Novel diamond shuttle to deliver flexible neural probe with reduced tissue compression

Abstract

The ability to deliver flexible biosensors through the toughest membranes of the central and peripheral nervous system is an important challenge in neuroscience and neural engineering. Bioelectronic devices implanted through dura mater and thick epineurium would ideally create minimal compression and acute damage as they reach the neurons of interest. We demonstrate that a three-dimensional diamond shuttle can be easily made with a vertical support to deliver ultra-compliant polymer microelectrodes (4.5-µm thick) through dura mater and thick epineurium. The diamond shuttle has 54% less cross-sectional area than an equivalently stiff silicon shuttle, which we simulated will result in a 37% reduction in blood vessel damage. We also discovered that higher frequency oscillation of the shuttle (200 Hz) significantly reduced tissue compression regardless of the insertion speed, while slow speeds also independently reduced tissue compression. Insertion and recording performance are demonstrated in rat and feline models, but the large design space of these tools are suitable for research in a variety of animal models and nervous system targets.

Keywords: Engineering; Materials science.

Fig. 1. Insertion concept, design concept, and fabrication of a ultra-nano-crystalline diamond (UNCD) shuttle.

a Progression of design improvement from a simple silicon shuttle, with a buckling load of P_cr, to an improvement of 13.3*P_cr by changing the Young’s Modulus, E, and moment of inertia, I. b Process flow of UNCD shuttle requiring only two masks (not shown) at steps 1 and 3. c SEM of released UNCD shuttle. d Insertion of flexible electrode array into a dorsal root ganglia (DRG) using a rigid shuttle and retraction

Fig. 2. Simulation result of blood vessel invasion comparing Si (dark gray) and T-UNCD (red) having an identical buckling strength.

Examples of invasion analysis with a T shape with 65 µm by 11 µm on the planar portion and 27.5 µm-deep, 16 µm to 2 µm-wide trapezoid for the vertical support, which has identical buckling strength with (b). b 65 µm by 34 µm rectangle. c Simulation result of 1000 insertions yielded a Gaussian distribution of the number of invaded blood vessels. Mean values of two geometries from (a, b) were 6.23 and 9.84, respectively. Invaded blood vessels in (a, b) indicated in magenta. X and Y labels in (a, b) indicated the coordinates in µm

Fig. 3. Mechanical characterization of various shuttles in a tissue phantom.

a Buckling load according to material, geometries and lengths, including two silicon geometries that are commercially available, N = 7. Dashed line shows calculated value. b Box plot results of insertion force of single shank shuttles, N = 10. c Box plot results of insertion force of 1, 2, and 4 shanks for three different probes, N = 7 (planar diamond not tested). For 4-shanks, the force per shank was 7.8, 8.2, and 9.8 mN for T-UNCD, Si-15, and Si-50, respectively. The inter-shank spacing was 250 µm. Insertion speed was 0.01 mm/s in all cases

Fig. 4. Result of ex-vivo insertions into cat DRG.

a 15-µm thick Si. “x” marks an insertion failure. b 11-µm thick planar UNCD. “x” marks an insertion failure. c T-shaped UNCD and 50-µm thick Si. d Generic boundary conditions for the buckling load calculation (K = 0.699, K = 1, and K = 2, respectively). The probe backend, at top, is fixed in each scenario and tissue is at the bottom. Theoretical limits noted as dashed lines in (a, b, c)

Fig. 5. Result of T-UNCD phantom insertion with and without oscillation according to insertion speed.

ANOVA results with an interaction term included. Both speed and oscillation were significant but the interaction term was not. Values for the maximum compression distance were taken from video analysis

Fig. 6. Assembly of high density flexible array mounted on UNCD shuttle.

a Photograph of entire assembly including the flexible polyimide array mounted on a UNCD shuttle. PCB is secured by a removable jacket that slides in a track of the shuttle jig. The insertion motor (not shown) is connected to the 3D printed shuttle jig. b Top view of PEG-coated UNCD shuttle with coating 1–2 µm thick. c Polyimide-based flexible array with 60 recording sites and (inset) zoomed view of the array tip

Fig. 7. Recorded neural signals from feline sacral DRG for evoked stimuli.

a Diagram of the array in DRG. On the left is a representative hematoxylin and eosin stained cross-section of an example DRG with the ventral root (VR) below. Overlaid is a CAD diagram of the array, with array sites in yellow. The array and DRG are to scale. b Wideband signal (highpass filter with 2 Hz cutoff) recorded during scrotum brushing. c High pass filtered signals recorded during two sensory input trials (scrotum and tail brush strokes indicated by black lines at top). Channels shown are indicated by the gray vertical line to the right of (b). Manually sorted single units are in color. d Peri-stimulus time histograms (PSTHs) for two channels and trial types, labeled and colored by channel numbers shown in (c). For both units shown, brushing starts at 0 s and continues through 5 s, including multiple individual strokes. Overlaid on each PSTH are the corresponding spike waveforms, with the mean waveform given as a dark thick line of the same color