Studies have found that 75% of patients struggle to regain their language skills after suffering from a stroke, often facing communication impediments for the rest of their lives. Finding more effective ways for patients to recover the use of their language capabilities is something that Mariachristina Musso, a doctor from the clinic for neurology and neurophysiology at the University Medical Center Freiburg, is dedicated to studying in her research work. In an interview with Morressier, she talks about her innovative study into how brain computer interfaces could help stroke patients significantly improve their language skills and potentially replace the need for speech therapy long-term.
Morressier: Please briefly explain the research you shared at the European Stroke Organisation Conference (ESOC)
Mariachristina Musso: We introduced a novel rehabilitation approach for stroke patients suffering from chronic aphasia, which is a type of language impairment that affects the way people express and understand language, along with how they read and write.
Our approach is based on techniques used for brain-computer interfaces (BCI), which is a system that automatically analyzes brain signals and translates them into control commands or user feedback. In our study, patients had to recognize a specific word in a rapid sequence of words while wearing an EEG cap. After a few seconds, patients received feedback indicating whether there was a difference in EEG signals between the target and other words. This feedback is designed to reflect the cognitive processes that underlie this training task and capture how well the patient executed the task and can then be used by patients to improve their results. One feature of our rehabilitation approach that is really exceptional is that patients do not actually have to speak at all to complete the task.
Morressier: Could you briefly explain what BCI refers to?
Musso: Brain-computer interfaces (BCIs) are systems capable of decoding and monitoring individual brain states in quasi real-time. After training on example data, machine learning algorithms extract information about the ongoing brain state of a subject based on the electrical activity of their brain. This information enables users to interact with physical devices or to control software applications via their brain activity. Various applications already exist for communication, motor rehabilitation, playing games and wheelchair control, to name just a couple of examples.
Morressier: What is the significance of your findings?
Musso: We evaluated our training concept in a pilot study involving 30 hours of high-intensity BCI-based training with eight stroke patients that had mild to moderate levels of chronic aphasia. The results were remarkable: The patients’ language skills significantly improved – even those who we did not train directly. For example, picture naming, reading and writing became easier for the patients post-training. In addition, five patients did not show aphasia on a linguistic test at all anymore.
Morressier: What inspired you to research this topic?
Musso: There is a huge amount of new aphasic patients every year and the number is expected to rise over the next decades. Patients with aphasia spontaneously recover to an extent within the first six months, but show only minimal improvements after this time. In this chronic phase, effect sizes of conventional language therapies are small to moderate at most. About 20% of patients do not completely recover even after language therapy. This is why new therapeutic approaches are urgently needed. We wanted to investigate the idea that examining the ongoing state of a patient’s brain can provide valuable information for aphasia rehabilitation.
The study is based on our decades-long research into stroke recovery. Stroke patients require external support as their internal closed-loop is disrupted by the stroke. In our research, we combine the external feed-back, which otherwise is provided by therapists with an internal signal, and hope to reinstall the internal closed-loop as well. The fact that functions other than the ones we trained also improved indicates that the brain has relearned how to use its remaining internal resources.
Morressier: Are there any ethical considerations that should be taken into account when doing research that involves BCIs?
Musso: A BCI is a very powerful tool that can provide more insights into the mental state of a patient than traditional approaches. If a BCI is part of feedback training you have to be careful to only reinforce brain states that are beneficial to the patient. This is a challenge due to the low quality of recorded brain signals and the enormous number of possible processes that could be reinforced. Reinforcing the wrong brain signals could potentially harm the patient, which is why patients must be closely monitored in the early research phase.
Morressier: Briefly outline your experience at ESOC
Musso: ESOC is an excellent exchange platform for outstanding scientists working with neurological patients. The discussions touch on different disciplines and provide an essential input for evaluating results and for developing hypotheses further. I plan on visiting ESOC again next year.
Image credit: Djneight