Fig 1.
Recording setup and data analysis.
Top left panel: Schematic of the position of the electrodes to record EEG (white), ESI (black), ECG (red), somatosensory activity at Erb’s points (green), and the positions of Common Mode Sense (CMS, blue) and Driven Right Leg (DRL, orange) electrodes. Note that Erb and ECG electrodes were placed on the chest, while ESI electrodes were placed on the back. Top right and bottom panels: Flowchart describing the analysis procedure. (1 and 2) Raw ESI signals are first re-referenced to the most caudal dorsal electrode (S64). (3) The artifact caused by the electrical stimulation of the median nerve is removed by linear interpolation. (4) The ECG is used to identify ESI time windows contaminated by the QRS complex (orange). This allows subsequent selection of ESI time windows to retain. (5) These time windows are epoched around the somatosensory stimulus. (6) An amplitude-based threshold is used to identify and remove artifactual epochs (red). (7) Resulting artifact-free epochs are averaged across stimuli of each block. Subject-level average waveforms are subsequently averaged across participants. (8) Images of the spatial distribution of spinal cord activity are calculated by spline interpolation across ESI electrodes. The figurine depicts spinal cord activity at the latency of the N13 wave.
Fig 2.
Neural activity along the somatosensory pathways in response to transcutaneous electrical stimulation of the median nerve at the wrist.
Responses recorded from the ipsilateral brachial plexus (eP8 and eN9, bottom row), the spinal cord (sP9, sN13, and sP22, middle row), and the cerebral cortex (cN20-cP25, top row). Left column: red waveforms are group-level averages, while the gray shade represents the standard deviation (±1 standard deviation). Right columns: dorsum and scalp maps obtained using cubic interpolation of multielectrode recordings (dorsum: 63 electrodes; scalp: 32 electrodes). Maps are shown at the latency of the eN9, sN13, and cN20 peaks. Red circles show the electrodes from which waveforms shown in the left column are extracted. Green circles show the reference electrode for both waveforms and maps: dorsal recordings are referenced to the most caudal electrode (S64), whereas scalp recordings are referenced to Fz. Enlargements show areas of maximal response amplitude. The data underlying this figure can be found at https://doi.org/10.17605/OSF.IO/KHJCG.
Fig 3.
Electoral Spinal Imaging (ESI): High-resolution images of spinal cord activity during somatosensory stimulation.
Bottom panel: Recordings obtained using 97 electrodes (2 electrodes located on the right and left ERB points, 63 dorsal electrodes spanning from the cervical (C2) to the thoracic (T8) segments, and 32 scalp electrodes). The electrode positions shown in the bottom right figurine and their recording of the response elicited by somatosensory stimulation (right) are color-coded such that neighboring channels have similar colors. Signals were referenced as described in Fig 1. Middle panel: Enlargement of the responses between 8 and 26 ms poststimulus. Top panel: Lower row shows dorsal and scalp maps with 1-ms resolution. The eN9, sP9, sN13, sP22, cN20 and cP25 peaks are labeled. Upper row enlarges the dorsum maps of spinal cord activity between 11 and 16 ms poststimulus. Note the peak of postsynaptic activity occurring around 13 ms poststimulus. The data underlying this figure can be found at https://doi.org/10.17605/OSF.IO/KHJCG.
Fig 4.
Latency distributions of the sP9 and sN13 waves.
Group-level data. Left column: sP9 (top) and sN13 (bottom) waves recorded from the top dorsal electrodes. Insets show the response over a larger time window; the box indicates the intervals around sP9 and sN13. Dots indicate the peak latency recorded from each electrode. The black dot and the horizontal black bar indicate the mean latency and latency range, respectively. Right column: Maps of the latency delays, with color-coding of time relative to the electrode with the shortest latency (highlighted in white, used as time 0). Note how, for the sP9 wave, the electrodes with shortest (whiter regions of spinal electrode topography, broadly corresponding to the C6-C7 segment) and longest latencies (redder regions) show a clear spatial gradient, indicating that the sP9 wave travels in the lateral-medial and caudal-rostral directions. Compare with the lack of any clear spatial gradient in the sN13. See also S3 Video. The data underlying this figure can be found at https://doi.org/10.17605/OSF.IO/KHJCG.
Fig 5.
Lateralization of neural responses along the somatosensory pathways.
Top and bottom panels: Neural responses recorded along the somatosensory pathways following the stimulation of the median nerve at the left and right wrist, respectively. Top and bottom panels display the group-level response superimposed from all 97 electrodes, referenced to S64 for the spinal electrodes, and to Fz for the scalp electrodes. Electrode position and responses are color-coded according to the scheme shown in the inset. These panels also contain dorsum and scalp maps at the latency of the eN9, sN13, cN20, sP22, and cP25 peaks. Middle panel: Difference maps obtained by subtracting the right response from the left response, at the same latencies as the maps shown in the top and bottom panels. Yellow circles identify electrodes showing statistically significant effects (p < 0.01, cluster-corrected in both time and space; see Methods for details). Note the clear lateralization of the response recorded from the brachial plexus (eN9) and the brain (cN20 and cP25), and the lack of lateralization of the spinal responses (sN13 and sP22). The data underlying this figure can be found at https://doi.org/10.17605/OSF.IO/KHJCG.
Fig 6.
Attentional modulation of spinal cord activity.
Top and bottom panels: Neural responses recorded along the somatosensory pathways under the conditions of letter memorization (top panel), and attention to somatosensory stimulation (counting omissions, bottom panel). Both panels show the group-level response distribution on the dorsum and scalp at the eN9, sN13, cN20, sP22, and cP25 peak latencies. Electrode position and responses are color-coded according to the scheme shown in the inset. Middle panel: Difference maps obtained by subtracting the responses elicited during the omission task from those elicited during the letter memorization task, at the same latencies as the maps in the top and bottom panels. Yellow circles identify electrodes showing statistically significant effects (p < 0.01, cluster-corrected in both time and space; see Methods for details). Note the significant difference in the cluster of spinal electrodes centered on the sP22, and the lack of differences in the early-latency cortical SEPs. The data underlying this figure can be found at https://doi.org/10.17605/OSF.IO/KHJCG.
Fig 7.
Neural generators of the recorded responses.
Schematic representations of the putative spinal mechanisms generating the electrocortical responses recorded with ESI. The sP9 appears as a traveling wave (see also the delay map in Fig 4) as it reflects the current sink entering the cord through the dorsal roots and traveling rostrally along the dorsal column tracts. In contrast, the stationary sN13 reflects the segmental postsynaptic activity occurring in the deep dorsal horn following the first arrival of somatosensory input. Finally, the late sP22 is not directly consequent to a local, segmental effect of the incoming somatosensory input, but instead reflects stimulus-triggered activation of a long-loop circuit involving supraspinal structures that, in turn, project top-down to the spinal cord. For this reason, the sP22 encompasses spinal segments rostral and caudal to those where the sP9 and sP13 are recorded (Fig 3 and S1, S2 and S3 Videos).