Waveforms for Neural Stimulation

Nondamaging

While generating the degree and pattern of activation required for therapeutic efficacy or functional restoration, the applied stimuli must not damage either the tissue where they reside or that they stimulate or the electrodes that are used to deliver stimulation. The mechanisms underlying stimulation-induced injury of neural tissue remain unclear, and empirical studies are required to document risks of new stimulation paradigms.

The most successful attempt to harmonize a range of empirical data is the Shannon model of stimulation-induced tissue injury ( ). The model defines a line separating stimulation regimes in which tissue damage was observed from regimes in which tissue damage was not observed, and the equation of this line is given by

where Q is the charge per phase of the stimulation pulse (equal to the current amplitude times the pulse duration for a rectangular pulse), Q/A is the charge density (the charge per phase divided by the geometric surface area of the stimulation electrode), and k is an empirically determined constant.

Although useful to provide guidance for selection of nondamaging stimulation parameters, the Shannon model is a fit to empirical data, and thus is valid only for the conditions under which those data were collected. In addition to charge density (Q/A) and charge per phase (Q), the interactions between the stimulation waveform and the potential risk for tissue damage are affected by other factors including stimulation pulse repetition rate (frequency), stimulation duty cycle, current density (equal to the current amplitude divided by the geometric surface area of the stimulation electrode), electrode size, electrode material(s), and the region of the nervous system that is stimulated ( ). For example, although the Shannon model was used to establish a suggested limit on the parameters of nondamaging deep brain stimulation (DBS), the data used to inform the form and parameterization of the model are quite different from both the stimulation parameters of DBS and the brain locations where DBS is typically delivered ( ). In this chapter I primarily focus on the efficacy of stimulation waveforms and while I do consider, for example, the importance of biphasic waveforms to reduce the risk of potential damage to the electrode or tissue, the reader is referred to other sources for a more comprehensive review of stimulation-induced damage ( ).

Selectivity

Selectivity refers to the ability to activate the neurons, neural elements (e.g., presynaptic terminals), or nerve fibers that are targeted for stimulation and intended to produce the desired therapeutic effect while not activating nontargeted neurons, which might, for example, produce unwanted side effects. When considering peripheral nerve stimulation, this selectivity could include both nerve fiber diameter selectivity, that is, the ability to stimulate selectively nerve fibers with a specific range of diameters (e.g., ), as well as nerve fiber position selectivity, that is, the ability to stimulate selectively nerve fibers lying in a particular position within a compound nerve (e.g., ). When considering spinal cord stimulation, selectivity can refer to the ability to activate dorsal column fibers without activation of dorsal root fibers (e.g., ), or the ability to activate dorsal column fibers with the targeted dermatome(s) where pain is experienced ( ). Finally, for brain stimulation, this could include stimulation location, including the subthalamus for tremor control ( ), discrete targets within the pallidum ( ) or subthalamic nucleus ( ), or a specific region of cortex ( ). Alternatively, selective brain stimulation could refer to the ability to activate selectively particular neural elements ( ), for example, antidromic activation of projection axons or activation of local cells without concomitant activation of neighboring or even intermingled axons of passage ( ). In all cases, such selectivity is influenced by the electrode geometry, as well as the stimulation waveform.

Efficiency

The energy required for a particular waveform to generate neural activation is an important consideration, as this will contribute to the battery life of an implanted pulse generator. An additional important consideration related to efficiency is the energy required by the hardware to generate the stimulation waveform. While there may be modest savings in threshold energy associated with stimulation waveform shapes ( ), these savings may be offset by the power consumed by the electronics to generate such waveforms. Thus, the design of the stimulation hardware is an integral component of the efficiency of any particular stimulation waveform ( ).