<p>The electrical interface to neural medical devices is researched from three perspectives, namely, the electronics within the device, the electrodes on the device, and the electromagnetic fields around the device. </p><p>A Brain-Machine Interface may allow paralyzed patients to control robotic limbs with neural signals sensed by fine wires inserted into the brain. The neural signals have an amplitude under one millivolt and must be amplified. A totally integrated amplifier is designed, manufactured, and characterized. The amplifier is fabricated in a standard half-micron CMOS process without capacitors or resistors. Two application issues not previously addressed are solved. First, the topology of the amplifier is shown to be less sensitive to long-term drift of transistor parameters than the standard topology. Second, a neural signal corrupted by 10 millivolts of powerline interference can be recovered. The amplifier has a gain of 58 dB, a bandwidth of 750 to 14k Hz, power consumption of 180 uW, and noise of 1.5 uV RMS. The design techniques proven in this amplifier are suitable for clinical Brain Machine Interfaces.</p><p>An implanted electroencephalogram (EEG) recorder may aid the diagnosis of infrequent seizure-like events that are currently diagnosed, without proof, as epilepsy. A proof-of-concept study quantifies the electrical characteristics of the electrodes planned for the recorder. The electrodes are implanted in an ovine model for eight weeks. Electrode impedance is less than 800 Ohm throughout the study. A frequency-domain determination of sedation performs similarly for surface versus implanted electrodes throughout the study. The time-domain correlation between an implanted electrode and a surface electrode is almost as high as between two surface electrodes (0.86 versus 0.92). EEG-certified clinicians judge that the implanted electrode quality is at least adequate and that the implanted electrodes provide the same clinical information as surface electrodes except for a noticeable amplitude difference. No significant issues are found that invalidate the concept of an implanted EEG recorder.</p><p>Transcranial stimulation may treat a multitude of neural and psychological illnesses. The stimulation may have higher repeatability and lower patient effort if an implanted device provides the stimulation. The shape of the device, 300 mm long by 1 mm in diameter, is unlike any present implanted device. Five techniques that deliver energy to the device are analyzed using computer simulations. The electrode for the techniques that employ an electric field to deliver the energy is a new design that exploits the anatomy of the scalp and skull. The electric field techniques deliver energy that is likely suitable for some stimulation protocols but not for all. The techniques that employ a magnetic field deliver more than the energy required, especially if the shape of the coil that creates the magnetic field is automatically optimized. However, the magnetic-field techniques heat the brain; the electric-field techniques do not heat the brain. This research validates the new delivery concepts and justifies future research.</p> / Dissertation
| Identifer | oai:union.ndltd.org:DUKE/oai:dukespace.lib.duke.edu:10161/13426 |
| Date | January 2016 |
| Creators | Jochum, Thomas |
| Contributors | Wolf, Patrick D |
| Source Sets | Duke University |
| Detected Language | English |
| Type | Dissertation |
Page generated in 0.0016 seconds