Page 53 - JSOM Spring 2025

a patient with tetraplegia and severe dysarthria due to amyo- or closed-loop, interfaces enable the patient to control and
trophic lateral sclerosis (ALS) received BCI implants connected modulate neural functions that normally are not under their
22
to computers. A software program displayed the words that volitional control. 20,24,29,43,45,46 By actively responding to BCI-
the patient was thinking onto a screen, then vocalized them us- based feedback, the patient can stimulate neural plasticity,
ing text-to-speech technology that mimicked the patient’s voice enabling, for example, brain regions to adopt functions that
22
before he developed dysarthria. In other clinical experiments, were previously carried out by injured areas, thereby facilitat-
BCIs have enabled patients with tetraplegia and patients with ing neurorehabilitation. 17,19,26,32,45
paraplegia to control devices such as light and other domestic
23
appliances, heating appliances, and television. As these exam- Future Relevance for Injured SOF Personnel
ples suggest, a key goal of medical BCIs is to empower patients
with significant motor impairments and neurodegenerative dis- Of particular interest to SOF medical personnel and injured
orders to lead more independent and self-determined lives. servicemembers is the potential of BCI technology to address
disabilities and conditions previously considered untreatable.
24
Scientists have developed various ways to classify BCIs. One Although successes have been limited to a few cases, they
of the most straightforward distinctions is based on the brain demonstrate proof of concept, and ongoing efforts are focused
signal acquisition mode, categorizing BCIs into noninvasive on achieving further advancements. A notable example was
and invasive. 25–27 Noninvasive BCIs are worn as caps, head- delivered in 2023, when a team successfully restored commu-
sets, and helmets, while surgically implanted BCIs are directly nication between the brain and spinal cord using a wireless
47
connected to the central nervous system, either epidurally, sub- BCI system. The 38-year-old patient with tetraplegia received
18
durally, or intracortically. Currently, invasive methods are two ECoG implants, in addition to a spinal implant. These im-
primarily used in clinical settings, particularly for patients with plants bypassed the damaged spinal cord region and provided
severe movement and communication impairments. Implants electrical stimulation to activate the relevant muscles. This
28
47
offer superior spatial and temporal resolution, along with im- allowed the person to stand, walk, and climb stairs. While
proved signal quality, compared to noninvasive options, en- challenges remain—such as some temporal mismatch between
abling higher accuracy in interpreting neural activity. 18 the patient’s intention and corresponding movements—the
study illustrates the potential for some spinal cord injuries to
Both invasive and noninvasive BCIs rely on different neuro- become less debilitating in the future.
imaging techniques to capture brain activity. 17,19,24,29 Wearable
BCIs commonly employ electroencephalography (EEG) and Furthermore, BCI-based smart prosthetics hold the potential
record brain activity through electrodes on the scalp. 24,27,30,31 for enabling more SOF servicemembers to return to active
Intracortical BCIs tend to rely on techniques such as intracor- duty after sustaining loss of a limb. Researchers have success-
tical electrophysiology by way of single electrodes or electrode fully decoded brain signals related to the speed, velocity, and
arrays, while electrocorticography (ECoG) typically records position of arm movements, highlighting the promise of neu-
brain signals from the surface of the cortex. 18,24,27,31–33 Because roprosthetics to support complex motor functions. 35,48 More-
different brain signals can be correlated with specific cogni- over, in 2021 scientists tested a closed-loop BCI system with
49
tive, physical, or sensory behaviors, it is possible to identify the a robotic arm. This system not only allowed the patient to
BCI user’s intention. 34,35 move the prosthetic through thought but also enabled them to
49
receive tactile feedback from the robotic arm. To achieve this,
During signal recording, unwanted signals, known as artifacts, the BCI implants stimulated specific sites within the somato-
can contaminate the desired cerebral signals, from which fea- sensory cortex. By integrating sensory feedback, the system
25
tures are extracted. These interferences originate from both enabled more precise and intuitive control of the prosthetic
27
the external environment and the body itself. Common en- limb. Since 2021, other studies have successfully integrated
dogenous artifacts include eye blinks, muscle twitches, and tactile feedback to enhance the control of prosthetic limbs. 50,51
cardiac rhythms, while exogenous artifacts can originate from Notably, such tactile feedback has the potential to transform
36
power line interference and equipment noise. Body move- prosthetics from purely functional tools into seamless exten-
ments, in particular, can generate complex artifacts. Algorithms sions of the individual’s body.
are increasingly used to filter them out. 18,24,37,38
Researchers have identified common injury and medical issue
BCIs rely on algorithms for all three key processes: acquiring patterns in SOF that include traumatic brain injuries (TBIs),
signals, extracting features, and translating the extracted fea- chronic joint and back pain, headaches, cognitive impairments
tures into output commands. 38–41 Recent advances in artificial such as memory and attention difficulties, depression, and
intelligence, including machine learning and deep learning, posttraumatic stress disorder (PTSD). 52,53 Given the current
have enabled significant improvement of these processes. 24,41–43 challenges in accurately diagnosing TBIs, BCI technology is
Future progress in BCI technology will depend heavily on in- emerging as a promising tool, not only for assessing the extent
terdisciplinary collaboration between neuroengineers—spe- of injury and aiding in diagnosis but also for therapy. 45,54,55
cializing in signal processing, machine learning, and neural Moreover, proof-of-concept research has demonstrated that
networks—and (computational) neuroscientists versed in EEG closed-loop BCIs can improve attention, while another proof
applications, other imagining techniques, and neural data of concept supports the application of BCI technology to
analysis. 19,25,28,42–44 Ultimately, this will enable real-time and restore memory functions. 8,56,57 BCIs have also been used to
accurate exchange between neurophysiologic processes and modulate and manage chronic pain and assist in treating mood
computer systems. and cognitive disorders, including depression and PTSD. 31,58–60
Generally speaking, by enabling precise modulation of brain
In addition to controlling technological devices, BCIs can also activity, BCIs can facilitate targeted interventions that promote
provide neurofeedback to their users. 18,43,45 These bidirectional, rehabilitative neural changes. 43,58

The Present and Future of BCI Technology | 51