 |
| A graphic abstract of the new approach. (Sorger 2012) |
Patients with “Locked-in Syndrome” suffer an extreme form of paralysis, one that leaves them quadriplegic and unable to speak. The condition can result from a number of factors, ranging from traumatic brain injury to stroke. Each pathology boils down to damage in the brain stem and lower brain that results in a loss of motor control while retaining consciousness and cognitive function. Researchers have been working for the past several decades to develop ways for these patients to communicate through non-verbal means by creating brain-computer interfaces (BCI’s). Much of the focus in the field has been on devices that measure the brain's electrical activity through electroencephalography (EEG) and translate that activity into actions, such as moving a cursor on the screen to select items. While relatively portable and inexpensive, EEG-based BCI’s are known to be rather difficult for a patient to learn how to use, requiring hours of training over multiple sessions. Alternative methods for these patients to communicate are needed and one team of researchers in the Netherlands believe they have such an approach.
A team of neuroscientists headed by Bettina Sorger at Maastricht University have been developing a way to use fMRI to encode and then, through the help of computer software, decode simple messages in real-time. The technology relies on the hemodynamic response, a measure of blood-oxygen concentration that is associated with brain activity, that is measured in fMRI. The technique allows participants to encode twenty-seven unique responses, the twenty-six letters of the English alphabet plus a space for separating words, allowing them to answer questions posed to them by the researchers. While in the MRI machine, participants are presented with the letter that they are to encode and then perform a task that is specific to the desired letter or space. The three tasks include motor imagery, which would entail tracing a geometric shape in your mind; mental calculation, practicing multiplication tables; and an inner monologue, such as reciting a poem. By manipulating the delay and duration of these three tasks, the researchers were able to elicit twenty-seven unique brain responses that they can then later associate with the appropriate character response.
 | |||
| Sorger and her team created the twenty-seven unique responses shown above by altering three conditions: 1.) The task to be performed, e.g. motor imagery, mental calculation, or inner speech. 2.) The onset delay of the task being formed, meaning how long after the letter to be encoded is placed on the screen do the subjects begin to perform the task. 3.) The duration of the task itself, ranging from ten to thirty seconds. (Sorger 2012) |
The whole encoding process usually only takes subjects about an hour, representing a significant reduction in training time from EEG-based BCI’s.
Data from the encoding process are then fed into a computer program that correlates certain hemodynamic responses with the twenty-seven characters. So then later in the experiment when the subject is asked questions, the decoding program can monitor input from the MRI and produce output in the form of the top three most likely letters that follow from the input it has received. The program can predict the correct letter with its first response with an accuracy of 82%, its second response with 95% accuracy, and its third response with up to 100% accuracy. The researchers can view this decoding and use contextual cues from the question to decipher the subject’s response. Below is an example of this decoding process.
 |
| Examples of questions posed to participants and their answers as interpreted first by the computer and then by a member of Sorger's team. (Sorger 2012) |
The current publication on Sorger's approach represents only a proof-of-concept, utilizing healthy volunteers, and the technique is still in its infancy. Sorger has high aspirations for the future application of her team's technique, hoping that it will allow locked-in patients that are unable to reach proficiency with EEG-based techniques be able to communicate with caregivers. Although the technology requires the use of an MRI machine in a clinical setting, the short conversations that it would enable could prove invaluable to individuals with no other way to express themselves to the outside world. Sorger also makes it clear that her team's approach is simply another option in group of ongoing innovations to improve the quality of life for the severely disabled. The technology may also serve as a way to establish consciousness in non-responsive patients and allow them to receive the appropriate care, something previously unattainable to physicians. An obvious drawback to the technique is its reliance on immobile and expensive MRI machines that are only available in hospitals and research centers. A possible circumvention of this limitation may lie in other brain-imaging techniques that similarly rely on monitoring fluctuations of blood-oxygen concentration within the brain. Technologies such as functional near-infrared spectroscopy, or fNIRS for short, are less expensive and more portable than their MRI counterparts. Overall, Sorger's and others attempts to, in a sense, look into the mind of individuals represents a growing movement in neuroscience to try and tease out salient aspects of thought by monitoring the actions of the brain. A tantalizing prospect to anyone who has ever fantasized about tapping into the minds of others and exploring the inner world hidden just beneath our skulls.
To watch this approach in action and get a first-hand account from the researchers, follow the link: http://www.scientificamerican.com/video.cfm?id=brain-scanner-may-help-2012-09-18
Original Research Article:
Bettina Sorger, Joel Reithler, Brigitte Dahmen, Rainer Goebel, A Real-Time fMRI-Based Spelling Device Immediately Enabling Robust
Motor-Independent Communication. Current Biology. Volume 22, Issue 14, 24 July 2012, Pages 1333-1338. (http://www.sciencedirect.com/science/article/pii/S0960982212005751)
No comments:
Post a Comment