Figure 3.
Signal chain diagram and summary of mitigation strategies of a generic aDBS system. (A) In the top row (light blue), proposed mitigation strategies to artifact susceptibility through signal chain. In the middle row (orange) the block diagram of key signal chain elements, with possible artifacts due to external triggers, such as ECG signal and movement artifacts; or due to stimulation. (B) Stimulation regimes: constant stimulation (e.g. 2 mA and 130 Hz) or varying stimulation following biomarker variations (aDBS). Rate of change in amplitude is defined by the stimulation ramp rate, with ‘slow’ ramps ranging 0.1–1 mA s−1 and ‘fast’ ramps from 1 to 10 mA s−1. Variable stimulation amplitude (red time-varying traces) results in transient step responses contaminating the LFP in the neighborhood of the stimulation electrode (black time-varying traces). Examples of biomarker signature (blue), stimulation artifact (green), and broadband artifact due to stimulation ramp (red) for a frequency-based detector/classifier. (C) Sense blanking is part of the analog chain and is always synchronously applied following the stimulation clock for a duration in the order of a few milliseconds (≫duration of stimulation pulse). By blanking the analog sense channel for a duration of ~ms the artifact in the sense channel due to the stimulation pulse can be rejected while the LFP signal is not missed. Typically the sense blanking duration falls in the order of the time between consecutive samples (1/sampling rate). Sense blanking is applied continuously, during both constant stimulation and adaptive stimulation. (D) Algorithm blanking is part of the digital chain and is synchronously applied to an algorithm detection event (e.g. biomarker input crosses predefined threshold Th). Algorithm blanking is defined with a certain duration based on biomarker physiology time/frequency dynamics and device specifications. By applying algorithm blanking to the input signal of the detector, the aDBS algorithm is blanked for that duration (hundred milliseconds to few seconds; typically at least for the duration of the stimulation ramp). By selecting appropriate algorithm blanking values, stimulation transition artifacts can be mitigated and detector performance improved. However, setting algorithm blanking parameters may be challenging for ‘fast’ aDBS because the duration of stimulation transients and time/frequency dynamics of the biomarker (e.g. beta band) converge at time scales of 1 s or less.