Deep brain stimulation (DBS) is a treatment for Parkinson’s disease (PD), and other neurological disorders. It uses electrodes implanted in the deep structures of the brain, involved in control of movements, and delivering electrical stimulation, generated by an implanted generator, to alleviate PD symptoms. While well-established and effective, DBS has several limitations. The clinical methods use fixed stimulation parameters, which do not adapt to the patient's needs in real time. This mismatch leads to decreased battery life, incomplete control of disease symptoms, and stimulation-induced side effects. Additionally, the process of parameter selection is long and manual, increasing the cost and lowering availability of this treatment.
Adaptive - or closed-loop - DBS (aDBS) is an extension of this method, using the measurement of the electrical activity of the brain to adjust the stimulation parameters in real time, responding to patient’s needs and changes in symptom severity. Several aDBS methods have been tested clinically with promising results but these studies are limited by low number of subjects and short duration of the experiment. This lack of data prevents us from drawing conclusions on what aDBS method offers the highest benefits while minimising the risks.
Computational modelling of deep brain stimulation allows for fast and reproducible testing of the proposed methods with no risk to the patients and additional possibility to investigate mechanisms of DBS action. While many computational studies of DBS have been published to date, they usually focus on a single method applied to a single model. Results from these studies are good indicators of the directions that the field can evolve towards, but they are limited by the assumptions present in the models. This calls for a systematic solution, enabling comparison of the efficacy and efficiency of the proposed methods, to ultimately make recommendations to the DBS device manufacturers, and the clinicians working in the field.
The aim of this project was to develop a systematic testing environment in which the proposed closed-loop algorithms can be tested against a range of computational models, to establish their relative strengths and weaknesses, and to use the results from this testing to propose novel aDBS algorithms.
As a result of this project, an adaptive DBS algorithm method has been implemented in a computational model of the brain structures implicated in Parkinson’s disease. The computational model has been developed in the direction of greater modularity, to further the goal of creating a systematic testing and verification environment. Additionally, conventional aDBS methods have been implemented in an embedded system, used in a study exploring effects of closed-loop DBS in rat model of Parkinson’s disease, creating a platform for future practical testing of the proposed methods, and proving that rats are a viable model organism for aDBS development.
