Formation of Anisotropic Conducting Interlayer for High-Resolution Epidermal Electromyography Using Mixed-Conducting Particulate Composite

Abstract

Epidermal electrophysiology is a non-invasive method used in research and clinical practices to study the electrical activity of the brain, heart, nerves, and muscles. However, electrode/tissue interlayer materials such as ionically conducting pastes can negatively affect recordings by introducing lateral electrode-to-electrode ionic crosstalk and reducing spatial resolution. To overcome this issue, biocompatible, anisotropic-conducting interlayer composites (ACI) that establish an electrically anisotropic interface with the skin are developed, enabling the application of dense cutaneous sensor arrays. High-density, conformable electrodes are also microfabricated that adhere to the ACI and follow the curvilinear surface of the skin. The results show that ACI significantly enhances the spatial resolution of epidermal electromyography (EMG) recording compared to conductive paste, permitting the acquisition of single muscle action potentials with distinct spatial profiles. The high-density EMG in developing mice, non-human primates, and humans is validated. Overall, high spatial-resolution epidermal electrophysiology enabled by ACI has the potential to advance clinical diagnostics of motor system disorders and enhance data quality for human-computer interface applications.

Keywords: EMG; anisotropic conductors; conducting polymers; organic bioelectronics.

Figure 1

EIS of cutaneous tissue. A) Schematic cross‐sectional circuit model demonstrating an EMG voltage source (V_EMG) and two overlying recording electrodes. Z_int, Z_Gel and Z_T denote electrode/tissue interface, ion‐conducting interlayer, and subcutaneous tissue impedances, respectively (top). Diagram of electrode placement. WE: work electrode, WS: work sense electrode, REF: reference electrode, CE: counter electrode; L marks the distance between WE and REF electrodes (bottom). B) In vivo EIS using 3‐ (black) and 4‐port (blue) configuration with various distances between electrodes (L). Darker shades represent shorter distances. C) Bulk tissue impedance (4‐port; blue) varies with electrodes distance whereas electrode/tissue interface (3‐port; black) impedance remains constant. Impedance values were taken at 1 kHz. D) In vivo EIS using 3‐ (black) and 4‐port (blue) configuration with various electrode sizes (D). Darker shades represent smaller sizes. E) Electrode/tissue interface (3‐port; black) impedance varies with electrode size whereas bulk tissue impedance (4‐port; blue) remains constant. Impedance values were taken at 1 kHz

Figure 2

ACI enables robust anisotropic conducting electrode/tissue interface. A) Illustration of ACI fabrication process: 1) conducting flakes are generated by baking a PEDOT:PSS‐based dispersion; 2) conducting flakes are pulverized using an IPA‐based ball mill; 3) particles are filtered and mixed with the adhesive scaffolding polymer (top). Cross‐sectional schematic illustrating ACI particles between skin and electrodes. Note there is no conducting path between neighboring electrodes. B) Histogram of skin roughness measured by 2D profilometry. Inset: tilted SEM of ACI particles applied on the surface of artificial skin. Scale bar, 50 µm. C) Optical micrograph of high‐density NeuroGrid for epidermal electrophysiology. Scale bar, 5 mm. Inset: optical microscopy of NeuroGrid. Scale bar, 100 µm. D) EIS of 450 × 450 µm² PEDOT:PSS electrodes with ACI (dashed blue) and conductive gel (dashed red) interlayers. Inter‐electrode impedance (L = 2 mm) with ACI (solid blue) and conductive gel (solid red) interlayers. Shaded areas represent standard errors (n = 3).

Figure 3

ACI enables high SNR, low‐crosstalk EMG recording in developing rodents. A) Optical image of the mouse pup EMG recording setup. Scale bar, 5 mm (top). Sample time‐frequency spectrogram of epidermal electrical signal showing both ECG and EMG. White trace shows raw acquired data. Scale bar, 50 ms, 200 µV (bottom). B) Comparison of normalized power of band‐passed EMG signal (50–1000 Hz; black) between ACI and conducting gel interlayers. Warmer colors represent higher power. ACI provides a more defined spatial extent (N_ACI = 1527, N_gel = 1483 EMG epochs). C) Power coherence of EMG signal as a function of electrode distance. ACI (blue) demonstrates rapid decay in power with distance compared with conductive paste (black). Shaded areas represent standard error (N_ACP = 1527, N_gel = 1483 EMG epochs).

Figure 4

ACI permits non‐invasive detection and clustering of single MU activity from behaving non‐human primates. A) Optical image of electrode array conforming to surface of non‐human primate arm. Scale bar, 2 mm. B) Sample raw traces of non‐human primate EMG. Scale bar, 10 ms. C) Example auto‐correlograms of clustered putative MUs acquired from non‐human primate during an isometric motor task. D) Inter‐spike interval (ISI) distribution of a MU active during four different applied force levels. N_25% = 1712, N_50% = 2587, N_75% = 1987, N_100% = 1929 spikes. Inset: scatter plot of ISI versus applied force.

Figure 5

ACI enables capturing of MUs from the surface of the human forearm with accuracy to decode individual finger movements. A) Optical image of NeuroGrids conforming to the forearm of the human subject. Scale bar, 1 cm (top). Sample raw traces of human EMG recorded from the forearm (red). Scale bar, 5 mV, 100 ms (bottom). B) Trigger‐averaged bandpass‐filtered (50‐1000 Hz) EMG traces reveal the spatial distribution of a MU with corresponding power indicated by the superimposed colormap. Scale bar, 100 ms, 1 mV. Channels with insufficient quality for analysis are noted in gray. C) Sample auto‐correlogram (red) of MUs acquired from the surface of forearm. D) Scatter plot of isolation distance versus L‐ratio of MUs with inter‐electrode spacing of 2 mm (blue, n = 46 MUs), 4 mm (red, n = 49 MUs), and 6 mm (yellow, n = 60 MUs). Inset: normalized amplitude of human MUs (n = 22 MUs) as a function of electrode distance. Shaded regions indicate standard error. E) Schematic of the experimental setup showing participant hand inside a 3D‐printed box with four load cells (left). Sample finger pressure measured by load‐cell. Scale bar, 10 s, 20 N (right). Task involved pressing on the load cells in sequence (1: index finger, 2: middle finger, 3: ring finger, 4: pinky). F) Scatter plot of firing rate as a function of force during index (blue), middle (orange), ring (yellow), and pinky (purple) finger pressing epochs. G) Goodness of fit (R²) values for measured versus decoded force values across sessions. Binning window = 250 ms. 1: 0.77671 ± 0.021582, 2: 0.81133 ± 0.016429, 3: 0.80792 ± 0.019255, 4: 0.71288 ± 0.017669, N_recording = 10 sessions, N_epochs = 49986 (Time bins). H) Raster plot of MU activity over the course of 4 sequential finger movements sorted according to the firing rate of the MUs (top). Decoded finger pressure (solid lines) from MU activity compared to actual pressure sensor output (dashed lines; bottom). Scale bar, 10 s.