New device looks to improve neural prosthetics for paralysis victims

For America's approximately 5,000 people who suffer cervical spinal cord injuries each year, typically resulting in quadriplegia, new brain-computer interfaces are not merely fun and games, they're a means to perhaps someday live a full life again and possibly regain some movement. Standing in the way is the complexity of the in-tissue implants needed to gain more complex control.

A newly designed implant from the California Institute of Technology (Caltech) aims to simplify the process of attaching miniature electrodes to brain neurons via robotic control using MEMS devices, tiny little motors. The Caltech Robotics Burdick group is running the project. The group is led by research engineers Michael Wolf, Joel Burdick, his mentor, Jorge Cham and Edward Branchaud. According to researchers, this is the "first robotic approach to establishing an interface between computers and the brain by positioning electrodes in neural tissue".

The research "could enhance the performance and longevity of emerging neural prosthetics, which allow paralyzed people to operate computers and robots with their minds" according to the researchers. An early prototype of the system has been constructed.

The prototype is currently undergoing testing on non-human primates. According to the researchers, the device "is designed to fit inside a standard laboratory cranial chamber, used for acute experiments in non-human primates, to allow semi-chronic operation. A semi-chronic design has the advantage that the device can be repositioned over a different region with minimal effort and without need for additional surgeries."

The device positions four electrodes to optimize action potentials. Wolf describes the overall design of the device, stating, "Our approach consists of implanting a small robotic device (and accompanying control algorithm) with many individually-motorized electrodes that each autonomously locate, isolate, and track a neuron for long periods of time. To further complicate matters, we wish to find signals only from neurons dedicated ('tuned') to a particular task, say controlling an 'arm reach.' While the primary aim of such technology is for a neural interface for neuroprostheses, such a device may also advance the state-of-the-art experimental techniques for electrophysiology."

While the Caltech team is still working on fine tuning the MEMS design for the final version of the device, the software algorithm is complete. The algorithm in many respects is the keystone of the project. It was actually adapted from algorithms the U.S. military uses to track airplanes. On a most basic level, the algorithm involves the motors slowly being powered to drive the probe down into a tissue. As it picks up a signal it pushes the probe deeper until the signal deepens, in which case it backs up to position itself on the active neuron.

Neuroprosthetics, the science of using brain implants to power robotic limb movement, is a budding field of science, buoyed by recent better understanding of the human brain and new nanoelectronic designs. However longevity is a major concern as cells in the brain can shift slightly and even slight shifts in an in-brain electrode probe could disconnect it.

The advance of medical science is frustratingly slow for the afflicted, but with improvements such as the new Caltech interface, quadriplegics and those suffering from other neurological conditions may someday be able to walk and lead mostly normal lives.

Source from DailyTech

AMD's next generation Radeon graphics card to get a boost from GDDR5

Although NVIDIA may be grabbing headlines lately with leaked details of its next generation GeForce graphics cards, AMD isn't exactly standing still with its Radeon offerings. ATI is preparing its Radeon HD 4800 series GPUs which will replace the existing HD 3800 lineup.

While the HD 4800 Series is rumored to feature GPGPU physics and HDMI 7.1 surround sound pass-through, today we were made privy of one concrete aspect of the new cards: the onboard memory. Qimonda contacted DailyTech earlier this morning with the news that they will supply AMD with GDDR5 memory chips for the reinvigorated Radeon family.

"The days of monolithic mega-chips are gone. Being first to market with GDDR in our next-generation architecture, AMD is able to deliver incredible performance using more cost-effective GPUs," remarked Rick Bergman, AMD Senior Vice President and General Manager, Graphics Product Group. "AMD believes that GDDR5 is the optimal way to drive performance gains while being mindful of power consumption. We’re excited about the potential GDDR5 brings to the table for innovative game development and even more exciting game play."

The high-speed memory chips are 512Mbit and offer bandwidth of up to 4.0Gbps. In preparing for AMD's June launch, mass production of the new GDDR5 chips has already commenced and are shipping in volume.

"We are very proud to supply AMD with GDDR5 volume shipments only six months after first product samples have been delivered," said Robert Feurle, Qimonda AG's Vice President of the DRAM Business Unit. "This is a further milestone in our successful GDDR5 roadmap and underlines our predominant position as innovator and leader in the graphics DRAM market."

"Qimonda's strong GDDR5 roadmap convinced us to choose them as a primary technology partner for our GDDR5 GPU launch," added AMD's Joe Macri. "Both the early availability of first samples and volume shipments added great value to the development and launch of our upcoming high-performance GPU."

Source from DailyTech