New algorithm greatly improves speed and accuracy of thought-controlled computer cursor

When a paralyzed person imagines moving a limb, cells in the part of the brain that controls movement still activate as if trying to make the immobile limb work again. Despite neurological injury or disease that has severed the pathway between brain and muscle, the region where the signals originate remains intact and functional.

In recent years, neuroscientists and neuroengineers working in prosthetics have begun to develop brain-implantable sensors that can measure signals from individual neurons, and after passing those signals through a mathematical decode algorithm, can use them to control computer cursors with thoughts. The work is part of a field known as neural prosthetics.

A team of Stanford researchers have now developed an algorithm, known as ReFIT, that vastly improves the speed and accuracy of neural prosthetics that control computer cursors. The results were published November 18 in the journal Nature Neuroscience in a paper by Krishna Shenoy, a professor of electrical engineering, bioengineering and neurobiology at Stanford, and a team led by research associate Dr. Vikash Gilja and bioengineering doctoral candidate Paul Nuyujukian.

In side-by-side demonstrations with rhesus monkeys, cursors controlled by the ReFIT algorithm doubled the performance of existing systems and approached performance of the real arm. Better yet, more than four years after implantation, the new system is still going strong, while previous systems have seen a steady decline in performance over time.

"These findings could lead to greatly improved prosthetic system performance and robustness in paralyzed people, which we are actively pursuing as part of the FDA Phase-I BrainGate2 clinical trial here at Stanford," said Shenoy.

New algorithm can analyze information from medical images to identify diseased areas of the brain and connections with other regions.

Disorders such as schizophrenia can originate in certain regions of the brain and then spread out to affect connected areas. Identifying these regions of the brain, and how they affect the other areas they communicate with, would allow drug companies to develop better treatments and could ultimately help doctors make a diagnosis. But interpreting the vast amounts of data produced by brain scans to identify these connecting regions has so far proved impossible.

Now, researchers in the Computer Science and Artificial Intelligence Laboratory at MIT have developed an algorithm that can analyze information from medical images to identify diseased areas of the brain and their connections with other regions.

The MIT researchers will present the work next month at the International Conference on Medical Image Computing and Computer Assisted Intervention in Nice, France.

The Superstitious Fund Project

The fund works like this: stock trades are carried out by an Automated Trading System (colloquially, a “robot”), which is a computer program that buys, sells or holds stocks based on a set of specifications encoded into the program’s governing algorithm. The code for Chung’s experiment was written by Jim Hunt, who runs a firm called Trading Gurus, and together with Chung they named it “Sid the Superstitious Robot”. (They also decided to make the source code completely transparent and free to download.)

Like many investment models, Sid is an automated speculator. But whereas other algorithms might take action based, for instance, on a stock’s recent performance or the price of oil, the criterion for this program are lunar phases and the affection and disaffection people have for certain numbers. “I wanted it to operate based on human characteristics,” Chung says.

Sid won’t buy anything on the 13th of the month, and steers clear of buying or selling any stock if its value happens to have a 13 in it. As for lunar phases, Chung explains with a hint of pride that the algorithm finds a new moon to be “good”, whereas a full moon is very, very bad. “The closer the moon is to being full, the more it effects us,” Chung says. So as the full moon approaches, the robot – instead of starting to grow claws and thick brown hair – sells more, as if it is nervous about the moon’s impact on multinational corporations and the decision-making capabilities of senior management. If you’re wondering how this automated yet temperamental trader handles an eclipse, one word: sell.