Deborah S. Won

Deep Brain Stimulation: An Evolving Technology

Deep brain stimulation (DBS) is widely used as a safe and effective medical treatment for certain neurological disorders. It continues to evolve with improving techniques in functional neurosurgery and biomedical device engineering. This paper provides an overview of the enabling science and technology that have allowed DBS to successfully treat certain neurological disorders. It also points toward some of the engineering advances that will enable DBS to yield a more predictable outcome for current indications and to be systematically developed as a treatment for new indications.

Robotic assistance that encourages the generation of stepping rather than fully assisting movements is best for learning to step in spinally contused rats

Robotic devices have been developed to assist body weight-supported treadmill training (BWSTT) in individuals with spinal cord injuries (SCIs) and stroke. Recent findings have raised questions about the effectiveness of robotic training that fully assisted (FA) stepping movements. The purpose of this study was to examine whether assist-as-needed robotic (AAN) training was better than FA movements in rats with incomplete SCI. Electromyography (EMG) electrodes were implanted in the tibialis anterior and medial gastrocnemius hindlimb muscles of 14 adult rats. Afterward, the rats received a severe midthoracic spinal cord contusion and began daily weight-supported treadmill training 1 wk later using a rodent robotic system. During training, assistive forces were applied to the ankle when it strayed from a desired stepping trajectory. The amount of force was proportional to the magnitude of the movement error, and this was multiplied by either a high or low scale factor to implement the FA (n = 7) or AAN algorithms (n = 7), respectively. Thus FA training drove the ankle along the desired trajectory, whereas greater variety in ankle movements occurred during AAN training. After 4 wk of training, locomotor recovery was greater in the AAN group, as demonstrated by the ability to generate steps without assistance, more normal-like kinematic characteristics, and greater EMG activity. The findings suggested that flexible robotic assistance facilitated learning to step after a SCI. These findings support the rationale for the use of AAN robotic training algorithms in human robotic-assisted BWSTT.

An EMG-based system for continuous monitoring of clinical efficacy of Parkinson's disease treatments

Current methods for assessing the efficacy of treatments for Parkinsons disease (PD) rely on physician rated scores. These methods pose three major shortcomings: 1) the subjectivity of the assessments, 2) the lack of precision on the rating scale (6 discrete levels), and 3) the inability to assess symptoms except under very specific conditions and/or for very specific tasks. To address these shortcomings, a portable system was developed to continuously monitor Parkinsonian symptoms with quantitative measures based on electrical signals from muscle activity (EMG). Here, we present the system design and the implementation of methods for system validation. This system was designed to provide continuous measures of tremor, rigidity, and bradykinesia which are related to the neurophysiological source without the need for multiple bulky experimental apparatuses, thus allowing more precise, quantitative indicators of the symptoms which can be measured during practical daily living tasks. This measurement system has the potential to improve the diagnosis of PD as well as the evaluation of PD treatments, which is an important step in the path to improving PD treatments.

A System to Integrate Electrical Stimulation With Robotically Controlled Treadmill Training to Rehabilitate Stepping After Spinal Cord Injury

A functional electrical stimulation (FES) system was engineered to integrate information from a robotically controlled position during stepping in order to time stimulation to continuous gait information in a rodent model of spinal cord injury (SCI). In contrast to conventional FES systems which have a fixed timing pattern relative to gait cycle onset (i.e., toe off/heel off or paw contact/heel strike), this system allows adaptation of stimulation to a robotically controlled position. Rationale for the system design is presented along with bench-test results verifying the timing of the stimulation with respect to hindlimb position. This robotically timed FES system will enable studies investigating the capability of this FES therapy to encourage rehabilitation by way of spinal plasticity.

A multichannel CMOS analog front end IC for neural recordings

A multichannel integrated circuit for processing extracellular neural signals has been designed and manufactured. The analog CMOS IC consists of 17 parallel channels, each comprised of three cascaded stages: bandpass filter with gain, switched capacitor filters, and output buffer with selectable gain. The bandpass filter stage features an opamp with non-inverting resistor feedback and an off-chip capacitor in the feedback pathway to provide gain (43 dB) and one high pass filter pole (220 Hz). The low pass pole is set by the gain-bandwidth product of the opamp. In the switched capacitor filter stage, a one-pole high pass filter (500 Hz) cascades into a two-pole biquadratic low pass filter (5 kHz). The switched capacitor filters may be controlled by either an onboard tunable ring oscillator centered at 50 kHz or an off-chip clock. A four-phase clock splitter provides the necessary filter control-signals; a phase delay of 180/spl deg/ between the high and low pass clock lines maximizes settling time between the filters. The output buffer stage provides selectable gain at 20 dB or 32 dB. The IC was manufactured by AMI using a 0.5 /spl mu/m triple metal double poly process, and measures 4.2 /spl times/ 3.8 mm. The die is designed to be packaged in a flip-chip sub-assembly.

16-channel neural pre-conditioning device

We present the mixed-signal circuit design, layout, implementation techniques, and test data for a 16-channel neural pre-conditioning device that is used to amplify and filter signals acquired from chronically implanted electrodes in an animals brain. Schematics and simulation data for each of the subcircuit macros are presented which include a high gain, continuous time first order bandpass filter pre-amplifier, a cascaded bandpass switch capacitor filter, a selectable gain output buffer, and a voltage controlled oscillator based clock generation circuitry. This device was implemented using AMIs, 0.5 /spl mu/m, double poly, triple level metal, 5 V, CMOS technology. The layout and floorplan, specifications and test data for this device conclude this paper.

Effects of spike sorting error on information content in multi-neuron recordings

In brain-machine interface (BMI) applications, multi-channel recordings are spike sorted before the single-unit spike trains are used in computational analysis to decode the neural response. Nevertheless, questions about the necessity and effectiveness of spike sorting remain. To address one of those questions regarding how sorting error affects the ability to use these neural recordings for BMIs, Shannon information theory was applied to spike trains simulated with random sorting error. Mutual information rate was found to decrease exponentially with spike sorting error, regardless of type, i.e. whether false negative or false positive. Less than 10% error could be tolerated before the information content dropped to half its maximum value with no error. Implications for BMI applications are discussed.

Effect of functional electrical stimulation (FES) combined with robotically assisted treadmill training on the EMG profile

Functional electrical stimulation (FES) is used to assist spinal cord injury patients during walking. However, FES has yet to be shown to have lasting effects on the underlying neurophysiology which lead to long-term rehabilitation. A new approach to FES has been developed by which stimulation is timed to robotically controlled movements in an attempt to promote long-term rehabilitation of walking. This approach was tested in a rodent model of spinal cord injury. Rats who received this FES therapy during a 2-week training period exhibited peak EMG activity during the appropriate phase of the gait cycle; whereas, rats who received stimulation which was randomly timed with respect to their motor activity exhibited no clear pattern in their EMG profile. These results from our newly developed FES system serve as a launching point for many future studies to test and understand the long-term effect of FES on spinal cord rehabilitation.

An analytical comparison of the information in sorted and non-sorted cosine-tuned spike activity

Spike sorting is a technologically expensive component of the signal processing chain required to interpret population spike activity acquired in a neuromotor prosthesis. No systematic analysis of the value of spike sorting has been carried out, and little is known about the effects of spike sorting error on the ability of a brain-machine interface (BMI) to decode intended motor commands. We developed a theoretical framework to examine the effects of spike processing on the information available to a BMI decoder. We computed the mutual information in neural activity in a simplified model of directional cosine tuning to compare the effects of pooling activity from up to four neurons to the effects of sorting with varying amounts of spike error. The results showed that information in a small population of cosine-tuned neurons is maximized when the responses are sorted and there is diverse tuning of units, but information was affected little when pooling units with similar preferred directions. Spike error had adverse effects on information, such that non-sorted population activity had 79-92% of the information in its sorted counterpart for reasonable amounts of detection and sorting error and for units with moderate differences in preferred direction. This quantification of information loss associated with pooling units and with spike detection and sorting error will help to guide the engineering decisions in designing a BMI spike processing system.

Robot-Applied Resistance Augments the Effects of Body Weight–Supported Treadmill Training on Stepping and Synaptic Plasticity in a Rodent Model of Spinal Cord Injury

Background. The application of resistive forces has been used during body weight–supported treadmill training (BWSTT) to improve walking function after spinal cord injury (SCI). Whether this form of training actually augments the effects of BWSTT is not yet known. Objective. To determine if robotic-applied resistance augments the effects of BWSTT using a controlled experimental design in a rodent model of SCI. Methods. Spinally contused rats were treadmill trained using robotic resistance against horizontal (n = 9) or vertical (n = 8) hind limb movements. Hind limb stepping was tested before and after 6 weeks of training. Two control groups, one receiving standard training (ie, without resistance; n = 9) and one untrained (n = 8), were also tested. At the terminal experiment, the spinal cords were prepared for immunohistochemical analysis of synaptophysin. Results. Six weeks of training with horizontal resistance increased step length, whereas training with vertical resistance enhanced step height and movement velocity. None of these changes occurred in the group that received standard (ie, no resistance) training or in the untrained group. Only standard training increased the number of step cycles and shortened cycle period toward normal values. Synaptophysin expression in the ventral horn was highest in rats trained with horizontal resistance and in untrained rats and was positively correlated with step length. Conclusions. Adding robotic-applied resistance to BWSTT produced gains in locomotor function over BWSTT alone. The impact of resistive forces on spinal connections may depend on the nature of the resistive forces and the synaptic milieu that is present after SCI.