Model-based rational feedback controller design for closed-loop deep brain stimulation of Parkinson's disease
Model-based rational feedback controller design for closed-loop deep brain stimulation of Parkinson's disease
复制标题
DOI：
10.1088/1741-2560/10/2/026016
复制
发表时间：
2013-04-01
影响因子：
4
通讯作者：
Sinha, A.
中科院分区：
文献类型：
--
作者：
Gorzelic, P.;Schiff, S. J.;Sinha, A.
客观的。探索使用经典反馈控制方法来实现改进的深部脑刺激（DBS）算法，以应用于帕金森病（PD）。方法。采用 PD 动力学计算模型来开发基于模型的理性反馈控制器设计。 PD 患者丘脑皮质中继能力的恢复被表述为反馈控制问题，其中 DBS 波形作为控制输入。测试了两种高级控制策略：一种由丘脑可靠性的在线估计驱动，另一种用于消除从苍白球内部（GPi）到丘脑的抑制的显着下降。受传统比例积分微分 (PID) 方法启发的控制律为每种策略规定，并在基底神经节网络的计算模型上进行模拟。主要结果。对于基于丘脑可靠性的控制，具有比例偏差的频率比例控制策略提供了针对给定能量消耗实现的最佳控制。相比之下，与基于可靠性的控制相比，基于 GPi 突触抑制输出的控制表现得非常好，并且相对于开环 DBS 的能量消耗进一步显着降低。最好的控制器性能是带有微分控制和积分偏置的幅度比例，这是完整的PID控制。我们演示了在这种情况下如何优化 PID 控制的三个组成部分是可行的，尽管这些优化函数的复杂性需要在实施中采用自适应方法。意义。我们的研究结果指出了基于模型的合理设计帕金森病反馈控制器的潜在价值。
Objective. To explore the use of classical feedback control methods to achieve an improved deep brain stimulation (DBS) algorithm for application to Parkinson's disease (PD). Approach. A computational model of PD dynamics was employed to develop model-based rational feedback controller design. The restoration of thalamocortical relay capabilities to patients suffering from PD is formulated as a feedback control problem with the DBS waveform serving as the control input. Two high-level control strategies are tested: one that is driven by an online estimate of thalamic reliability, and another that acts to eliminate substantial decreases in the inhibition from the globus pallidus interna (GPi) to the thalamus. Control laws inspired by traditional proportional-integral-derivative (PID) methodology are prescribed for each strategy and simulated on this computational model of the basal ganglia network. Main Results. For control based upon thalamic reliability, a strategy of frequency proportional control with proportional bias delivered the optimal control achieved for a given energy expenditure. In comparison, control based upon synaptic inhibitory output from the GPi performed very well in comparison with those of reliability-based control, with considerable further reduction in energy expenditure relative to that of open-loop DBS. The best controller performance was amplitude proportional with derivative control and integral bias, which is full PID control. We demonstrated how optimizing the three components of PID control is feasible in this setting, although the complexity of these optimization functions argues for adaptive methods in implementation. Significance. Our findings point to the potential value of model-based rational design of feedback controllers for Parkinson's disease.