A Computational Study of Lower Urinary Tract Nerve Recruitment with Epidural Stimulation of the Lumbosacral Spinal Cord

Maria Jantz, Lucy Liang, Arianna Damiani, Lee E. Fisher, Taylor Newton, Esra Neufeld, T. Kevin Hitchens, Elvira Pirondini, Marco Capogrosso, Robert A. Gaunt2022 · DOI

This study presents a sophisticated computational framework for optimizing epidural spinal cord stimulation (SCS) to restore bladder function after spinal cord injury. By combining high-resolution MRI with biophysical nerve modeling, the researchers demonstrate how to selectively target the neural pathways responsible for voiding and continence.

TL;DR

Researchers developed a realistic MRI-based computational model of the cat spinal cord to predict how epidural stimulation recruits specific nerves involved in bladder control. The model successfully identifies stimulation parameters that selectively activate sensory fibers (pudendal afferents) while avoiding motor fibers, providing a non-invasive roadmap for improving clinical bladder therapies.

What Problem Does This Paper Address?

Lower urinary tract dysfunction is a debilitating consequence of spinal cord injury (SCI). While epidural spinal cord stimulation (SCS) has shown potential in triggering bladder reflexes, it is often a "black box" process. To achieve functional results—like controlled voiding or maintaining continence—clinicians must selectively recruit pudendal afferents (sensory fibers).

However, the spinal cord is a crowded environment. Stimulating the wrong area or using the wrong intensity can recruit pelvic efferents (motor fibers), which might cause the bladder to contract prematurely or interfere with the desired reflex. Finding the "sweet spot" for electrode placement and stimulation amplitude through surgical trial-and-error is difficult and inefficient. This paper addresses the need for a predictive tool to optimize these parameters before a patient ever reaches the operating table.

Methods

The authors built a "functionalized" finite element model (FEM) using a multi-step pipeline:

  • Anatomical Modeling: They used ultra-high-resolution (11.7 Tesla) postmortem MRI of a cat lumbosacral spinal cord to segment gray matter, white matter, CSF, and epidural fat.
  • Tractography: Diffusion Tensor Imaging (DTI) was used to map 200 unique axon trajectories per root level, ensuring the model reflected realistic neural geometry.
  • Electromagnetic Simulation: Using Sim4Life, they modeled the electric fields generated by SCS electrodes at two locations: Rostral (L6 vertebra) and Caudal (L7 vertebra).
  • Biophysical Axon Models: They coupled the E-field data with the NEURON simulator. They used different membrane models based on fiber size: the McIntyre-Richardson-Grill (MRG) model for large fibers (>4 µm), an adapted version for medium fibers (2-4 µm), and the Sundt model for unmyelinated small fibers (<2 µm).

Datasets

  • Imaging Data: High-resolution T2-weighted anatomical scans (50 μm resolution) and DTI datasets (200 μm resolution) of a cat sacral spinal cord.
  • Validation Data: The model's predictions were compared against previously collected in vivo experimental data from 6 cats, where nerve cuff electrodes recorded actual recruitment during SCS.
  • Anatomical Literature: Axon diameter distributions and nerve fiber counts were pulled from established neuroanatomical studies of the feline sacral roots.

Key Insights and Ideas

  • The "Safety Window": A critical finding is that afferents (sensory) are consistently recruited at lower amplitudes than efferents (motor). This suggests a "dynamic range" where SCS can trigger bladder reflexes without causing unwanted direct muscle contractions.
  • CSF as a Conductor: The model highlighted that moving electrodes caudally (toward the tail) lowers the stimulation threshold. This is because the increased volume of highly-conductive Cerebrospinal Fluid (CSF) and the decreased volume of the spinal cord in that region alter the electric field distribution.
  • Selectivity is Possible: The model confirmed that pudendal afferents can be recruited selectively over pelvic afferents at specific rostral locations, matching experimental observations. This is vital because different nerves trigger different components of bladder control.

Key Results

  • Model Accuracy: The simulated recruitment curves closely matched the experimental data obtained from cats, validating the model’s predictive power.
  • Rostral vs. Caudal:
    • At the rostral site (L6), pudendal afferents were recruited at lower amplitudes than pelvic afferents, allowing for better selectivity.
  • At the caudal site (L7), thresholds were lower overall. Pelvic afferents were recruited at ~50 μA, while pudendal afferents followed at ~100 μA.
  • Afferent Dominance: In all simulations, sensory fibers were activated before motor fibers, confirming that initial nerve recruitment via SCS has a negligible efferent component.

Limitations and Future Work

  • Missing Anatomy: The current model does not include the dura mater, which acts as an electrical insulator and could influence the absolute voltage thresholds required for recruitment.
  • Fiber Modeling: Small sensory axons (A-delta fibers) were modeled using an adapted MRG model rather than a specialized A-delta model, which might affect the precision of small-fiber recruitment predictions.
  • Future Directions: The authors plan to incorporate the dura mater and more physiologically accurate A-delta models to further refine the simulation's accuracy for human clinical translation.

Summary updated

Abstract

Bladder dysfunction is a major health risk for people with spinal cord injury. Recently, we have demonstrated that epidural sacral spinal cord stimulation (SCS) can be used to activate lower urinary tract nerves and provide both major components of bladder control: voiding and continence. To effectively control these functions, it is necessary to selectively recruit the afferents of the pudendal nerve that evoke these distinct bladder reflexes. Translation of this innovation to clinical practice requires an understanding of optimal electrode placements and stimulation parameters to guide surgical practice and therapy design. Computational modeling is an important tool to address many of these experimentally intractable stimulation optimization questions. Here, we built a realistic MRI-based finite element computational model of the feline sacral spinal cord which included realistic axon trajectories in the dorsal and ventral roots. We coupled the model with biophysical simulations of membrane dynamics of afferent and efferent axons that project to the lower urinary tract through the pelvic and pudendal nerves. We simulated the electromagnetic fields arising from stimulation through SCS electrodes and calculated the expected recruitment of pelvic and pudendal fibers. We found that SCS can selectively recruit pudendal afferents, in agreement with our experimental data in cats. Our results suggest that SCS is a promising technology to improve bladder function after spinal cord injury, and computational modeling unlocks the potential for highly optimized, selective stimulation. Clinical Relevance - This model provides a method to non-invasively establish electrode placement and stimulation parameters for improving bladder function with epidural spinal cord stimulation.
BibTeX
@inproceedings{jantz2022computational,
  author = {Maria Jantz and Lucy Liang and Arianna Damiani and Lee E. Fisher and Taylor Newton and Esra Neufeld and T. Kevin Hitchens and Elvira Pirondini and Marco Capogrosso and Robert A. Gaunt},
  title = {{A Computational Study of Lower Urinary Tract Nerve Recruitment with Epidural Stimulation of the Lumbosacral Spinal Cord}},
  booktitle = {2022},
  year = {2022},
  doi = {10.1109/embc48229.2022.9871292},
}
Figure