a rapid pace, but few of these technological improvements make their way to neuroscience in vivo.
A new strategy for interfacing chips with three-dimensional micro-wire arrays
This neural interface consists of a bundle of isolated micro-wires coupled perpendicularly to imaging networks such as those found in camera chips.
By organizing them into bundles, the authors control the three-dimensional structure of the distal end, with a robust parallel contact plane on the proximal side which is coupled to an array of pixels.
The density of the microwires for the proximal end and the distal end can be independently adjusted, allowing wire-to-wire spacing to be customized as required.
Design and Manufacturing
- (A) Procedure for manufacturing bundles of micro-wires.
- (i) The individual micro-wires are electrically insulated with a robust ceramic or polymer coating.
- (ii) A sacrificial layer is applied to the wires to ensure spacing.
- (iii) The tips of the micro-wires can be shaped with an angular tip..
- (iv) The wires are then grouped together by winding the wire or by mechanical aggregation. The threads pile up naturally in a honeycomb network.
- (v) The bundle is infiltrated with biomedical epoxy to hold the wires together, then the upper (proximal) end is polished to mate with the CMOS chip.
- (vi) The proximal end is etched from 10 to 20 μm to mate with the CMOS chip and the distal end of the wires is released by etching.
- (B) An electron microscope backscatter view of an individual microwire.
- (C) The wires are grouped in a structure in honeycomb and epoxy is infiltrated in between to fill in the gaps.
- (D) Proximal end of a 177 bundle wires after etching to expose the common thread.
- (E) Expected volumetric displacement of the micro-wire bundles as a function of the wire-to-wire distance, determined by the wire size and the thickness sacrificial coating.
- (F) The distal end of a 600 bundle 7.5 μm W wires covered with 1 μm glass after etching (G and H). The distal end can be precisely shaped to simultaneously access different depths in the tissue.
Compatibility with different imaging chips and tests on a retina and on a motor cortex
Tests with different imaging matrices
The process described is very flexible and agnostic as to the identity of the chip; the authors successfully interface with the imaging matrix chip of a Xenics Cheetah camera, with an organic light emitting diode display chip from Olightek and with a multi-electrode matrix device.
Retinal tests
To test the ability of the completed device to record neural activity on a flat surface, the authors used an ex vivo preparation of rat retina. A dialysis membrane held a small piece of isolated retina against the bundle in an infusion chamber, then a 138-thread bundle was lowered into contact with the retina. The recorded spikes exhibited typical unit signatures, i.e. a detected action potential located on a wire with smaller peaks on the adjacent wires. These retinal recordings demonstrate the system's ability to record individual units at high data acquisition rates and a high signal-to-noise ratio.
Test on a motor cortex
Next, the researchers tested whether it was possible to record neural activity in the deep cortical and subcortical areas across a large spatial region in rodents in vivo. The recordings were made within 2 hours of implantation of the bundle into deep layers of the motor and somatosensory cortexes and of the dorsal striatum. The mice were allowed to run on a spherical treadmill, in a state of head restraint during the recording. Neural activity was easily observed in most of the wires in the bundle through a horizontal layer. Approximately 100 to more than 200 putative neurons were reliably identified over a large horizontally extended area in each recording during a typical 5-minute recording session.
Conclusion
Conventional two-photon imaging is generally limited in time and space, so bundles of micro-wires coupled to CMOS arrays can simultaneously record the activity of hundreds of neurons. In addition, the flexibility along the length of the distal end of the beam allows precise recordings from normally inaccessible areas, such as the striatum.
