Objective: Despite the widespread use of spinal cord stimulation (SCS) for chronic pain management, its neuromodulatory effects remain poorly understood. Computational models provide a valuable tool to study SCS and its effects on axonal pathways within the spinal cord. However, these models must include sufficient detail to correlate model predictions with clinical effects, including patient-specific data. Therefore, the goal of this study was to investigate axonal activation at clinically relevant SCS parameters using a computer model that incorporated patient-specific anatomy and electrode locations. Methods: We developed a patient-specific computer model for a patient undergoing SCS to treat chronic pain. This computer model consisted of two main components: 1) finite element model of the extracellular voltages generated by SCS and 2) multicompartment cable models of axons in the spinal cord. To determine the potential significance of a patient-specific approach, we also performed simulations with standard canonical models of SCS. We used the computer models to estimate axonal activation at clinically measured sensory, comfort, and discomfort thresholds. Results: The patient-specific and canonical models predicted significantly different axonal activation. Relative to the canonical models, the patient-specific model predicted sensory threshold estimates that were more consistent with the corresponding clinical measurements. These results suggest that it is important to account for sources of interpatient variability (e.g., anatomy, electrode locations) in model-based analysis of SCS. Conclusions: This study demonstrates the potential for patient-specific computer models to quantitatively describe the axonal response to SCS and to address scientific questions related to clinical SCS.

were responsible for the study concept and design. Hans Zander developed the finite element model, and Robert Graham developed the sensory axon models. Carlos Anaya conducted the study, analyzed the data and prepared the manuscript draft, tables, and figures with input and guidance from Dr.

Objective. Spinal cord stimulation (SCS) is a common neurostimulation therapy to treat chronic pain. Computational models represent a valuable tool to study the potential mechanisms of action of SCS and to optimize the design and implementation of SCS technologies. However, it is imperative that these computational models include the appropriate level of detail to accurately predict the neural response to SCS and to correlate model predictions with clinical outcomes. Therefore, the goal of this study was to investigate several anatomic and technical factors that may affect model-based predictions of neural activation during thoracic SCS. Approach. We developed computational models that consisted of detailed finite element models of the lower thoracic spinal cord, surrounding tissues, and implanted SCS electrode arrays. We positioned multicompartment models of sensory axons within the spinal cord to calculate the activation threshold for each sensory axon. We then investigated how activation thresholds changed as a function of several anatomical variables (e.g. spine geometry, dorsal rootlet anatomy), stimulation type (i.e. voltage-controlled vs. current-controlled), electrode impedance, lead position, lead type, and electrical properties of surrounding tissues (e.g. dura conductivity, frequency-dependent conductivity). Main results. Several anatomic and modeling factors produced significant percent differences or errors in activation thresholds. Rostrocaudal positioning of the cathode with respect to the vertebrae had a large effect (up to 32%) on activation thresholds. Variability in electrode impedance produced significant changes in activation thresholds for voltage-controlled stimulation (38% to 51%), but had little effect on activation thresholds for current-controlled stimulation (less than 13%). Changing the dura conductivity also produced significant differences in activation thresholds. Significance. This study demonstrates several anatomic and technical factors that can affect the neural response to SCS. These factors should be considered in clinical implementation and in future computational modeling studies of thoracic SCS.