Professional Documents
Culture Documents
Submitted in the partial fulfilment of the requirements for the award of the degree of
BACHELOR OF TECHNOLOGY
IN
Submitted By
K. VANITHA
H.T. No:18701A04F8
Department of ECE
Submitted to
DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING
ANNAMACHARYA INSTITUTE OF TECHNOLOGY AND SCIENCES::RAJAMPET
(AN AUTONOMOUS INSTITUTION)
(Approved by AICTE, New Delhi, Accredited By NAAC, Bangalore) (Affiliated To J.N.T.U.A,
Anantapuram)
2021-2022
CERTIFICATE
This is to certify that the Technical Seminar Report entitled “BRAINGATE TECHNOLOGY” is a bonafide
record submitted by K.VANITHA bearing H.T.
No:1870A04F8 in partial fulfillment for the award of Degree of Bachelor of Technology in
“ELECTRONICS & COMMUNICATION ENGINEERING” for the year 2021-2022.
1. Introduction 1-2
2. Neuroprosthetic device 8
3.1 Chip 8
3.2 Connectror 8
3.3 Converter 9
3.4 Computer 10
4.1 Detection 10
4.2 Controll 11
4.3 Training 11
6. Preprocessing 13
9. Advantages 17
10. Disadvantages 18
13. Conclusion 23
Thousands of people around the world suffer from 100 hair-thin electrodes that sense the electro-
paralysis, rendering them dependent on others to magnetic signature of neurons firing in specific areas
perform even the most basic tasks. But that could of the brain.
system brain-computer interface is to enable those systems underlying core technology in the neuroport
with severe paralysis and other neurological system to enable improved diagnosis and treatment
conditions to live more productively and for a number of neurological conditions, such
independently. The computer chip, which is as epilepsy and brain trauma. Braingate will be the
implanted into the brain, monitors brain activity in first human device that has been designed to record,
the patient and converts the intention of the user into filter, and amplify multiple channels of
simultaneously recorded neural activity at a very This can be achieved either by reducing the
high spatial and temporal resolution. diameter of the cylindrical scanner down to
approximately 350 mm [9], [10], [11], [12], [13] or
Recent advances in radiochemistry have by extending the axial field of view (FOV) [14], or
boosted renewed progress in dedicated brain PET by designing compact geometries adapted to the
instrumentation with improved sensitivity [6], [7]. In shape of the brain [15], [16], [17], [18]. It was shown
fact, though PET has high specificity and sensitivity in simulations that the sensitivity of a full helmet-
compared with other imaging modalities, the signal shaped scanner is about 4.2 times higher than that of
to noise ratio (SNR) of PET image is still low due to a classical cylindrical brain PET [19].
the limited resolution and low coincident-photon
counts detected during a given scan time. The paradigm of current developments in
brain PET is that the compact design, often with
The low SNR also limits the achievable additional scintillator layers to stabilize the parallax
temporal resolution in dynamic PET studies, because error, requires an high density of readout channels.
one has either to use long frame durations or to apply The design of fast readout electronics, capable of
temporal smoothing to reduce noise, which makes it handling this complex configuration, is thus an
hard to capture fast dynamic changes. In other important issue in the development of compact brain
applications, such as the imaging of multiple PET scanners [20], [21].
sclerosis using TSPO tracer [11C]PK11195, the SNR
is poor also due to low brain penetration and high A large number of multi-channel application
non-specific binding of the tracer. specific integrated circuits (ASICs) for PET were
developed, providing different solutions for the
Higher sensitivity allows a reduction of the analogue-to-digital conversion of the energy and
radiation dose so that longitudinal studies, such as arrival time of the detected gamma photons and for
those employed in Alzheimer’s disease early the transmission to the data acquisition backend.
prediction, become more feasible, making the Several techniques for the calculation of the energy
technique potentially acceptable for the screening of were proposed, such as pulse peak conversion [22],
vulnerable patients [8]. A straightforward way to [23], current integration [24], non-linear [25] and
improve the sensitivity of dedicated brain PET linear [26] time-overthreshold techniques.
scanners is to increase the solid angle coverage.
Time stamping was performed with simple or such as energy and timing pick up [32], [33], [34],
dual thresholds [27], constant fraction discrimination [35].
[28] and more advanced pattern matching [29]. By
way of example, the TRIROC ASIC is used in the For example, in a proposed brain PET
TRIMAGE brain PET and provides a high-gain and prototype [36] the amplified signals are sent to a
low-gain path for the time stamping and charge-to- position decoder circuit to reduce the number of
digital conversion of PET signals, with singles readout channels by a factor of 256 and three FPGA-
sorting and coincidence detection by time stamp embedded data acquisition (DAQ) boards are used
comparison performed on a field programmable gate for the digitization of analog output signals. In the
array (FPGA) in the backend acquisition system recent MINDVIEW brain PET system a readout per
[30], [2]. block consists of custom ADC electronic boards with
12 bit precision, capable of processing up to 66
In another recent concept of brain PET [31], a channels.
single-event 64bit data includes the crystal segment
and depth of interaction (DOI) addresses, the energy A main trigger board with programmable
and time stamp, and a valid flag indicating coincidence windows (3, 5, 7 and 9 ns) manages the
”acceptable: within the range of the set energy detectors whose signals are digitized. The number of
window” or ”not acceptable out of the range of the readout channels is reduced using a charge division
set energy window”. On-the-fly coincidence network providing information for each row and
detection is performed by the software after each column output, which are further digitized [13]. In
single-event data set is sorted in time order. In [12] recent years we proposed a multi voltage threshold
an improved 48-channel ASIC for brain PET is (MVT) sampling method that takes samples of a
proposed, with a suppressed noise level to achieve pulse with respect to a set of reference voltages.
lower discrimination energy at 60 keV.
We have shown that for PET detectors it is simultaneously. The system also decodes these
sufficient to use only a few reference voltages to signals in real time to control a computer or other
obtain a small number of samples. As the pulses are external devices. The brain gate technology platform
instantaneously digitized and the number of samples was designed to take advantage of the fact that many
per pulse is small, this sampling method can have a patients with motor impairment have an intact brain
high count-rate capability without requiring a large that can produce movement commands allowing the
on-board storage buffer [37], [38]. Due to its high brain gate system to create an output signal directly
modularity, the digital PET technology based on from the brain, bypassing the route through the
MVT readout was given the name of Plug&Imaging nerves to the muscles that cannot be used in
(P&I) sensor technology [39]. paralyzed people.
Dendrites: Signals sent through dendrites neurons. And interneurons tie everything together by
cause chemical changes that result in an connecting the various neurons within the brain and
electrical signal in the cell body. spinal cord. The part of the brain that controls motor
Axons: Nerve impulses are carried through skills is located at the ear of the frontal lobe.
axons away from the neurons cell body.
How does this communication happen?
Neuron muscular junction: The signal is
Muscles in the body's limbs contain embedded
passed by neuron transmitters from synaptic
sensors called muscle spindles that measure the
bulbs on the neurons to muscle fibers. The
length and speed of the muscles as they stretch and
muscle fibers then react to the signal.
contract as you move. Other sensors in the skin
respond to stretching and pressure. Even if paralysis
or disease damages the part of the brain that
1.1 How does the brain control motor
processes movement, the brain still makes neural
function?
signals. They're just not being sent to the arms, hands
which are made by billions of neurons that make A technique called neurofeedback uses
electricity whenever they are stimulated. The
connecting sensors on the scalp to translate brain
electrical patterns are called brain waves. Neurons waves into information a person can learn from. The
act like the wires and gates in a computer, gathering
sensors register different frequencies of the signals
and transmitting electrochemical signals over produced in the brain. These changes in brain wave
distances as far as several feet. to new circumstances.
patterns indicate whether someone is concentrating
The brain encodes information not by relying on or suppressing his impulses, or whether he is relaxed
single neurons, but by spreading it across large
or tense. Even if paralysis or disease damages the
populations of neurons, and by rapidly adapting part of the brain that processes movement, the brain
Motor neurons carry signals from the central still makes neural signals. They're just not being sent
nervous system to the muscles, skin and glands of the to the arms, hands and legs.
hooked up to wires that travel to a pedestal on the is transmitted through the pedestal plug
scalp. From there, a fiber optic cable carries the brain attached to the skull.
activity data to a nearby computer. 3.3 The converter: The signal travels to an
amplifier where it is converted to optical data
The sensors register different frequencies of
and bounced by fiber optic cable to a
the signals produced in the brain. These changes in
computer.
brain wave patterns indicate whether someone is
3.4 The computer: Brain gate learns to associate
concentrating or suppressing his impulses, or
patterns of brain activity with particular
whether he is relaxed or tense.
imagined movements up, down, left, right
The brain encodes information not by and to connect those movements to a cursor.
relying on single neurons, but by spreading it across
large populations of neurons, and by rapidly adapting
. Receptors sense things like chemicals, light, and
sound and encode this information into
electrochemical signals transmitted by the sensory
neurons
WORKING
the silicon chip implanted into the person’s computer cursor, electrodes on the silicon chip
brain detect neural activity. His cortical implanted into the person’s brain detect neural
neurons fire in a distinctive pattern, the signal activity from an array of neural impulses in the brains
motor cortex. The impulses transfer from the chip to
a pedestal protruding from the scalp through through thin gold wires to a titanium pedestal that
connection wires. protrudes about an inch above the patient’s scalp.
The pedestal filters out unwanted signals or An external cable connects the pedestal to
noise and then transfers the signal to an amplifier. computers, signal processors, and monitors. The
The signal is captured by acquisition system and is technology is able to sense the electrical activity of
sent through a fiber optic cable to a computer. The many individual neurons at one time the data is
computer then translates the signal into an action, transmitted from the neurons in the brain to
causing the cursor to move. computers where it is analyzed and the thoughts are
used to control an external device.
The braingate system is a neuromotor
prosthetic device consisting of an array of one 4) PRINCIPLE
hundred silicon microelectrodes; each electrode is
1mm long and thinner than a human hair. The 4.1) DETECTION
electrodes are arranged less than half a millimeter
apart in the array, which is attached to a 13cm-long
The detection of the input from the user and
cable ribbon cable connecting it to a computer.
them translating it into an action could be considered
as key part of any BCI system. This detection means
The BrainGate neural interface system is a to try to find out these mental tasks from the EEG
proprietary, investigational Brain-Computer signal. It can be done in time-domain, e.g. by
Interface (BCI) that consists of an internal sensor to comparing amplitudes of the EEG and in frequency-
detect brain cell activity and external processors that domain. This involves usually digital signal
convert these brain signals into a computer-mediated processing for sampling and band pass filtering the
output under the person’s own control.The sensor is signal, then calculating these time -or frequency
implanted on the surface of the area of the brain domain features and then classifying them.
responsible for voluntary movement, the motor
cortex. The electrodes penetrate about 1mm into the
These classification algorithms include
surface of the brain where they pick up electrical
simple comparison of amplitudes linear and
signals known as neural spiking, the language of the
nonlinear equations and artificial neural networks.
brain from nearby neurons and transmit them
By constant feedback from user to the system and
4.2) CONTROLL
4.3) TRAINING
The training is the part where the user adapts Fig:4.Multichannel Acquisition
to the BCI system. This training begins with very
simple exercises where the user is familiarized with Our brains are filled with neurons, individual
mental activity which is used to relay the information nerve cells connected to one another by dendrites and
to the computer. This training begins with very axons. Every time we think, move, feel or remember
simple exercises where the user is familiarized with something, our neurons are at work. That work is
mental activity which is used to relay the information carried out by small electric signals that zip from
to the computer. neuron to neuron as fast as 250 mph. The signals are
generated by differences in electric potential carried
Bio-Feedback: The definition of the biofeedback is by ions on the membrane of each neuron.
biological information which is returned to the
source that created it, so that source can understand Although, the paths, signals take are insulated
it and have control over it. This biofeedback in BCI by something called myelin, some of the electric
systems is usually provided visually, e.g. the user signal escapes. Scientists can detect those signals.
sees cursor moving up or down or letter being
selected from the alphabet.
Brain Gate Technology Page 11
BRAINGATE TECHNOLOGY
. They could rig a camera that would send those what signals are sent to the brain by the optic nerve
exactly when someone sees the color red. They could rig a
camera that would send those exact signals into
someone’s brain whenever the camera saw red,
allowing a blind person to “see” without eyes. They
could rig a camera that would send those exact
signals into someone’s brain whenever the camera
saw red, allowing a blind person to “see” without
eyes.
6) PREPROCESSING
This approach has many problems, however. It
requires invasive surgery to implant the electrodes, The raw EEG signal requires some
and devices left in the brain long-term tend to cause preprocessing before the feature extraction. This
the formation of scar tissue in the gray matter. This preprocessing includes removing unnecessary
scar tissue ultimately blocks signals. frequency bands, averaging the current brain activity
level, transforming the measured scalp potentials to
cortex potentials and denoising. Frequency bands of The results showed that a human with
the EEG : tetraplegia was able to control a cursor on a computer
screen just by thinking, enabling him to open emails,
.
and to operate devices such as a television[7].
Band Frequen- Amplit- Location
cy [Hz] -ude
One participant, Matt Nagle, had a spinal
[_V]
cord injury, whilst another had advanced ALS. In
Alpha 8-12 10 -150 Occipital/
July 2009, a second clinical trial (dubbed
(_) Parietal
"BrainGate2") was initiated. In May 2012, Brain
regions
Gate researchers published a study in
µ- 9-11 varies Precentral/
Nature[8]demonstrating that two people paralyzed
rhythm Postcentra
by brainstem stroke several years earlier were able to
l regions
control robotic arms for reaching and grasping.Since
Beta 14 -30 25 typically
2009, clinical trials are being conducted under the
(_) frontal
name "BrainGate2 Neural Interface System” .
regions
Theta 4-7 varies varies
As of October 14, 2014, Stanford University,
(_)
Massachusetts General Hospital, Case Western
Delta
<3 varies varies Reserve University (Ohio) and Providence VA
(_)
Medical Center are actively recruiting participants
for the ongoing BrainGate2 clinical trial
Interface System
He was ultimately able to open and close the
The first patient, Matthew Nagle, a 25-year- fingers of a prosthetic hand and use a robotic limb to
old Massachusetts man with a severe spinal cord grasp and move objects.
injury, has been paralyzed from the neck down since
2001. Nagle is unable to move his arms and legs after PERFORMANCE EVALUTION
he was stabbed in the neck. During 57 sessions, at
New England Sinai Hospital and Rehabilitation
Because of the lack of specific NEMA
Center, Nagle learned to open simulated e-mail, draw
procedures for dedicated brain systems, following
circular shapes using a paint program on the
[41], NEMA standards for small animal and clinical
computer and play a simple videogame, "neural
PET systems have been adapted case by case [42],
Pong," using only his thoughts.
[43].
The mean and the variance σ of the 511 keV We placed one face of the cylinder at the
photoelectric peak were used for the calibration of beginning of the axial FOV. We calculated the
each channel. After applying an energy window rate of total, true, scatter random and Noise
selection of 2 σ around the 511 keV peak, we Equivalent Count (NEC) following the procedure
calculated the difference of the detection times of the in [42] for the first smaller phantom and in [43]
selected events for each line of response (LOR). We for the larger phantom.
defined the time resolution for each LOR as the
FWHM of the coincidence time distribution. The scattered counts were calculated by
subtraction of random, intrinsic and true counts
2) Count rate performance: We produced two from the total number of counts. The random and
phantoms with increasing size. They intrinsic counts were computed according to the
consisted of a high density polyethylene delayed window technique.
(0.96±0.1) g/cm3 cylinder with a length of
(400.0±0.5) mm, (700.0±0.5) mm and a All the delayed coincidences within a 100
diameter of (100.0±0.5) mm, (202.0±0.5) ns time window are recorded as random
mm respectively. coincidences. When the FDG source is not
present inside the FOV, the random coincidences
The volume of the two phantoms was 3140 are considered intrinsic to the Lu-176 radioactive
ml and 22421.8 ml respectively. We carved a decay the random coincidences.
cylindrical hole with a diameter of 3.2 mm
parallel to the longitudinal axis at a radial 3) Sensitivity: We placed a 22Na point source
distance from the center of the FOV of 29.5 mm with 30 µCi activity first in the center of the
and 45 mm respectively. FOV and then we moved it on both sides
along the axial direction in steps of 5 mm in
We filled the hole with 7.9 mCi and 15 mCi order to cover a total distance of 200mm. The
[18F]-Fludeoxyglucose ([18F]-FDG), acquisitiontime for each
respectively, diluted in saline solution. We pointwas3minutesinordertoguaranteeapproxi
placed the phantom on the scanning bed with its mately10000 events.
axial direction coinciding with the axial direction
of the detector ring. We performed a background acquisition
removing the source from the FOV, in order to
Brain Gate Technology Page 16
BRAINGATE TECHNOLOGY
estimate the intrinsic coincidence rate. We 10% chloral hydrate solution, we placed it on the
calculated the sensitivity and the absolute movable stage bed and we injected 5.42 mCi
sensitivity following the procedure in [42]. [18F]-FDG. 50 minutes after the injection we
performed a 20 min PET scan.
4) Spatialresolution: Weusedapoint-like
22Naradioactive source with 30 µCi activity. We performed the image reconstruction
We studied the space resolution at the center with a 3DOSEM/PSF algorithm, with 5 iterations
of the FOV and at a distance of 50 mm from and without scatter and attenuation corrections.
the center along the longitudinal axis. For Human experiments were performed on a 33
both axial positions we placed the source at a years old male human subject (height 176 cm,
series of radial points ranging between 5 mm weight 67 kg) under approval of the First
and 150 mm radial distance from the FOV. Affiliated Hospital, Sun Yat-sen University,
Guangzhou, China.
The acquisition time for each point was 5
minutes. We quoted the axial, tangent and The subject was kept fasting for 24 hr and
longitudinal profiles of the reconstructed image was injected with 4.7 mCi [18F]-FDG. 40
with the technique reported in [42] We quoted the minutes after the injection we performed a 10
axial, tangent and longitudinal profiles with the min PET scan. In order to exploit the potential of
technique. the digital PET technique, the postprocessing of
the acquired digital data was performed with 2 ns
5) Animal and human imaging: For the and 10 ns coincidence time window. We
evaluation of the imaging capabilities of the performed the image reconstruction with a 3D
digital brain PET scanner we performed an OSEM PSF algorithm, with 5 iterations and
animal imaging and a human imaging without scatter and attenuation corrections. We
experiment. did not apply any normalization to the data.
We calculated the contrast ratio (CR) and the physically disabled people by allowing them
contrast to noise ratio (CNR) as CR = to quickly and reliably control a wide range
µwhite/σwhite and CNR = (µgray of devices by thought, including computers,
−µwhite)/σwhite. environmental controls, robotics and medical
devices.
1. The brain gate neural interface system is an allow for robotic control, such as a thought-
developed to improve the quality of life for products may be able to provide an individual
with the ability to control devices that allow fraction of the signals from neurons that fire. It is
breathing, bladder and bowel movements. difficult for the computer to convert these signals
5. The brain gate system has allowed people resulting in the cursor jiggling and making it difficult
with paralysis to operate a computer in order to select icons on the screen with accuracy.
to read e-mail, control a wheelchair and
operate a robotic hand. Other BrainGate short comings include:
6. The system will connect the brain gate sensor
with functional electrical stimulation (FES) 1. Size: Brain gate right now has a bulky look
system, which uses electrical impulses to with cables and processors. The device has to
trigger muscle and limb movement. The first be less bulky to make the technology
version will allow users to make simple mainstream.
movements that could be used to perform
tasks such as eating or drinking using their Cyber kinetics is developing a
own arms and hands and under the natural prototype of a device that would fit behind
control of their own brains. The initial the ear of the patient, much like the cochlear
version of this FES system would use arm implant, and connect via a magnet to the
supports. Later versions, however, won’t computer equipment, thus eliminating the
require supports and will allow users to do need to cross the skin. This will lead to a
activities that require more dexterity, such as wireless Brain Gate, giving the patient
using cell phones or remote controls. greater freedom.
7. The device can be used in an interactive
2. Calibration: In its current form, it is
environment; activity surrounding the patient
essential to recalibrate the device before each
will not affect the accuracy of the device.
use by the patient. The team is working on
could connect muscles to the functioning Limb movement developments are currently
brain, thus bypassing the damaged spinal at the research stage and are not available for use
cord. with the existing Brain Gate System. In addition,
4. The brain is incredibly complex. To say products are in development to allow for robotic
that all thoughts or actions are the result of control, such as a thought-controlled wheelchair. In
simple electric signals in the brain is a gross the future, the Brain Gate System could be used by
understatement. There are about 100 billion those individuals whose injuries are less severe.
neurons in a human brain. Each neuron is
Next generation products may be able to
constantly sending and receiving signals
provide an individual with the ability to control
through a complex web of connections. There
devices that allow breathing, bladder and bowel
are chemical processes involved as well,
movements. Currently The Brain Gate Neural
which EEGs can’t pick up on.
Interface System is an investigational device and is
Single events are collected by each detector concentration of22.7 kBq/ml. The peak of the NEC
module at a rate that is similar to other systems for rate was at 30.7×103
the same injected activity. The challenge here is to counts/satanactivitylevelof7.3kBq/ml.Similarly, the
obtain a digitization of the signal without introducing NEC peak was measured between 50 kcps and 75.1
a significant fractional dead time. Furthermore, as the kcps at an activity of approximately 50 kBq/ml in
data stream includes all prompts, without any [10], [45].
specific event selection, such as time and energy
In a brain PET employing CdTe
window, the additional dead time introduced by
semiconductor detectors [12], the maximal reported
conventional time coincidence electronics is
value of the NEC rate was 48 kcps in the range
removed.
between 16 kBq/ml and 31 kBq/ml. More recently
As a result, as shown in the human acquisition [31], in a brain PET system using offline software
example above, offline analysis for the improvement coincidence detection methods similar to the one
of image quality in clinics may include, among presented in this paper, a NEC peak of 27.9 kcps at
others, optimization of the coincidence time window 11.9 kBq/ml was reported.
and energy window.
Despite the high true coincidence rate, the
Thanks to the small size of the data package, result was mainly limited by an high rate of random
the fractional dead time of the digital sampling coincidences. Finally, in a brain PET prototype based
electronics does not impact significantly the scanner on LySO/SiPM sensors and on a digital signal
performance, allowing achieving optimal count rate processing with FPGA-embedded ADC board [46],
and sensitivity, with respect to the geometry of the the best achieved NEC rate was 6.5 kcps at an
scanner. While in previous studies [38], [39] we have activity of 5.5 kBq/ml.
demonstrated the applicability of this technique to
This low value was mainly due to a short axial
small scale prototypes, in this paper we show for the
FOV. The maximal NEC rate of the digital brain PET
first time the performances of the P&I techniques in
system proposed in this paper is thus approximately
a large-scale medical imaging system.
two times better than the currently available state-of-
The figure of merit of the digital brain PET the art systems. The count rate performance could be
scanner is the count rate performance. In brain PET improved considering that the scatter fraction of 53%
systems with conventional analog electronics measured for the human-sized phantom reflects the
readout, such as in [44], the peak count rate for the absence of any tungsten shielding in the prototype.
trues was at 323.5×103 counts/s at an activity
Brain Gate Technology Page 21
BRAINGATE TECHNOLOGY
By way of example, a value as low as 34.7% LabPET4 system with 1:1 coupling reports 1.51 mm
may be obtained in dedicated brain PET scanners, resolution since the crystal pitch is 2 mm [49].
suggesting that it could be of interest to add a
The OSEM-3D reconstruction method
tungsten septa for reduction of external radiation
allows improving the spatial resolution to
[46].
approximately 1.7 mm (FWHM) at the center of the
As for the scanner sensitivity, the best FOV. However, a disadvantage of the digital brain
possible value expected in compact brain PET PET is the absence of the DOI correction.
systems with cylindrical shape was determined in
The effect of DOI is most notably in
dedicated simulations as 61 cps/kBq [41], [1].
transaxial images at large distance from the axis and
High sensitivity scanners obtained with then primarily in the tails of the point-spread
analog electronics readout were designed. For function, with a consequent degradation of the space
example, the human brain PET resolution with large distortions reaching more than
insertsystembySIEMENShasachievedthehighestsens 100% increase in FWHM at 100 mm off axis.
itivity in the center of the FOV of approximately 64
This feature was observed in brain PET
cps/kBq [44]. Higher sensitivity in the range between
scanners without DOI measurement. In [46], the
150 cps/kBq and 400 cps/kBq were obtained in [10],
radial and tangential resolutions varied from 3.1 mm
[45].
to 3.5 mm and 3.1 mm to 5.4 mm at the center of the
Similarly a maximal sensitivity of 57 FOV and at 120 mm off-center respectively.
cps/kBq was reached in [47]. More recent brain PET
Most of recent brain PET systems are based
designs reached sensitivities ranging between 19
on a correction of the DOI, which is instrumental to
cps/kBq [31] and 21 cps/kBq [13]. The sensitivity of
achieve a better space resolution. By way of
the digital brain PET proposed in this paper is thus
example, in [31] a four-layer depth of interaction
close to the best expectation for modern brain PET
detector was designed containing five detector units
scanners.
axially lined up per layer board.
The spatial resolution at the center of the
Each of the detector units consisted of a
FOV obtained with the FBP reconstruction is
finely segmented (1.2 mm) LYSO scintillator array
approximately 3.3 mm (FWHM) and is consistent
and an 8×8 array of multi-pixel photon counters,
with the expected resolution of a PET system with
allowing a space resolution of approximately 0.9 mm
3.95 mm pitch [48]. By way of example, the
(FWHM). In similar approaches, the possible
Brain Gate Technology Page 22
BRAINGATE TECHNOLOGY
achieved space resolution thanks to the DOI high resolution brain PET systems was already
correction could be in the range between 0.7 mm and appreciated in [9], where, for example, the NEMA-
1.7 mm [13]. 2001 count rate study with 370 MBq (a clinical
relevant dose) in the 700 mm phantom generated in
The drawback of such scanners, as mentioned
1 h a list mode file of approximately 100 Gbyte in
above, is the lower sensitivity, because of the higher
size.
complexity of the system. Imaging performance is
optimized offline after data acquisition. This unique This not only demanded large data storage
feature of the digital brain PET prototype proposed discs and backup facilities, but also required very fast
in this paper represents a clear advantage with data handling from the PET acquisition computer,
respect to traditional PET systems, where the considering the rate at which the data stream came
coincidence time window is applied during data in.
taking and an offline optimization of the image
CONCLUSION
quality changing the basic acquisition parameters
such as energy or coincidence time window is not
possible. The technology driving this breakthrough in
the Brain-Machine-Interface field has a myriad of
In the reported brain imaging cases a change potential applications, including the development of
of the coincidence time window from 10 ns to 2 ns human augmentation for military and commercial
improves the noise level (CR) of approximately purposes The primary goal of this technology and
50%, due to the lower component of random devices like brain gate is to help those are who are
coincidences, without adversely affecting the image paralyzed to perform routine activities that are part
contrast (CNR). Finally, it is important to note that of normal human existence. The brain gate can be
the digital readout scheme proposed in this paper used to replace the memory center in patients
requires a novel approach to computing for PET. affected by strokes, epilepsy or Alzheimers disease.
The high sensitivity and count rate The ‘BrainGate’ device can provide paralyzed or
performance, together with the four threshold time motor-impaired patients a mode of communication
stamping technique, implies to collect approximately through the translation of thought into direct
1.7 Gbytes/s at the maximal prompts count rate of computer control. Normal humans may also be able
100 Mcps. The raw data of a typical digital brain PET to utilize BrainGate technology to enhance their
scan may thus approach a volume of few hundreds relationship with the digital world provided they are
Gbytes. This trend in modern high sensitivity and willing to receive the implant.
This work was supported in part by the [4]. Cuthbertson, Anthony (3 April 2021). "Scientists
Natural Science Foundation of China (NSFC) connect human brain to computer wirelessly for first
(Grants 61425001, 61210003, 61604059, 61601190 time ever". The Independent. Retrieved 5 April 2021
and 61671215); in part by the Natural Science [5] A. M. Scarf and M. Kassiou, “The translocator
Foundation of Hubei Province (Grant 2016CFA005); protein,” Journal of nuclear medicine, vol. 52, pp.
in part by National Key Scientific Instrument and 677–680, 2011.
Equipment Development Program of China
(2013YQ030923); in part by National Key Scientific [6] B. F. Huton, T. Yamaya, and L. R. Furenlid,
Instrument and Equipment Development Program of “Dedicated molecular imaging systems for human
Hubei Province (2013BEC050); in part by National neurological studies,” IEEE Trans Rad Plasma Med
Key Research and Development Program of China Sci, vol. 3, no. 3, pp. 252–253, 2019. [7]
Intracortical Neural Interface System for Persons Nohara, T. Tomitani, M. Yamamoto, H. Murayama,
and M. Endo, “Development of a high resolution Badawi, “Optimal whole-body pet scanner
PET,” IEEE Transactions on Nuclear Science, vol. configurations for different volumes of lso
37, no. 2, pp. 594–599, 1990. scintillator: a simulation study,” Phys. Med. Biol.,
vol. 57, p. 4077, 2012.
[11] R. Boellaard, F. Buijs, H. W. A. M. de Jong, M.
Lenox, T. Gremillion, and A. A. Lammertsma, [15] A. Gonzalez, J. Gonzalez-Montoro, L. F. Vidal,
“Characterization of a single LSO crystal layer High J. Barbera, S. Aussenhofer, L. Hernandez, L.
Resolution Research Tomograph,” Physics in Moliner, F. Sanchez, C. Correcher, E. Pincay, G.
Medicine and Biology, vol. 48, no. 4, pp. 429–448, Canizares, E. Lamprou, S. Sanchez, J. Catret, S.
2003. Jim´enez-Serrano, J. Cabello, M. Schwaiger, A.
Iborra, T. Merlin, D. Visvikis, and J. Benlloch,
[12] Y. Morimoto, Y. Ueno, W. Takeuchi, S.
“Initial results of the mindview pet insert inside the
Kojima, K. Matsuzaki, T. Ishitsu, K. Umegaki, Y.
3t mmr,” IEEE Trans Rad Plasma Med Sci, vol. 3,
Kiyanagi, N. Kubo, C. Katoh, T. Shiga, H. Shirato,
no. 3, pp. 343–351, 2019.
and N. Tamaki, “Development of a 3D brain PET
scanner using CdTe semiconductor detectors and its [16] C. E. Bauer, J. Brefczynski-lewis, G. Marano,
first clinical application,” IEEE Transactions on M.-b. M. A. Stolin, P. Martone, J. W. Lewis, G.
Nuclear Science, 2011. Jaliparthi, R. R. Raylman, and S. Majewski,
“Concept of an upright wearable positron emission
[13] A. J. Gonz´alez, S. Majewski, F. S´anchez, S.
tomography imager in humans,” no. May, pp. 1–10,
Aussenhofer, A. Aguilar, P. Conde, L. Hern´andez,
2016.
L. F. Vidal, R. Pani, M. Bettiol, A. Fabbri, J. Bert, D.
Visvikis, C. Jackson, J. Murphy, K. O. Neill, and J. [17] S. Majewski, “HelmetPET: a silicon
M. Benlloch, “Nuclear Instruments and Methods in photomultiplier based wearable brain imager,” in
Physics Research A The MINDView IEEE Nuclear Science Symposium Conference
brainPETdetector,feasibilitystudy based Record, 2011, pp. 4030–4034.
onSiPMarrays,” Nuclear Inst. and Methods in
[18] C. Bauer, J. Brefczynski-Lewis, J. W. Lewis, M.
Physics Research, A, vol. 818, pp. 82–90, 2016.
B. Mandich, and S. Majewski, “A novel upright low-
[Online]. Available:
dose brain imager: a new tool in neuroscience and its
http://dx.doi.org/10.1016/j.nima.2016.02.046
potential applications,” Society Neuroscience, vol.
[14] J. K. Poon, M. L. Dahlbom, W. W. Moses, K. 658, pp. 13–16, 2014.
Balakrishnan, W. Wang, S. R. Cherry, and R. D.