The Combination of Brain-Computer Interfaces and Artificial Intelligence: Applications and Challenges

The Combination of Brain-Computer Interfaces and Artificial Intelligence: Applications and Challenges

712 Review Article on Medical Artificial Intelligent Research Page 1 of 9 The combination of brain-computer interfaces and artificial intelligence: applications and challenges Xiayin Zhang1#, Ziyue Ma2#, Huaijin Zheng2#, Tongkeng Li2, Kexin Chen2, Xun Wang1, Chenting Liu2, Linxi Xu2, Xiaohang Wu1, Duoru Lin1, Haotian Lin1,3 1State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangzhou, China; 2Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China; 3Center of Precision Medicine, Sun Yat-sen University, Guangzhou, China Contributions: (I) Conception and design: H Lin, X Zhang; (II) Administrative support: H Lin; (III) Provision of study materials or patients: T Li, K Chen, X Wang, C Liu, L Xu, X Wu, D Lin; (VI) Collection and assembly of data: Z Ma, H Zheng, X Zhang; (V) Data analysis and interpretation: Z Ma, H Zheng, X Zhang; (VI) Manuscript writing: All authors; (VII) Final approval of manuscript: All authors. #These authors are considered as co-first authors. Correspondence to: Prof. Haotian Lin. Zhongshan Ophthalmic Center, Sun Yat-sen University, Xian Lie South Road 54#, Guangzhou, China. Email: [email protected]. Abstract: Brain-computer interfaces (BCIs) have shown great prospects as real-time bidirectional links between living brains and actuators. Artificial intelligence (AI), which can advance the analysis and decoding of neural activity, has turbocharged the field of BCIs. Over the past decade, a wide range of BCI applications with AI assistance have emerged. These “smart” BCIs including motor and sensory BCIs have shown notable clinical success, improved the quality of paralyzed patients’ lives, expanded the athletic ability of common people and accelerated the evolution of robots and neurophysiological discoveries. However, despite technological improvements, challenges remain with regard to the long training periods, real-time feedback, and monitoring of BCIs. In this article, the authors review the current state of AI as applied to BCIs and describe advances in BCI applications, their challenges and where they could be headed in the future. Keywords: Brain-computer interface (BCI); artificial intelligence (AI); prosthesis; encoding and decoding; computational neuroscience; machine learning Submitted Sep 11, 2019. Accepted for publication Nov 19, 2019. doi: 10.21037/atm.2019.11.109 View this article at: http://dx.doi.org/10.21037/atm.2019.11.109 Introduction to communicate with the central nervous system, and neural sensory organs can provide a muscle independent With the large explosion in technology, the border between communication channel for people with neurodegenerative humans and machines has begun to narrow. Our spectacular diseases, such as amyotrophic lateral sclerosis, or acquired science fictions describing “mind control” have gradually come true with the help of machines. The frontiers of these brain injuries (3). The history of BCIs is intimately related new techniques are brain-computer interfaces (BCIs) and to the effort of developing new electrophysiological artificial intelligence (AI). Experimental paradigms for BCIs techniques to record extracellular electrical activity, and AI were usually developed and applied independently which is generated by differences in electric potential from each other. However, scientists now prefer to combine carried by ions across the membranes of each neuron (4). BCIs and AI, which makes it possible to efficiently use the The methods of detecting different types of brain brain’s electric signals to maneuver external devices (1). signals can be classified as invasive or noninvasive (5). For severely disabled people, the development of BCIs Invasive recording systems include electrocorticography could be the most important technological breakthrough in (ECoG), microelectrode arrays (MEAs), and so on (6). decades (2). BCIs, which represent technologies designed Noninvasive BCIs including electroencephalography © Annals of Translational Medicine. All rights reserved. Ann Transl Med 2020;8(11):712 | http://dx.doi.org/10.21037/atm.2019.11.109 Page 2 of 9 Zhang et al. BCIs based on AI (EEG), magnetoencephalography, functional magnetic mouse cursors for paralyzed patients through BCIs with resonance imaging (fMRI), and functional near-infrared high feasibility (16,17). The fundamental components of a spectroscopy, do not carry risks of tissue damage and can cursor control BCI include a sensor to record neural signals, be implemented rather easily (7). With the help of these a decoder to interpret movement intentions, and a computer electrophysiological techniques, BCIs can be quickly cursor that interacts with the external environment (18). applied to ‘read’ the brain to record its activity and decode A pioneering study published in 2000 by Kennedy its meaning and to ‘write’ to the brain to manipulate activity and colleagues first demonstrated that an invasive BCI in specific regions and affect their function (8). However, device with a special electrode implanted into the outer the development of BCIs has limitations. Although we have layers of the human neocortex could be decoded to drive a obtained much information from multiple extracellular cursor on a computer monitor (19). Studies in nonhuman electrodes, this large amount of information cannot primates have shown that cursor control BCIs can achieve be transferred efficiently (9). Neuroscientists cannot multidimensional neural integration with two or more unambiguously discern a person’s intentions from the degrees of freedom (20). One-dimensional (1D) cursor background electrical activity recorded in the brain and control is achievable using EEG with event-related match it to the actions of robotic arm (10). The reason for desynchronization, using decision trees sequentially string this limitation is that the neural correlates of psychological selections together to make a final choice (21,22). Two- phenomena are inexact and poorly understood (11). dimensional (2D) cursor control can be achieved using Fortunately, recent advances in AI methodologies have techniques such as fMRI or EEG (23). Recent work in 2017 made great strides, verifying that AI outperforms humans reported the development of a high-performance, invasive in decoding and encoding neural signals (12). This provides BCI for communication, using two algorithms to translate AI a great opportunity to be to an ideal helper in processing signals into point-and-click commands: the ReFIT Kalman signals from the brain before they reach the prostheses. Filter for continuous two-dimensional cursor control and a AI is a set of general approaches that uses a computer Hidden Markov Model-based state classifier for clicking (24). to model intelligent behavior with minimal human By providing at least 2D neural control of the computer intervention, eventually matches and even surpasses cursor and a parallel selection method such as a click, the human performance in task-specific applications (13). user can not only type in self-selected characters but also When AI works within BCIs, internal parameters are use native computer application with the cursors, just like provided to the algorithms constantly, such as pulse a healthy person can with a mouse (25). Moreover, this durations and amplitudes, stimulation frequencies, energy BCI improved the communication rate to 32 letters/min, consumption by the device, stimulation or recording making cursor control more efficient (26). densities, and electrical properties of the neural tissues (14). On the basis of scalp EEG, ECoG and synchronous After receiving the information, AI algorithms can identify evoked potentials, many BCI systems for cursor control useful parts and logic in the data and then simultaneously have been developed, such as the P300 matrix speller and produce the desired functional outcomes (15). Although the rapid serial visual presentation method (27). The Brain these studies remain largely in the preclinical research Gate group conducted the first human trial of a motor BCI starting in June 2004, which recorded signals from a domain, the continued development may highlight clinically Blackrock 96-channel MEA implanted in the arm area of actionable changes in BCIs. M1 in a patient with tetraplegia following cervical spinal At the dawn of technological transformation, a tendency cord injury (28). They achieved two-dimensional movement to combine BCIs and AI has also attracted our attention. of a cursor on a screen and subsequently used this “neural Here, we review current applications with a focus on the cursor” to direct the movement of a robotic limb (28). state of BCIs, the role that AI plays and future directions of BCIs based on AI (Figure 1). Applications in neuroprosthetics and limb rehabilitation Applications of BCIs based on AI The task complexity of BCI studies rapidly has advanced from 2D and 3D control of a cursor on a computer screen (29) Applications in cursor control to the control of more natural behaviors, such as reaching Early studies focused on the control of personal computer and grasping (30), self-feeding (31), and bimanual arm © Annals of Translational Medicine. All rights reserved. Ann Transl Med 2020;8(11):712 | http://dx.doi.org/10.21037/atm.2019.11.109 Annals of Translational Medicine, Vol 8, No 11 June 2020 Page 3 of 9 Auditory sensation Signal processing Cursor control AI Control signal Limb control Spelling device Signal acquisition Rehabilitation Neuroprosthetic Electrode Somatic sensation Visual prosthesis

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    9 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us