Locked-in syndrome is a condition in which a patient is consciously aware, yet cannot move most voluntary muscle groups. In some cases, these patients are unable to communicate leading to a poor quality of life. Other potentially less inhibiting conditions such as quadriplegia or amyotrophic lateral sclerosis (ALS) still leave patients with the inability to enjoy the freedom they once had. Quadriplegia, typically the result of injury to the spinal cord, leaves patients with partial or total loss of control to all four limbs and the torso. ALS is a neuro degenerative disease that causes loss of control and atrophy of voluntary muscles. The disease progresses until the patient can no longer breathe without the use of a ventilator.
Research in prosthesis has led to the development of many devices aimed at improving the quality of life for many individuals that suffer from the above irreversible conditions. The first advanced computerized speech generation software, called Equalizer, was developed by Walter Woltosz for theoretical physicist Stephen Hawking in the mid 80’s. Hawking operates the device through a tedious process of choosing among a rolling list of options by clicking a single button with his hand. The software has allowed Hawking to continue making the contributions to theoretical physics that has made him famous. But many patients are not lucky to have even limited use of their hands. Research has turned to the development of assistive devices that communicate directly with the brain, bypassing faulty or degenerating neural systems.
Regions of the whole brain are constantly buzzing with activity while they do everything from regulating heartbeat to performing abstract mathematical calculations. In order to translate brain activity into movement, assistive devices use an algorithm that “filters” out the parts of the signals necessary to understand intended motor movement. One such algorithm is called the Kalman filter (KF). Originally developed for spacecraft navigation, the filter is particularly useful to model system states that can only be indirectly observed by the system itself. Kalman filters are the most promising method of decoding signals for motor movement because they minimize error better than any other type of noise filter.
Currently, the best method of recording brain signals uses and array of electrodes surgically implanted into the brain. Due to the risks involved with experimental brain implantation, much of the research has been done on monkeys trained to complete a specific task in anticipation of a reward. In the case of prosthetic training, monkeys are first trained to move a hand to a specific location indicated on a screen while their neural activity is recorded. Information about the task along with neural activity is integrated to estimate the intentions for movement represented by the neural activity. Although many helpful prosthetic devices use this method, they do a very poor job of replicating natural movement. The reason for this is that very little is understood about exactly how the brain commands muscles to move.
Up until now, the most successful algorithms attempted only to understand the intended velocity of the motor movement (Velocity-KF). However researchers have recently shown that the actual location of the moving object, such as a cursor, is very important in estimating intended action. Vikash Gilja et al. describe a new type of kalmin filter called ReFIT-KF, which uses a two-step training method. Because real-time visual feedback of the moving object's location is very important for successful natural motor movement, the authors added this information to the Velocity-KF algorithm. Before performing the above training task, the monkeys watched cursors that autonomously moved toward a series of targets while their neural signals were recorded. The information from this first training session was thought to represent the location of the cursor as it moved toward the target. The results obtained after integrating the cursor's actual location are very promising. For ReFIT-KF, not only were the recorded cursor movements straighter, targets were successfully acquired in half the time as Velocity-KF.
The prosthetic benefits provided by intracortical implants are not without risks. Researchers are still evaluating the long-term effects of placing electrodes directly into brain tissue. Even if the devices are shown to be safe, patients still are under risk of common surgical complications such as infection or adverse reactions to anesthesia. As our understanding of brain signaling improves, electroencephalography (EEG), might provide the information suitable to prosthesis. EEG uses a series of tiny electrodes that rest on the top of the head to read different patters of brain activity. Currently, while intracortical arrays can be implanted to measure activity deep within the brain, the detail provided by EEG is limited by surface placement of the electrodes. The skull the separates the brain and EEG electrodes causes dispersion of the signals, making them very difficult to decode. But non-invasive methods such as EEG appear to be just around the corner. The U.S. Army recently invested $6.3 million into EEG research with the intent of developing a device for communication through the imagining of words, also known as synthetic telepathy. Gerwin Shalk of the Wadsworth Center found that it is possible to discriminate the vowels and consonants of speech through the use of a modified EEG in which electrodes are placed directly on the surface of the brain.
As we learn more about how different regions of the brain talk to one another, it is likely that scientists will develop algorithms that correct for signal distortion caused by the skull, allowing for the use of non-invasive EEG. Not only could this enhance the functioning of prosthesis, it could make the whole processes a lot smoother for the patient.