Algorithmic Progress in Neural Prosthetic Devices for Paralyzed and Amputated Patients

Author:  Gowda Shilpa
Date:  October 2007

A case of swords into ploughshares? Perhaps. This week, researchers at Massachusetts General Hospital (MGH), Massachusetts Institute of Technology (MIT), and Harvard University reported in the cover article of the Journal of Neurophysiology on their progress in the design of algorithms in neural prosthetic devices to restore function in patients who suffer from motor deficits. The central themes – graphical models and statistical signal processing – have their roots in the flurry of research that followed America's victory in World War II. WWII researchers' discoveries brought together these themes with modern neuroscience to form a cohesive mathematical approach to designing neural prosthetic devices. This new type of technology could one day dramatically improve the quality of life of patients with neurological disease, where current medical and surgical therapies have fallen short.

The researchers are trying to help people who have intact cognitive abilities but severe motor impairment. Surprisingly, these people comprise a substantial portion of the population, especially those suffering from the degeneration of motor neurons, such as in patients with ayomyotrophic lateral sclerosis (ALS). More than thirty thousand Americans live with ALS, known more commonly as Lou Gehrig's disease for the former Yankees superstar who suffered this progressive, debilitating condition. According to the ALS Association, around three to five thousand new cases are diagnosed each year alone. Other diseases that could benefit from neural prosthetic devices include spinal cord injury leading to paralysis, brainstem infarcts, muscle degeneration as in muscular dystrophy, or from disease of the neuromuscular junction where are used by nerves to trigger muscle movement.

Lakshminarayan "Ram" Srinivasan is the lead author of the paper. Currently a postdoctoral researcher at MGH in Boston and a medical student in the Harvard-MIT Division of Health Sciences and Technology, Srinivasan's research area brings together basic neuroscience with engineering principles. Along with colleagues Uri Eden (Boston University), Sanjoy Mitter (MIT), and Emery Brown (Harvard/MIT/MGH), he reported on a unifying framework that would allow patients to control their artificial limbs with their own brain activity.

While several groups have already prototyped initial brain-driven interfaces in monkeys and humans, each group developed neural signal processing algorithms customized to their particular application, brain region, and hardware. The new framework transcends these divisions to determine a common structure in neural prosthetic algorithm design. At its core, the framework still reflects a two-stage process of estimation followed by control. First, the user's intentions are estimated from their neural signals. Second, those estimates are used to drive the prosthetic device.

Srinivasan expects that this will not be the final or only solution to the design of prosthetic devices. Moreover, there is still a long and fascinating road of discovery between our current technical abilities and the ultimate clinical treatments. Nevertheless, the researchers hope that along the way, intermediate technologies will still provide meaningful improvements to the most severely affected patients.

Ultimately, research in neural prosthetic device algorithms may very well help research in other areas of neuroscience and medical technology. "My hope for the long term," said Srinivasan, "is that lessons from this current research focus will facilitate emerging efforts with other areas of brain function in health and disease."

Author: Shilpa Gowda

Reviewed by: HoiSee Tsao and Dean Corbaley

Published by: Konrad Sawicki