Characteristics and stability of sensorimotor activity driven by isolated-muscle group activation in a human with tetraplegia

Abstract

Understanding the cortical representations of movements and their stability can shed light on improved brain-machine interface (BMI) approaches to decode these representations without frequent recalibration. Here, we characterize the spatial organization (somatotopy) and stability of the bilateral sensorimotor map of forearm muscles in an incomplete-high spinal-cord injury study participant implanted bilaterally in the primary motor and sensory cortices with Utah microelectrode arrays (MEAs). We built representation maps by recording bilateral multiunit activity (MUA) and surface electromyography (EMG) as the participant executed voluntary contractions of the extensor carpi radialis (ECR), and attempted motions in the flexor carpi radialis (FCR), which was paralytic. To assess stability, we repeatedly mapped and compared left- and right-wrist-extensor-related activity throughout several sessions, comparing somatotopy of active electrodes, as well as neural signals both at the within-electrode (multiunit) and cross-electrode (network) levels. Wrist motions showed significant activation in motor and sensory cortical electrodes. Within electrodes, firing strength stability diminished as the time increased between consecutive measurements (hours within a session, or days across sessions), with higher stability observed in sensory cortex than in motor, and in the contralateral hemisphere than in the ipsilateral. However, we observed no differences at network level, and no evidence of decoding instabilities for wrist EMG, either across timespans of hours or days, or across recording area. While map stability differs between brain area and hemisphere at multiunit/electrode level, these differences are nullified at ensemble level.

Trial registration: ClinicalTrials.gov NCT03161067.

Figure 1

Cortical recording sites and experimental methods. (A) Sites of the six bilaterally-implanted microelectrode arrays (MEAs), overlaid on MRI reconstruction of participant’s brain (CS: central sulcus). Arrays are labeled by anatomical target and pedestal (e.g. MA: primary motor cortex, pedestal A; SB: primary sensory cortex, pedestal B). (B) Experimental paradigm. Isolated muscle contractions (or contraction attempts) were cued by metronome ticks (click and pixel flash) at 4-s intervals. Electromyography (EMG) traces are from 4 example trials (repetitions), where the participant executed left extensor carpi radialis (ECR) contractions without co-contracting neighboring or opposite-limb muscles. (C) Temporal referencing and synchronization of neuromuscular and cortical data for a representative EMG-producing muscle (the left extensor carpi radialis: ECR). The upper plot shows event-referenced raw EMG in mV (reference line). Box plot shows movement cue time distributions relative to burst onset; pink regions are the interquartile range (IQR) of this distribution. The lower plot shows the peri-event time histograms (PETH) of multiunit activity (MUA) from an example motor channel on the contralateral motor array, in spikes/sec. Signals were referenced to EMG burst onset (dashed line; see Supplementary Information), and are trial averaged. Shaded regions show bootstrapped 95%–confidence intervals.

Figure 2

Regional body map for wrists, showing activity patterns across individual muscles. A: Overview of assessed muscles, color-coded by group. Targeted groups were: wrist extensors–-extensor carpi radialis; wrist flexors: flexor carpi radialis. B: Summary of channel-level representation of each muscle by brain area (M1: motor cortex arrays; S1: sensory cortex arrays) and laterality (contralateral, ipsilateral, or both). Counts are raw numbers, and percentages are with respect to all channels active for that region within the indicated brain area. C Activity maps for right hemisphere (MA: motor array, pedestal A; MB: motor array, pedestal B; SA: sensory array, pedestal A; SB: sensory array, pedestal B) and left hemisphere (MC: motor array, pedestal C; SC: sensory array, pedestal C). Channels with significant MUA responses to contractions in a body region are colored as in panel A. Solid colors indicate activity for only the contralateral side of the body (right side for MA/SA/MB/SB, left side for MC/SC); diagonal lines denote activity for the ipsilateral side; and black triangle overlays denote that both sides of the body elicited a response.

Figure 3

Spatial patterning and longitudinal stability activity from contractions of left wrist extensor (ECR: extensor carpi radialis). Longitudinal stability is considered relative to how often a channel is measured as active relative to the number of experiment sessions (11 total). (A) Frequency of activity across sessions for each channel distributed over arrays B and C (top: motor; bottom: sensory). Color code denotes the percentage of sessions (of 11) that a given channel was active, with higher percentages corresponding to greater longitudinal stability. (B) Probabilities that any active channel on Pedestals B and C responded for more than n sessions, within contralateral (left panel) and ipsilateral (right) hemispheres. Motor and sensory areas are pooled. Dashed lines mark the median number of sessions responsive among channels within the active footprint of each hemisphere. (C) Probabilities that any active channel on Pedestals B and C (mapped in panel A) responds for more than n sessions, measured for motor (left panel) and sensory (right) areas (hemispheres pooled). For example, a channel active for exactly 2 out of 11 possible sessions would be counted in the bars for both n = 1 and n = 2. Dashed lines mark the median number of sessions that a given channel in the active footprint of each area responds. (D) Distributions of the number of total sessions a channel responds to attempted left wrist extensions, by hemisphere. The median number of sessions a channel was observed responsive was greater within the contralateral than the ipsilateral hemisphere (n_{Median, Co}–n_{Median, Ip} = 5 sessions; p < 0.001). (E) Distributions of the number of total sessions a channel responded to attempted left wrist extensions (among all channels in active footprint), by area. The median number of sessions a channel within the active footprint was observed responsive was greater among sensory than motor arrays (n_{Median, M1}–n_{Median, S1} = -2 days, p = 0.183).

Figure 4

Within-channel (MUA) stabilities of left-ECR-related activity over time. X-axis denotes time scale of comparison (H: hours; D: days). Y-axis measures are normalized to the range [0, 1], higher values denoting greater stability. Bars show mean values + /- 1 SE. (*), (**), and (***) denote significance levels of 0.05, 0.01, and 0.001 respectively. (A) Firing strength stability significantly decreased from hours to days, and was higher for contralateral than ipsilateral channels. There was a significant hemisphere—by–timescale interaction, with a significant decrease in contralateral channel stability as the period between recordings varied from hours to days, but no corresponding significant difference in ipsilateral stability over timescale. (B) Firing dynamic stability (cross-correlation between z-scored PETHs) did not significantly decrease from hours to days in both hemispheres. There was a significant main effect of brain hemisphere, with contralateral channels being more stable than ipsilateral across hours and days. (C–D) Comparative stability across areas, pooled over brain hemisphere. (C) Firing strength stability (relative change in z-score of firing rate) by cortical area, over time. Stability significantly decreased over time from hours to days, with no significant difference between cortical areas. (D) Firing dynamic stability was invariant within areas from hours to days, but was significantly higher in sensory than motor channels.

Figure 5

Ensemble-level (network) stabilities of left-ECR-related activity over time. (A–B) Representative principal-component (PC) trajectories of neural ensemble, visualized in the PC1-PC2 plane, during left wrist extensions across typical sets of consecutive–hour (A) and consecutive–day (B) recordings, for the right (contralateral) hemisphere (i.e. motor and sensory channels aggregated). Trajectories reflect average neural activity within EMG bursts only, with endpoints designated at the onset (filled circles, variables t⁰), and terminal points (filled boxes and variables t^f) of the burst. Error is a normalized Euclidean distance between trajectories, with higher values indicating greater instability (see Materials and Methods). (C–D) Cumulative error between trajectory representations of ensemble activity, comprising the first six PCs, between consecutive hours, and consecutive days (Mean + /− 1 s.e). Channel ensembles are grouped by brain hemisphere (C) and area (D). (E–F) EMG of left wrist extensor (ECR: extensor carpi radialis), as measured from surface electrodes (blue) and as predicted from PCA trajectories (PCs 1–6: red) using Wiener filters trained on separate data. Data shown are for test (measurement) and training datasets recorded over consecutive hours (E) and days (sessions) (F). All envelopes are restricted to between burst onsets and terminations, and were time warped to an equal length to facilitate analysis. The x-axis shows the resulting time axis (in samples). The y-axis gives normalized envelope amplitude (see Supplementary Information). R² signifies the correlation between the actual EMG and the prediction. (G–H) Correlations (R²) between measured (actual) and predicted EMG, arranged by brain hemisphere (G) and area (H). As above, R² is based on training and test sessions spaced over consecutive hours and days (sessions).