Howard Jay Chizeck

Prototype closed-loop deep brain stimulation systems inspired by Norbert Wiener

Implantable neurostimulators have rapidly become established methods of treating a variety of neurological disor-ders. The development of implantable neural interfaces enable the testing of Norbert Wieners hypotheses regarding neural disorders and their relationship to ideas of cybernetics. However, currently deployed medical devices of this kind are open-loop. For example, DBS for treatment of tremor does not take into account the variable and intermittent nature of the tremor. Closing the loop through sensors and real-time communication to the implanted neurostimulator could result in lower average power dissipation and reduced side effects from unneeded stimulation. In this paper we present a closed-loop DBS platform for investigating control strategies for the management of essential tremor. We demonstrate a system capable using a variety of sensors including inertial measurements, electromyography and neurostimulator electrode readings. This sensed data is used to modify stimulation (within limits pre-set by a clinician), thus resulting in a closed-loop system.

Closed-loop DBS with movement intention

In this paper we present a prototype proof-of-concept for a closed-loop deep brain stimulation system for patients with essential tremor. This system makes use of sensed movement intentions via EEG to determine when stimulation is required and automatically enables stimulation only when needed. We demonstrate this system using a healthy subject and a benchtop experimental prototype. By limiting stimulation to only when it is therapeutically required, implanted neurostimulators can be more power efficient and potentially limit the period where patients experience side-effects to only the time when therapy is needed.

Experimental analysis of denial-of-service attacks on teleoperated robotic systems

Applications of robotic systems have had an explosive growth in recent years. In 2008, more than eight million robots were deployed worldwide in factories, battlefields, and medical services. The number and the applications of robotic systems are expected to continue growing, and many future robots will be controlled by distant operators through wired and wireless communication networks. The open and uncontrollable nature of communication media between robots and operators renders these cyber-physical systems vulnerable to a variety of cyber-security threats, many of which cannot be prevented using traditional cryptographic methods. A question thus arises: what if teleoperated robots are attacked, compromised or taken over? In this paper, we systematically analyze cyber-security attacks against Raven II R, an advanced teleoperated robotic surgery system. We classify possible threats, and focus on denial-of-service (DoS) attacks, which cannot be prevented using available cryptographic solutions. Through a series of experiments involving human subjects, we analyze the impact of these attacks on teleoperated procedures. We use the Fitts law as a way of quantifying the impact, and measure the increase in tasks difficulty when under DoS attacks. We then consider possible steps to mitigate the identified DoS attacks, and evaluate the applicability of these solutions for teleoperated robotics. The broader goal of our paper is to raise awareness, and increase understanding of emerging cyber-security threats against teleoperated robotic systems.

Securing the exocortex: A twenty-first century cybernetics challenge

An exocortex is a wearable (or implanted) computer, used to augment a brains biological high-level cognitive processes and inform a users decisions and actions. In this paper, we focus on Brain-Computer Interfaces (BCIs), a special type of exocortex used to interact with the environment via neural signals. BCI use ranges from medical applications and rehabilitation to operation of assistive devices. They can also be used for marketing, gaming, and entertainment, where BCIs are used to provide users with a more personalized experience. BCI-enabled technology carries a great potential to improve and enhance the quality of human lives. This technology, however, is not without risk. In this paper, we address a specific class of privacy issues, brain spyware, shown to be feasible against currently available non-invasive BCIs. We show this attack can be mapped into a communication-theoretic setting. We then show that the problem of preventing it is similar to the problem of information hiding in communications. We address it in an information-theoretic framework. Finally, influenced by Professor Wieners computer ethics work, we propose a set of principles regarding appropriate use of exocortex.

Securing the Exocortex: A Twenty-First Century Cybernetics Challenge

An exocortex is a wearable (or implanted) computer, used to augment a brains biological high-level cognitive processes and inform a users decisions and actions. In this paper, we focus on Brain-Computer Interfaces (BCIs), a special type of exocortex used to interact with the environment via neural signals. BCI use ranges from medical applications and rehabilitation to operation of assistive devices. They can also be used for marketing, gaming, and entertainment, where BCIs are used to provide users with a more personalized experience. BCI-enabled technology carries a great potential to improve and enhance the quality of human lives. This technology, however, is not without risk. In this paper, we address a specific class of privacy issues, brain spyware, shown to be feasible against currently available non-invasive BCIs. We show this attack can be mapped into a communication-theoretic setting. We then show that the problem of preventing it is similar to the problem of information hiding in communications. We address it in an information-theoretic framework. Finally, influenced by Professor Wieners computer ethics work, we propose a set of principles regarding appropriate use of exocortex.

On potential security threats against rescue robotic systems

Remotely operated robots can be used for rescue and recovery in natural disasters and man-made catastrophes, including battlefield environments. But, what if the robot is taken over and turned into a weapon? In this paper, we consider the type of attacks that might occur and their implications on rescue and recovery missions. From this, we introduce a notion of telerobotic security and propose some ideas to ensure that rescue systems are “teleoperation secure” against one likely exploit.

Empirical Movement Models for Brain Computer Interfaces

For brain–computer interfaces (BCIs) which provide the user continuous position control, there is little standardization of performance metrics or evaluative tasks. One candidate metric is Fitts’s law, which has been used to describe aimed movements across a range of computer interfaces, and has recently been applied to BCI tasks. Reviewing selected studies, we identify two basic problems with Fitts’s law: its predictive performance is fragile, and the estimation of ‘information transfer rate’ from the model is unsupported. Our main contribution is the adaptation and validation of an alternative model to Fitts’s law in the BCI context. We show that the Shannon–Welford model outperforms Fitts’s law, showing robust predictive power when target distance and width have disproportionate effects on difficulty. Building on a prior study of the Shannon–Welford model, we show that identified model parameters offer a novel approach to quantitatively assess the role of control-display gain in speed/accuracy performance tradeoffs during brain control.

Personal Responsibility in the Age of User-Controlled Neuroprosthetics

Deep-brain stimulation systems are an accepted and clinically effective form of neuroprosthetic treatment for a variety of common and debilitating neurological movement disorders: Essential Tremor, Parkinsons, and others. Most current implementations of DBS are open-loop: they remain active continuously, whether or not the user is experiencing symptoms. Recent research suggests that it is possible to devise more advanced systems where stimulation is delivered on demand. Their research offers a proof-of-concept for a Brain-Computer Interface-triggered DBS (BCI-DBS) system capable of detecting either signs of tremor or the users neural commands through an additional set of co-implanted sensors. The system then delivers stimulation to meet the users needs or demands. These technologies, however, come with a set of moral problems - in particular, problems for personal responsibility. This paper investigates whether giving users moment-to-moment neural control over their DBS system is ethically responsible given that users can make bad choices and thus harm others. We also ask what responsibilities medical professionals have have to support users as they learn to adapt to neuroprosthetic use. We guide our exploration of these issues through a series of hypothetical scenarios that BCI-DBS users may face.

Model predictive control of deep brain stimulation for Parkinsonian tremor

Deep brain stimulation (DBS) is an established therapy for a variety of neurological disorders, including Parkinsons disease, essential tremor, and dystonia. Recent DBS research has pursued methods for closed-loop control to provide more effective management of symptoms, side effects, and device power consumption. Most closed-loop DBS (CLDBS) studies to date use simple threshold-based controllers to trigger DBS and, as a result, any optimization of symptoms and device power consumption is only incident. In this paper, we demonstrate the utility of an approach based on identifying patient-specific models of symptom response to DBS and using these models to formulate a model predictive control strategy for CLDBS, which explicitly solves an optimization problem. We simulate the model predictive controller for various parameters and find that this approach yields a range of performances for the competing objectives of minimizing patient symptoms and device power consumption. We examine this fundamental tradeoff using the concept of Pareto optimality and conclude with a discussion about incorporating patient, clinician, and other stakeholder preferences in the design of CLDBS systems.

Classifier-based closed-loop deep brain stimulation for essential tremor

Deep brain stimulation (DBS) is a common therapy for the treatment of essential tremor (ET). Currently, this technology continuously delivers stimulation to deep brain regions to mitigate symptoms. Closed-loop DBS aims to deliver stimulation only when symptoms are present, thus improving battery life and decreasing potential side effects. In this study, we used an investigational DBS device implanted with an electrocorticography strip in a subject with essential tremor. Using local field potentials sensed from motor cortex, we built a system of classifiers capable of detecting tremor-inducing movement. These classifiers were incorporated into a closed-loop DBS system which changed stimulation voltage in real time to ameliorate tremor. This is the first time that machine learning has been used in a CLDBS system to detect symptoms and change DBS parameters in real time.