The differences among IPG waveforms and their resulting effects on neural activation have two important consequences. First, the presence of different IPG models within a patient population is a previously unidentified source of variance in clinical studies of DBS. This could be particularly evident when examining changes in activation thresholds for side effects or therapeutic parameter settings. Second, the waveforms generated by the IPGs differ from the idealized stimulation parameters specified by the clinical programming device. This creates errors when calculating the voltage distribution in the tissue medium or the amount of charge or energy injected into the tissue. Both of these effects are exacerbated for higher frequencies and longer pulse widths.
Three general characteristics define the stimulus waveforms generated by the Soletra and Kinetra. First, the charge injected into the tissue during the cathodic and anodic phases is equal (charge-balanced) for tissue safety considerations. Second, in contrast to the behavior of the external pulse generators used during surgery (Trottenberg et al., 2004
), the anodic phase duration is shorter than the stimulation period in ways that are IPG and frequency dependent (). Third, and most importantly, the programmed stimulation voltage is roughly equal to the difference between the cathodic and anodic peaks rather than the magnitude of the cathodic pulse alone. These characteristics, when combined together, can result in cathodic pulse magnitudes that are substantially smaller than the programmed stimulation values.
Several important differences exist between the two IPG models. The most obvious clinical difference is that the Kinetra supports two DBS leads for bilateral stimulation, while the Soletra supports only one lead. The two IPG models also have different battery characteristics. The Soletra battery voltage is 3.7V (new) and uses a voltage multiplier circuit to achieve amplitudes above the battery voltage. The multiplier circuit, which is employed for stimulus amplitudes >3.6V, substantially increases power consumption. The Kinetra battery voltage is 3.2V (new) and battery power consumption increases linearly as a function of voltage across the entire stimulus amplitude range of the device.
When comparing the stimulation waveform construction, the Soletra and Kinetra use qualitatively similar strategies. For scientific studies concerned with explicit representation of the details, our analysis suggests that the overall waveform can be broken into four distinct phases: 1) cathodic phase (duration set by the user), 2) an inter-pulse delay at 0 V between the cathodic and anodic phase, 3) anodic phase (duration dependent on stimulation pulse width and frequency), and 4) a final delay at 0 V between the end of the anodic phase and beginning of the next waveform. The inter-pulse delay of the Soletra lasts ~0.5 ms and the final delay lasts ~3.8 ms. In contrast, the Kinetra interleaves two stimulation waveforms (one for each lead in bilateral stimulation) in each cycle, so the effective time interval used to construct the overall waveform is half the period defined by the stimulation frequency. The inter-pulse delay of the Kinetra lasts ~0.2 ms and the final delay is ~0.4 ms (until the start of the pulse on the bilateral lead). In both IPGs, the duration of the anodic phase is defined by the time remaining in the total period after subtracting the cathodic pulse width, inter-pulse delay, and final delay. Because of charge balancing, the ratio of cathodic to anodic magnitude can vary substantially across stimulation parameter settings. And because the amplitude registered on the programming devices is actually representative of the peak-to-peak voltage difference between the cathodic and anodic phases, the effective cathodic pulse amplitude is reduced.
The monophasic cathodic square wave used in our analysis would not be used to stimulate living tissue due to safety considerations. We used the monophasic waveform to illustrate the differences between the expected waveform parameters, as indicated by the programming device, and the actual waveforms produced by the IPGs (). In our experience, explicit representation of the actual IPG waveform is an important component of accurate models of the neural response to DBS (Butson and McIntyre, 2005
; Butson et al., 2006b
; Miocinovic et al., 2006
; Butson et al., 2007
In summary, the actual IPG waveforms should be taken into account for three aspects of DBS research: 1) developing computer models of DBS, 2) calculating charge or energy injected into the tissue during DBS, and 3) correlating DBS parameter settings with clinical outcomes and/or side effects. Further, the IPG model may be an unrecognized source of variance within a patient population, and these differences should be considered in multi-patient studies and patient programming.