Signals of brain activity

Scientists, decoding brain signals in real time, have learned how to define the image visible to the human

Signals of brain activity

Using electrodes embedded in the temporal lobe (temporal lobes) of the brain of people with epilepsy, scientists from the University of Washington managed in real time to decode signals associated with the images shown to these people. Scientists used methods and technologies enabled by the signals of the brain to recognize images of faces of different people, of buildings and other objects, the recognition accuracy exceeded 95 percent. But in practice, these technologies will help to communicate with people who were "locked within his brain" as a result of paralysis, attack or other diseases.

The studies involved volunteers who are being treated at the Medical center Harborview in Seattle. Each of these people was subject to epileptic seizures, this medication treatments has not brought any positive results. Therefore, each patient with their consent during brain surgery have been introduced temporary electrodes, with which the doctors tried to determine the exact location of the origin of epileptic seizures.

The temporal lobes are brain areas that perform pre-processing the most part of sensory information and in most cases, within them lie the sources of epilepsy. In addition, violations in these areas will be the sources and other diseases of dementia, Alzheimer's, etc. From this point of view the temporal lobe are among the most vulnerable parts of the brain.

During the experiment, signals from elektrokardiograficheskie (electrocorticographic, ECoG) electrodes, placed at various locations in the temporal lobes, are passed through complex algorithmic processing. With the help of this treatment was subdivided into two main components, "the event-related potentials" (event-related potentials), the values of electric potentials of hundreds of thousands of neurons activated by images, and "broadband spectral change" (broadband spectral changes), the change of energy signals in a wide frequency range.

Volunteers, the brain activity which was controlled by computer, showed a sequence of images that alternated every 400 milliseconds. These were the images of people's faces, buildings and other objects that sometimes alternated with empty screens filled with uniform gray. The software that executes the processing of digitized signals, executes the loop processing at a speed 1000 times per second. In addition, the program automatically selects the combination of signals from electrodes located in different places that best correlates with the image that demonstrates the person at a given time.

Task to be solved by the computer system, was to determine the images of the house, flipped upside down. These images appeared on average in 3 cases and 300 and the system successfully identified all of this, making a mistake, only in 3 percent of cases.

After passing through the preliminary stage of learning is known in advance sequence of images, researchers have begun to demonstrate to patients the sequence in which about one-third of the contents were chosen randomly. And in this case, the system demonstrated 96 percent accuracy, giving the result after 20 milliseconds after the man was shown another picture.

This high accuracy was only achieved with a combination of two different components of the brain signals, referred to just above. This combination has allowed scientists to gain some amount of additional information. "Traditionally, scientists considered the signals from individual neurons," write the researchers, "Our method gives a more complete picture on the level quite large and complex neural networks. This is the only way to determine if a wide-awake man focuses on complex visual objects".

"We have developed hardware technology and algorithmic methods can be used in studies of motor functions, cases and research related to the functioning of memory. In addition, all mathematical methods can be used not only to man but also to every living being with a developed brain."