Three people with paralysis of all four limbs used their thoughts to steer a wheelchair through a cluttered room with a reasonably high level of accuracy. This suggests people with paralysis could move independently through certain rooms, but the technology may not be advanced enough to navigate a busy street.
A range of different researchers have previously used two main strategies to test mind-controlled wheelchairs on non-disabled people. The first involves a person focusing on a flickering light in a particular location. This generates brain signals that an artificial intelligence translates into wheelchair movements towards that location, but this approach often leads to eyestrain.
The second strategy involves implanting electrodes in the brain. These accurately transmit brain signals to an AI, but only following a highly invasive procedure that carries a risk of infection.
Testing a third strategy, José Millán at the University of Texas in Austin and his colleagues recruited three people with little or no movement in any of their limbs. The team assessed whether a brain-computer interface could steer an electric wheelchair based on brain activity generated when these individuals imagine moving their limbs.
Each participant wore a skullcap containing 31 electrodes, which could non-invasively detect signals from a brain region that regulates movement, called the sensorimotor cortex. These signals were transmitted to a laptop fixed on the back of the wheelchair, where an AI translated them into wheel movements.
To move right, the participants imagined moving both arms. To move left, they imagined moving both legs. The wheelchair otherwise moved forwards.
In the two other strategies used to steer mind-controlled wheelchairs, the ability to navigate the chair mainly relies on how well brain-computer interfaces retrieve and interpret brain signals from a user over a training-and-testing session that lasts a few hours.
In the latest research, the team trained the participants to generate clearer brain signals over a period of two to five months, with three training sessions per week.
During each session, the team asked the participants to command the wheelchair to move left or right 60 times, on average.
“Person 1” delivered correct commands 37 per cent of the time, on average, across their first 10 training sessions, increasing to 87 per cent accuracy by their final 10 training sessions.
The steering accuracy of “Person 3” also improved, from 67 per cent to 91 per cent. “Person 2” consistently steered with an average accuracy of 68 per cent over their training sessions.
“There will be people who will learn it very fast and very well, then there will be others who will need more time to learn, such as Person 2, but I think anyone can learn to do it,” says Millán.
By analysing the participants’ brain signals over the training period, the team found the “left” and “right” brain signals of Person 1 and Person 3 became more distinct.
Next, the team tested how well the participants could navigate the wheelchair through four checkpoints across a 15-metre hospital room containing beds, chairs and medical equipment.
Person 1 completed the circuit in about 4 minutes with 80 per cent success, on average, over 29 attempts. Success was defined as passing through the circuit checkpoints.
Person 3 completed the circuit in roughly 7 minutes with 20 per cent success, on average, across 11 attempts. Person 2 reached the third checkpoint in about 5 minutes during 75 per cent of their attempts but couldn’t complete the whole course.
“I wouldn’t say the approach is useful on busy streets or less controlled environments, but the ability to move independently at all can be a huge benefit to these people,” says Millán.
However, the skullcap must be stuck to the head via a gel that dries out after a few hours, limiting how long the wheelchair can be controlled at one time.
The use of gels could one day be avoided due to rapid advancements in dry and skin-printed electrodes, as well as ones that fit within the ear, says Palaniappan Ramaswamy at the University of Kent, UK. Combining this latest research with the gel-free technology could push mind-controlled wheelchairs out into the real world in the next decade, he says.
Journal reference: iScience, DOI: doi.org/10.1016/j.isci.2022.105418
More on these topics: