An Electrospinography Database of Spinal Cord Activity in Spinal Cord Injury Patients During Gait-Related Tasks and Motor Imagery

Description

The electroespinography (ESG) signals dataset comprises recordings from five patients with spinal cord injury (age: 45.20 ± 10.62, 20 % female participants).  The participants were provided with information about the experimental procedure and signed a written informed consent form, as well as consent for video and image recording. The study was carried out in accordance with the Declaration of Helsinki of the World Medical Association and was approved by the ethics committees of the University Hospital Complex of Toledo (Ref. No. 986; 29/03/2023) and Institut Guttmann (CEI 23/16).

The ESG signals were acquired using high-density surface electromyography (hHD-sEMG) equipment operating at a sampling frequency of 2 kHz. Signal amplification and acquisition were conducted with the Sessantaquattro+ system (OT Bioelettronica, Italy) and OTBioLab+ software. A reference electrode was placed on the right scapula. Spatial orientation data were recorded concurrently using the Sessantaquattro system.

Data were collected using two 32-electrode matrices (1 mm diameter, 8x4 arrangement, 1 cm inter-electrode distance) were positioned at the C4 vertebra (brachial plexus) and L2 vertebra (lumbar plexus).

In addition to ESG, the Sessantaquattro+ system simultaneously recorded orientation data from its integrated inertial measurement unit (IMU). The IMU consisted of a Bosch BNO055 sensor operating in Fusion (NDOF, Nine Degrees of Freedom) mode, providing synchronized quaternion estimates of body orientation. Orientation was represented by the four quaternion components (W, X, Y and Z) Although all data were acquired and stored at 2 kHz, the quaternion estimates were internally updated at 100 Hz and therefore repeated between successive sensor updates.

Experimental Protocol:

The experimental protocol was designed to capture detailed spinal cord activity under three conditions: baseline (resting), active movement execution, and motor imagery. It consisted of four stages.

1. Baseline Recording (Stages 1 and 4): Participants stood with eyes open for two minutes at the beginning and end of the protocol to record resting spinal cord activity.
2. Active Movement Execution (Stage 2): Participants alternated between walking and standing still, with each task lasting 7.5 seconds. Tasks were initiated by a one-second auditory signal and repeated four times per cycle.
3. Motor Imagery (Stage 3, with static tasks corresponding to even-numbered trials and movement tasks to odd-numbered trials): Participants alternated between mental relaxation for 7.5 seconds and kinesthetic motor imagery for 15 seconds. During motor imagery, participants imagined sensations in their lower limbs while walking. Tasks were performed under both stationary and walking conditions, alternatively.

The number of trials conducted at each stage was adjusted according to the participant’s comfort, ensuring that the entire experiment could be completed without undue strain. In this cohort, Stage 3 (motor imagery) trials were performed prior to Stage 2 (active movement execution); however, the original stage numbering has been retained to maintain comparability with the able-bodied dataset. A machine-readable manifest file, trial_description.csv, is included in the repository, specifying the stage to which each trial corresponds.

Data structure:

The dataset is organized into compressed zip archives, with each archive corresponding to a specific experiment and participant. The naming convention for the zip files follows the format: PXX_YYYYMMDD, where:

• PXX represents the participant code. 
• YYYYMMDD specifies the experiment date. 

The dataset is organized into compressed zip archives, with each archive corresponding to a specific experiment and participant: PXX_YYYYMMDD_Trial_XX", where Trial_XX ranges from 1 to 16.

A trial consists of a Matlab structure, data_structure, that stores all information related to the trial experiment. 

• experiment_data: Metadata about the experiment.
  • sample_frequency: Sampling frequency of the recordings.
  • time_vector: Time vector (seconds).
  • tasks_description: Structure describing task numbers and their descriptions.
  • task_vector: Vector indicating the task assigned to each sample.
• quaternions: Data related to spatial orientation.
  • labels: Labels for each quaternion recorded.
  • data: Matrix containing quaternion data (quaternions × samples).
• brachial_matrix:
  • ied_mm: Inter-electrode distance in millimeters.
  • electrode_configuration: Layout of the electrode matrix as seen on the participant’s back.
  • matrix_position: Reference point used to position the matrix.
  • matrix_description: Information about the matrix design.
  • null_electrodes: Binary vector indicating null electrodes (1 for null).
  • raw_data: Raw data matrix (electrodes × samples) in microvolts (µV).
  • ATS_data: Preprocessed data matrix (electrodes × samples) in µV.
  • outlier_mask: Binary matrix (electrodes × samples) indicating outlier samples (1 for outlier).
  • noisy_electrodes: Binary vector indicating noisy electrodes (1 for noisy).
  • filtered_data: ATS filtered data matrix (electrodes × samples) after substituting the outlier samples and noisy electrodes samples with NaN values, in µV.
• lumbar_matrix:
  • Contains the same fields as the brachial matrix: matrix_description (ied_mm, electrode_configuration, matrix_position), null_electrodes, raw_data, ATS_data, outlier_mask, noisy_electrodes and filtered_data.

Data records:

Due to device malfunction IMU data were not recorded for subject P02.