Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593

A Detailed Overview of Brain-Computer and

Brain-Machine Interfaces
1 2
Uday S. Yeshi Atharva A. Khode
Electronics and Telecommunications SPIT, Electronics and Telecommunications SPIT,
Andheri Mumbai, India Andheri Mumbai, India

3 4
Shashvat Sangle Surabhi Vishwasrao
Electronics and Telecommunications SPIT, Electronics and Telecommunications SPIT,
Andheri Mumbai, India Andheri Mumbai, India

5
Gautami Salve
Electronics and Telecommunications SPIT,
Andheri Mumbai, India

Abstract:- Brain-Computer Interfaces (BCIs) and Brain- techniques such as electroencephalography (EEG), while
Machine Interfaces (BMIs) represent trans-formative BMIs often involve invasive methods, including
technologies capable of enabling communication and electrocorticography (ECoG) and microelectrode arrays, to
control for individuals with severe disabilities. These achieve high-resolution neural decoding.
systems employ a series of intricate processes, including
signal acquisition, feature extraction, feature translation, The development of BCIs and BMIs is rooted in a multi-
and device output, to translate neural activity into step methodology encompassing signal acquisition, feature
actionable commands. While BCIs predominantly focus extraction, feature translation, and device output. These pro-
on noninvasive applications, BMIs often involve invasive cesses transform raw neural signals into commands that
methods, with preclinical studies on animal models control external devices, creating a closed-loop system where
advancing the un- derstanding of neural decoding. feedback enhances user interaction. Preclinical studies,
Despite their promise, several technical challenges particularly those involving primates, have played a pivotal
remain, including signal reliability, adaptive user role in demonstrating the feasibility of these systems. For
interfaces, feedback mechanisms, and economic instance, implanted electrodes in the motor cortex of monkeys
scalability. Addressing these gaps through have enabled precise control over robotic arms, offering
interdisciplinary research is critical to unlocking the full insights into the real-world potential of BMIs.
potential of BCIs and BMIs for real-world applications.
This paper reviews current methodologies, highlights Despite these advancements, significant technical and
technical limitations, and proposes future directions to prac- tical challenges hinder the widespread adoption of BCIs
enhance the reliability, usability, and accessibility of these and BMIs. Signal noise, system reliability, adaptive control,
groundbreaking technologies. and limited feedback mechanisms are key obstacles that must
be addressed. Furthermore, the high costs and invasive
Keywords:- Brain-Computer Interfaces (BCIs), Brain- nature of certain systems, coupled with the lack of robust
Machine Interfaces (BMIs), Neural Decoding, Signal business models, limit accessibility for many potential users.
Acquisition, Feature Extraction, Device Output, Invasive
Technologies, Non- Invasive Technologies, Technical This paper explores the methodologies underpinning
Challenges, Feedback Mechanisms, Economic Feasibility, BCIs and BMIs, delves into the technical gaps that
Interdisciplinary Research, Real-World Applications. constrain their utility, and discusses emerging strategies to
overcome these barriers. By fostering interdisciplinary
I. INTRODUCTION collaboration and innovation, BCIs and BMIs can evolve
from laboratory prototypes to practical solutions,
Brain-Computer Interfaces (BCIs) and Brain-Machine transforming the lives of individuals with disabilities.
Inter- faces (BMIs) are at the forefront of technological
innovation, offering unprecedented possibilities for restoring II. FUNDAMENTALS OF BRAIN-COMPUTER
communica- tion and motor functions in individuals with
severe disabilities. By leveraging neural activity to control A. Interfaces (BCIs) and Brain-Machine Interfaces (BMIs)
external devices, these systems have shown immense
potential in applications ranging from prosthetic control to  Neural Signal Acquisition:
rehabilitation. BCIs predominantly utilize noninvasive The foundation of both BCIs and BMIs lies in the ability
to acquire neural signals from the brain. These signals

IJISRT24DEC1801 www.ijisrt.com 2202

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593
represent the electrical activity produced by neurons and can  Real-Time Processing:
be measured using different techniques: The system must process the signals rapidly (often in
real- time) to ensure smooth control of devices. This is where
Electroencephalography (EEG): Non-invasive method machine learning and adaptive algorithms come into play,
using electrodes placed on the scalp to measure brain activity. continually adjusting to changes in the brain’s patterns over
While it offers relatively low spatial resolution, it is widely time.
used due to its affordability and non-invasive nature.
 Closed-Loop Control:
Electrocorticography (ECoG): Invasive technique Most BCIs and BMIs function in a closed-loop system,
where electrodes are placed directly on the surface of the meaning the system provides feedback to the user about
brain, providing higher spatial resolution and better signal the action being performed (e.g., visual feedback for cursor
fidelity. movement), which allows the user to refine their control.

Intracortical Micro-electrode Arrays (MEA): These are  Neuroplasticity and Learning:

implanted within the brain tissue to record individual neuron One of the most important concepts in the operation of
activity or small groups of neurons, allowing for precise BMIs, in particular, is neuroplasticity, the brain’s ability to
control in BMIs. reorganize itself and form new neural connections. BMIs
often leverage this plasticity:
Magneto-encephalography (MEG) and Functional
Magnetic Resonance Imaging (fMRI): These are more Adaptation: As a user interacts with a BMI, their brain
advanced tech- niques that offer high-resolution insights into gradually adapts to the new way of controlling devices. This
brain activity, although they are not yet commonly used in adaptation can improve the precision and speed of control
real-time control applications. over time, just as the brain adapts to new physical skills.

 Signal Processing and Feature Extraction:  Motor Learning:

The electrical signals acquired from the brain are raw The brain learns new motor skills through repetitive
and noisy, requiring extensive signal processing before they practice. In the context of BMIs, this learning process
can be used effectively. This processing includes: involves the brain adjusting to the feedback from the BMI
system, which in turn improves control.
Preprocessing: Raw neural signals are filtered to remove
artifacts from external sources like eye blinks or muscle  Feedback Mechanisms:
movements. Feedback is essential to the success of BCIs and BMIs,
as it helps users adjust their brain activity to control devices
Feature Extraction: The next step is to identify relevant more effectively. Feedback can be:
patterns in the brain activity that correspond to specific
thoughts or intentions. Common features include amplitude, Visual Feedback: Most common in BCIs, where users
frequency, and latency of neural oscillations. In BMIs, the see the results of their brain activity (e.g., moving a cursor
focus is on identifying motor-related patterns (e.g., on a screen). However, this type of feedback can be slow and
movement intention). Time Frequency Analysis: Many brains sometimes counterintuitive.
signal change over time and frequency. Techniques like
wavelet transforms or Fourier transforms are used to analyze  Proprioceptive or Tactile Feedback:
these dynamic signals. Especially important in BMIs, where users need to
control a prosthetic or robotic limb. Sensory feedback, such as
Pattern Recognition: Using algorithms like Support vibrations or forces, informs the user about the state of the
Vector Machines (SVM) or Neural Networks, these features device (e.g., whether the prosthetic hand is gripping an
are classified into specific categories, representing distinct object).
user intentions (e.g., move a cursor, open a prosthetic hand).
 Multimodal Feedback:
 Decoding and Feature Translation Combining different types of feedback (e.g., visual,
The key challenge in both BCIs and BMIs is decoding auditory, and tactile) can offer a more intuitive and effective
brain signals into meaningful commands that can control control experience, especially for complex tasks requiring
external devices. This step involves translating extracted fine motor precision.
features into actionable instructions:
 System Adaptability and Personalization:
 Neural Decoding: Both BCIs and BMIs must be adaptable to individual
The brain’s signals, once processed, are mapped to users because:
motor actions or control commands. In BCIs, these might
translate to cursor movements on a screen, whereas in BMIs, Individual Variability: Brain signals vary greatly
they could control robotic arms or prosthetics. between individuals, meaning a one-size-fits-all approach is
not feasible. Customizing the system to each user’s neural
patterns is essential.

IJISRT24DEC1801 www.ijisrt.com 2203

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593
Learning Algorithms: Advanced machine learning tech- Signal Noise and Interference: Neural signals are often
niques, such as neural networks and deep learning, are weak and mixed with noise (from muscle contractions or
increasingly used to create systems that personalize other sources), making decoding challenging.
themselves based on the user’s brain activity. These
algorithms adapt to the user’s evolving brain patterns and Long-Term Reliability of Implants: For invasive BMIs,
improve control over time. ensuring that implanted devices continue to work effectively
over long periods is a major challenge, as the body’s immune
 Challenges and Limitations system may cause inflammation or scar tissue buildup.
Several theoretical and technical barriers hinder the
widespread use of BCIs and BMIs:

III. PROPOSED SYSTEM DESIGN

A. Let us Understand the Function of each Block in our System,

Fig 1 Proposed Block Diagram of Brain Computer Inter-Face BCI

 Brain Signal Source.  Command Translational unit.

When we think action we want to perform our brain It converts the classified output to executable
generates the electrical signals, such as electroencephalogram commands. Basically this commands are transmitted to
(EEG) signals. Such signals are produced due to neural external device via various communication protocols like
activity and are captured from the scalp by using dedicated Bluetooth , wi-fi or USB.
sensors or electrodes. The brain’s electrical activity varies
based on cognitive or motor tasks, which forms the basis of  External Device or Actuator
BCI functionality. This is nothing but the device we have targeted to
control by using overall system. These devices can be
 Signal Acquisition System. anything for ex: Robotic arms , moving cursors etc.
In signal acquisition part we placed the electrodes on the
scalp to collect raw brain signals. These electrodes can be  Feedback Loop
(wet or dry EEG sensors) which picks up very weak signals The feedback loop provides real-time feedback to the
in the range of microvolts (µV). These signals are highly user, which helps user to implement their mental strategies
prone to interference and noise which requires amplification. for improved system performance. This feedback provided by
The signal acquisition block also digitizes the analog signals the system can be vary it can be visual , audio as well as
for further processing. haptic.

 Pre-processing Unit IV. ALGORITHM

This block removes unwanted noise and artifacts from
the raw brain (EEG signals) signals to improve their quality. A. For Brain Computer Task Interface Algorithm can be
Usually it contains Bandpass filters, artifacts removal and Proposed as :
amplification of these signals.
 Data Acquisition
 Featured Extraction Unit. Input given to this part are the signals from the brain
This block identifies and extracts the meaningful which can be EEG, MEG, ECoG. There are certain specific
patterns from the preprocessed signals. This implements steps by which we can acquire this signals.We need to
various meth- ods like FFT, PCA and Wavelet Transform. connect the sensors or electrodes at appropriate locations.
Signals acquired will be weak and in the analog form we Will
 Signal Classification Unit use Analog to digital converter which will convert this
This unit is implemented to classify the clear intend of acquired signals into digital for further used for processing.
the signals by using various complex algorithms like K- Sampling rate used will be varying as per the signals if signals
Nearest Neighbors(KNN) , Recurrent Neural Networks acquired are EEG we need sampling rate of 256 Hz or more.
(RNN). Which declares the clear user intent to target the
controlling device.

IJISRT24DEC1801 www.ijisrt.com 2204

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593
 Preprocessing  Feature Extraction
Usually the signal we acquired is weak with This stage identifies the important patterns and
introduction of noise we need this stage to remove the noise. characteristics of the preprocessed brain signals. This feature
We use filters specially Bandpass filters in this stage to pass can be derived in different domains like Time-domain
the specific bands of frequency a desired . Now this range of features which include information of amplitude signals
frequency will be also varying as per the signals . For ex: 0.5 behavior with respect to time. Frequency domain features in
– 4.0 Hz for EEG. In this stage we also remove the unwanted which features extracted by using method like Fast Fourier
noise such as powerline noise using a notch filter created due Transform(FFT) Which tells about energy distribution and
to eye blinks, muscle movement or other environmental frequency bands. Then there is Time frequency analysis
factors like the Independent Component Analysis(ICA). This which combines the strengths of both domains by methods
stage ensures that extracted data is reliable and accurate. like Short-Time Fourier Transform (STFT).

 Feature Extraction  Classification

This stage identifies the important patterns and This stage maps the selected features to specific brain
characteristics of the preprocessed brain signals. This feature states or required task using advanced machine learning
can be derived in different domains like Time-domain algorithms. Various machine learning algorithms which
features which include information of amplitude signals includes Ramdom forestes , Support vector machines (SVM),
behavior with respect to time. Frequency domain features in K-Nearest Neigh- bors (KNN) are mostly used for
which features extracted by using method like Fast Fourier interpretability and robustness. Also for other tedious tasks
Transform(FFT) Which tells about energy distribution and models like Convolutional Neural Networks (CNNs) are
frequency bands. Then there is Time frequency analysis implemented.
which combines the strengths of both domains by methods
like Short-Time Fourier Transform (STFT). V. EXPERIMENTATION AND
RESULT ANALYSIS
 Classification
This stage maps the selected features to specific brain In this experiment, we explored the concept of Brain-
states or required task using advanced machine learning Computer Interface (BCI) by utilizing MATLAB to simulate
algorithms. Various machine learning algorithms which signal processing and cursor movement. The primary
includes Ramdom forestes , Support vector machines (SVM), objective was to create a signal, process it, and control the
K-Nearest Neigh- bors (KNN) are mostly used for movement of a cursor based on the signal’s values.
interpretability and robustness. Also for other tedious tasks
models like Convolutional Neural Networks (CNNs) are
implemented.

Fig 2 Brain Signals

IJISRT24DEC1801 www.ijisrt.com 2205

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593

Fig 3 Output of Our Proposed System

The experiment began by generating a random signal reliable and responsive interface. The experiment also
using MATLAB’s rand function. This function created a highlighted the potential of using such signal-processing
set of random values which were then fed into the BCI techniques as a foundation for more advanced BCI systems
processing code. The core functionality of the system in real-world applications, such as communication for
involved applying predefined thresholds to the generated individuals with severe disabilities.
signal. Based on the signal’s value exceeding or falling
below these thresholds, the cursor was moved either to the left This experiment serves as an initial step in exploring the
or to the right on the screen. capabilities of BCIs and highlights the importance of signal
processing in enabling real-time control of external devices
The thresholds were designed to detect significant through neural or artificial signals.
variations in the signal, and once these were identified, the
system translated the signal into cursor movements. The The results can be viewed at the following link:
behavior of the cursor was directly linked to the values of the https://drive.google.com/file/d/16IluwJ8BqsXJ4FEb1dvG29Aj
signal, and the thresholds played a critical role in determining Y pQHomvZ/view?usp=sharing
the direction of movement. This setup effectively mimics a
basic BCI system where the signal (which can be thought of VI. CONCLUSION
as neural input) controls an external device (the cursor in this
case). This project highlights the feasibility of integrating
simu- lated EEG signals with computer activity control,
A. Results and Analysis effectively showcasing a basic brain-computer interface
The results from the experiment demonstrated that the (BCI) prototype. By combining sine wave-based EEG signal
cursor successfully moved in the left and right directions simulation with random noise to mimic realistic neural
based on the variations in the generated signal. As the signal activity, the system demonstrated the ability to translate
fluctuated, it consistently triggered the movement of the signal amplitude into actionable commands, such as moving
cursor when crossing the specified thresholds. This confirms a computer mouse cursor.
the feasibility of using signal processing techniques for simple
BCI applications. The implementation of a threshold-based trigger for
cursor movement exemplifies how EEG signal features can be
Further analysis showed that the threshold values utilized for interactive applications. This proof of concept
played a significant role in the accuracy and responsiveness underscores the potential of BCIs in enabling real-time
of the cursor movement. By fine-tuning these thresholds, the system interactions, paving the way for further exploration
system’s performance could be enhanced, providing a more into more sophisticated and practical use cases.

IJISRT24DEC1801 www.ijisrt.com 2206

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593
Future work could focus on enhancing signal processing [8]. J. M. Carmena, M. A. Lebedev, R. E. Crist, et al.,
techniques, incorporating adaptive algorithms for more “Learning to control a brain-machine interface for
accurate control, and expanding the scope of computer reaching and grasping by primates,” PLoS Biology,
activities that can be regulated through EEG signals. This vol. 1, pp. 193–208, 2003.
project serves as a foundation for exploring advanced BCI [9]. J. R. Wolpaw and D. J. McFarland, “Control of a two-
designs and their applications in assistive technology and dimensional movement signal by a noninvasive brain-
human-computer interaction. computer interface in humans,” Proc. Natl. Acad. Sci.
U.S.A., vol. 101, pp. 17849–17854, 2004.
The development of Brain-Computer Interfaces (BCIs) [10]. N. Birbaumer, A. Kubler, N. Ghanayim, et al., “The
and Brain-Machine Interfaces (BMIs) has unlocked new thought translation device (TTD) for completely
possibili- ties in bridging neural activity with external paralyzed patients,” IEEE Trans. Rehabil. Eng., vol.
systems, offering life-changing applications for individuals 8, pp. 190–193, 2000.
with disabilities. Despite the advancements in neural signal [11]. N. Birbaumer, T. Hinterberger, A. Kubler, and N.
acquisition, feature processing, and device control, there Neumann, “The thought- translation device (TTD):
remain hurdles such as signal noise, system adaptability, neurobehavioral mechanisms and clinical outcome,”
long-term usability, and cost barriers. IEEE Trans. Neural Syst. Rehabil. Eng., vol. 11, pp.
120–123, 2003.
Overcoming these challenges will require a concerted [12]. P. R. Kennedy, R. A. Bakay, M. M. Moore, K. Adams,
interdisciplinary effort, integrating innovations in and J. Goldwaithe, “Direct control of a computer from
neuroscience, machine learning, and real-time system design. the human central nervous system,” IEEE Trans.
Enhancing adaptability through personalized algorithms and Rehabil. Eng., vol. 8, pp. 198–202, 2000.
leveraging neuroplasticity can significantly improve user [13]. P. H. Peckham, M. W. Keith, K. L. Kilgore, et al.,
experience and performance. Furthermore, ethical “Efficacy of an implanted neuroprosthesis for
considerations, including pri- vacy protection and equitable restoring hand grasp in tetraplegia: a multicenter
access, must remain a cornerstone of future developments. study,” Arch. Phys. Med. Rehabil., vol. 82, pp. 1380–
1388, 2001.
By fostering collaboration across diverse fields, BCIs [14]. D. R. Kipke, “Implantable neural probe systems for
and BMIs can progress from experimental prototypes to cortical neuropros- theses,” Conf. Proc. IEEE Eng.
accessible, reliable technologies. Achieving this goal will Med. Biol. Soc., vol. 7, pp. 5344–5347, 2004.
not only transform assistive technology but also expand the [15]. J. D. Kralik, D. F. Dimitrov, D. J. Krupa, D. B. Katz,
potential applications of neural interfacing, paving the way D. Cohen, and M. A. Nicolelis, “Techniques for long-
for widespread societal benefit. term multisite neuronal ensemble recordings in
behaving animals,” Methods, vol. 25, pp. 121–150,
REFERENCES 2001.
[16]. P. J. Rousche and R. A. Normann, “Chronic recording
[1]. A. B. Schwartz, X. T. Cui, D. J. Weber, and D. W. capability of the Utah intracortical electrode array in
Moran, “Brain- controlled interfaces: movement cat sensory cortex,” J. Neurosci. Methods, vol. 82, pp.
restoration with neural prosthetics,” Neuron, vol. 52, 1–15, 1998.
pp. 205–220, 2006. [17]. R. J. Vetter, J. C. Williams, J. F. Hetke, E. A.
[2]. E. C. Leuthardt, G. Schalk, D. Moran, and J. G. Nunamaker, and D. R. Kipke, “Chronic neural
Ojemann, “The emerging world of motor recording using silicon-substrate microelectrode
neuroprosthetics: a neurosurgical perspective,” arrays implanted in cerebral cortex,” IEEE Trans.
Neurosurgery, vol. 59, pp. 1–14; discussion 11–14, Biomed. Eng., vol. 51, pp. 896–904, 2004.
2006. [18]. P. R. Kennedy, M. T. Kirby, M. M. Moore, B. King,
[3]. M. A. Lebedev and M. A. Nicolelis, “Brain-machine and A. Mallory, “Computer control using human
interfaces: past, present and future,” Trends Neurosci, intracortical local field potentials,” IEEE Trans.
vol. 29, pp. 536–546, 2006. Neural Syst. Rehabil. Eng., vol. 12, pp. 339–344,
[4]. J. P. Donoghue, A. Nurmikko, M. Black, and L. R. 2004.
Hochberg, “Assistive technology and robotic control [19]. P. R. Kennedy, R. A. Bakay, M. M. Moore, K. Adams,
using motor cortex ensemble-based neural interface and J. Goldwaithe, “Direct control of a computer from
systems in humans with tetraplegia,” J. Physiol, vol. the human central nervous system,” IEEE Trans.
579, pp. 603–611, 2007. Rehabil. Eng., vol. 8, pp. 198–202, 2000.
[5]. M. A. Nicolelis, “Actions from thoughts,” Nature, vol. [20]. P. R. Kennedy and R. A. Bakay, “Restoration of
409, pp. 403–407, 2001. neural output from a paralyzed patient by a direct
[6]. L. R. Hochberg, M. D. Serruya, G. M. Friehs, et al., brain connection,” Neuroreport, vol. 9, pp. 1707–
“Neuronal ensemble control of prosthetic devices by 1711, 1998.
a human with tetraplegia,” Nature, vol. 442, pp. 164– [21]. T. D. B. Nguyen-Vu, H. Chen, A. M. Cassell, R. J.
171, 2006. Andrews, M. Meyyappan, and J. Li, “Vertically
[7]. D. M. Taylor, S. I. Tillery, and A. B. Schwartz, “Direct aligned carbon nanofiber architecture as a
cortical control of 3D neuroprosthetic devices,” multifunctional 3-D neural electrical interface,” IEEE
Science, vol. 296, pp. 1829–1832, 2002. Trans. Biomed. Eng., vol. 54, pp. 1121–1128, 2007.

IJISRT24DEC1801 www.ijisrt.com 2207

Volume 9, Issue 12, December – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14598593
[22]. A. P. Georgopoulos, R. E. Kettner, and A. B.
Schwartz, “Primate motor cortex and free arm
movements to visual targets in three-dimensional
space. II. Coding of the direction of movement by a
neuronal population,” J. Neurosci., vol. 8, pp. 2928–
2937, 1988.
[23]. J. C. Sanchez, J. C. Principe, J. M. Carmena, M. A.
Lebedev, and M. A. Nicolelis, “Simultaneous
prediction of four kinematic variables for a brain-
machine interface using a single recurrent neural
network,” Conf. Proc. IEEE Eng. Med. Biol. Soc.,
vol. 7, pp. 5321–5324, 2004.
[24]. E. E. Fetz and D. V. Finocchio, “Operant conditioning
of specific patterns of neural and muscular activity,”
Science, vol. 174, pp. 431–435, 1971.
[25]. E. E. Fetz, “Real-time control of a robotic arm by
neuronal ensembles,” Nat. Neurosci., vol. 2, pp. 583–
584, 1999.
[26]. J. K. Chapin, K. A. Moxon, R. S. Markowitz, and M.
A. L. Nicolelis, “Real-time control of a robot arm
using simultaneously recorded neurons in the motor
cortex,” Nat. Neurosci., vol. 2, pp. 664–670, 1999.
[27]. J. Wessberg, C. R. Stambaugh, J. D. Kralik, et al.,
“Real-time prediction of hand trajectory by ensembles
of cortical neurons in primates,” Nature, vol. 408, pp.
361–365, 2000.
[28]. M. D. Serruya, N. G. Hatsopoulos, L. Paninski, M. R.
Fellows, and J. P. Donoghue, “Instant neural control
of a movement signal,” Nature, vol. 416, pp. 141–142,
2002.

IJISRT24DEC1801 www.ijisrt.com 2208