Event Date:
September 12th 9:00 AM - 10:00 AM
NEC Seminar, 12 September 2025
Speaker: Ben Alexander
- Sears 43
Lab: Laboratory for Intelligent Machine Brain Systems (LIMBS)
PI: A. Bolu Ajiboye
Title: Fast and Accurate Mapping of Functional Electrical Stimulation Parameters Using the Strength-Duration Curve
Abstract: Functional electrical stimulation (FES) can restore motor control to people affected by spinal cord injury (SCI) or stroke through stimulation of peripheral nerves. Because every human nerve is unique, the inputs (stimulation parameters) and outputs (muscle activations) of FES systems must be characterized for each participant before functional movements can be created. Most existing systems use 2D recruitment curves (RCs), modulating either pulse amplitude (PA) or pulse width (PW), while holding the other parameter constant. This is because it has historically been too time intensive to characterize the full PA-PW space with a 3D recruitment surface (RS). However, there are many potential benefits of full RS characterization, including greater muscle selectivity, better fatigue mitigation, and higher precision of movement. We have described and validated a fast, accurate, and automatic method of characterizing the full PA-PW stimulation space by leveraging the strength-duration curve.
Stimulation parameters were systematically tested in a participant with tetraplegia due to SCI who is enrolled in the Reconnecting the Hand and Arm to the Brain (ReHAB) clinical trial. Multi-contact nerve cuff electrodes delivered low frequency twitch stimuli and electromyography (EMG) was recorded to determine the individual muscle responses. Three RS acquisition methods were tested with all of the available nerve cuff electrodes. An equally spaced grid (ESG) method naively sampled the entire stimulation space. A Gompertz function inspired (GFI) method algorithmically sampled the space assuming a specific surface fit equation (Freeberg et al. 2011). Finally, our novel SD curve (SDC) based method acquired one PW modulated RC at maximum PA and one PA modulated RC at maximum PW. SD curves were constructed between the two RCs and a surface was fit to them. R^2 values for each acquisition method were calculated by comparing the predictions of a given method to sampled test points from the middle 80% of the normalized EMG range. Our novel method enables faster characterization of FES parameters, making it feasible to map the full parameter space and laying the groundwork for more dexterous FES stimulation patterns.