San Francisco -- Progress in the microminiaturization of prosthetic devices for the blind and the hard of hearing was reported at the International Solid-State Circuits Conference here last week, as were advances in neural implants.
A Massachusetts team is working on a retinal implant in hopes of restoring vision to those suffering from age-related macular degeneration and retinitis pigmentosa. "The goal is to stimulate the remaining healthy layers of retinal neurons using brief biphasic current pulses," MIT researcher Luke Theogarajan said at the ISSCC.
Researchers from the Massachusetts Institute of Technology and the Massachusetts Eye and Ear Infirmary at Harvard University noted that the design requirements for the implant include an external power source; wireless communication of external commands to the implant; and the ability to provide wireless tuning of pulse amplitude, duration and interpulse timing.
They met source and wireless requirements by using an inductively coupled power-and-data link. The team developed a flexible stimulator-chip architecture that allows for seamless scaling of the number of electrodes. The device can be physically implanted in a minimally invasive approach, the scientists said.
Only the electrode array is placed in the eye, beneath the retina, while the secondary coils and stimulator chip are surgically attached to the eyeball. "Surgical trauma to the eye is greatly minimized by placing the bulky electronics in the more compliant eye socket, rather than the delicate retina," said Theogarajan.
The implant was made from a paralyene-encapsulated flexible polyimide substrate, onto which the chip and electrode array were stud-bump bonded. The stimulator-chip architecture provides frequency-independent operation and compensates for nonidealities due to process, temperature and voltage. The 2.3 x 2.2-mm chip, which can drive 15 electrodes, contains 30,000 transistors in a 0.5-micron technology. It consumes 1.3 milliwatts at a data rate of 100 kbits/second (excluding the current sources).
Meanwhile, researchers from two German companies, design house sci-worx and medical electronics firm IIP Technologies, are working on an epiretinal prosthesis that restores basic vision through electrical nerve stimulation within the eyeballs of those blinded by retinal degeneration. The German design is an array of fully digitally interfaced and programmable stimulation-pad cells for a retinal implant "Currently, a complete retinal-stimulator chip is being fabricated, which includes all global functions and 116 stimulation-pad cells," the IIP Technologies researchers reported. Clinical trials have started.
As for cochlear prostheses, researchers at the University of Michigan see a potential solution to hearing problems in electrode arrays that increase the number of stimulating sites in the ear, so that the arrays can more easily adapt to differing patterns of nerve survival. The arrays use multipolar current shaping to increase pitch perception.
Increasing the number of wire electrodes is precluded by the size of the cochlea, which tapers from a diameter of about 200 mm to about 1 mm over its length. Researcher Pamela Bhatti reported on a thin-film cochlear electrode array that achieves high site density and incorporates on-board circuitry for stimulus generation and position sensing. "The array is designed for use in guinea pig studies but offers the same features and site densities needed for a 128-site, 16-channel human array," said Bhatti.
The 8-mm-long substrate tapers from a width of 500 to 200 microns and supports thirty-two 180-micron-diameter iridium-oxide stimulating sites on 250-micron centers. An eight-lead polymeric cable connects the array to a hermetically sealed electronics package containing a microcontroller along with a wireless interface for power and bidirectional data transfer.
A research team at Stanford University, meanwhile, proposes an integrated silicon implant technology that combines research on cortical electrophysiology, algorithms and circuit design to achieve high levels of prosthetic performance while minimizing power consumption.
"Our proposed implantable prosthesis processor [IPP] consists of four major building blocks: an amplification stage and a variable-resolution analog-to-digital converter array; a digital spike sorter; a maximum-likelihood neural decoder; and a wireless data-and-power transceiver," reported Teresa Meng, a Stanford EE professor, and her research team.
The overall compression factor attained is on the order of 106, translating raw neural data at a rate of 80 Mbits/s down to less than 20 bits/s, indicating the intended movement signals. The IPP's total power budget is limited to 1 mW--less than 2 µW for a multistage amplifier that provides 60-dB gain and anti-aliasing filtering; 1 µW for the variable-resolution A/D; and the rest for digital signal processing and wireless communication. The 100-channel A/D array occupies 2.6 x 1.8 mm in 0.13-micron CMOS technology.
Furthest out in boldness is the development work going on in implantable devices for recording neural patterns. In the past decade, neuroscientists and clinicians have begun to use implantable MEMS multi-electrode arrays to observe the simultaneous activity of many neurons. By observing the action potentials, or "spikes," of multiple neurons in a localized region of the brain, it is possible to gather enough data to predict hand trajectories in real-time during reaching tasks. Recent experiments have shown that it is possible to develop neuroprosthetic devices--machines controlled directly by thoughts--if the activity of multiple neurons can be observed.
Researchers at the University of Utah are developing a wireless, fully implantable neural recording system to facilitate research and neuroprosthetic applications. As reported at the ISSCC, the system is based on the Utah Microelectrode Array (UEA), a 10 x 10 array of platinum-tipped silicon extracellular electrodes. Researchers described the development of a mixed-signal IC that will be flip-chip bonded to the back of the UEA and directly connected to all 100 electrodes. There, it will amplify the neural signals from each electrode, digitize this data and transmit it over an RF link outside the body. Power is to be delivered to a 6-mm coil mounted on the back of the chip using an inductive link.