GOVERNMENT POLYTECHNIC COLLEGE

VECHOOCHIRA

SEMINAR REPORT

ON

FUNCTIONAL NEAR INFRARED SPECTROSCOPY

SUBMITTED BY

CHRISTIN A KOCHUMON

REG NO. 20240534

DEPARTMENT OF BIOMEDICAL ENGINEERING

2022 - 2023
 GOVERNMENT POLYTECHNIC COLLEGE

VECHOOCHIRA

CERTIFICATE

This is to certify that Mr. Christin A. Kochumon (Reg. No. 20240534) fifth semester
student of biomedical engineering has satisfactorily completed the seminar report entitled
"FUNCTIONAL NEAR INFRARED SPECTROSCOPY" in accordance with the
syllabus of technical department of Kerala.

HEAD OF DEPARTMENT GUIDE

EXTERNAL EXAMINER INTERNAL EXAMINER

ii
 ACKNOWLEDGEMENT

Al the outset I think good for making my endeavour a success. I also express my gratitude to
Mrs. RENU B JOSE, principle of the institution, for providing me with adequate facilities
ways and means by which I was able to complete this seminar.

I express my sincere gratitude to my seminar guide Mrs. ASHA N.D, for this constant
support and valuable suggestions without which the successful completion of this seminar
would not have been possible.

I would like to extend my sincere appreciation to Mrs. ASHA N.D, head of biomedical
engineering for their co-operation and support.

I express my immense pleasure and thankfulness to all teachers and staff of the department of
Biomedical Engineering.

Last but not the least I thank nil other and especially my classmates and my family members
who on one way or another helped me in the successful Completion of this.

CHRISTIN A KOCHUMON

iii
 ABSTRACT

Functional Near-Infrared Spectroscopy (fNIRS) maps human brain function by

measuring and imaging local changes in hemoglobin concentrations in the brain that arise
from the modulation of cerebral blood flow and oxygen metabolism by neural activity. Since
its advent over 20 years ago, researchers have exploited and continuously advanced the
ability of near infrared light to penetrate through the scalp and skull in order to non-
invasively monitor changes in cerebral hemoglobin concentrations that reflect brain activity.
~We=review recent advances in signal processing and hardware that significantly improve
the capabilities of fNIRS by reducing the impact of confounding signals to improve statistical
robustness of the brain signals and by enhancing the density, spatial coverage, and
wearability of measuring devices respectively. Then summarize the application areas that are
experiencing rapid growth as fNIRS begins to enable routine functional brain imaging.

iv
 CONTENTS

1. INTRODUCTION 1
2. FUNCTIONAL NEAR-INFRARED SPECTROSCOPY

2.1 FNRI MACHINE 2

2.1.1 PRINCIPLES OF FUNCTIONAL NEAR-INFRARED SPECTROSCOPY 2

2.1.2 EVENT-RELATED OPTICAL SIGNAL 4

2.1.3 FNIRS SYSTEM TYPES 6

2.1.4 FNIR SYSTEM DESIGNS 6

2.1.5 FNIR SENSOR DESIGNS 7

2.2 COMPARISON OF FNIRS WITH OTHER NEUROIMAGING MODALITIES

2.2.1. COMPARISON OF FNIRS AND EEG, ERP 9

2.2.2 COMPARISON OF FNIRS AND FMRI, PET 9

2.3 RECENTLY ADVANCE IN FNIRS

2.3.1 HARDWARE ADVANCES 12

2.3.1.1 HIGH-DENSITY FNIRS 12

2.3.1.2 WEARABLE FNIRS 13

2.3.2 SIGNAL PROCESSING ADVANCES 14

2.3.2.1 STATISTICS 14

2.3.2.2 SHORT-SEPARATION REGRESSION OF INTERFERING

PHYSIOLOGICAL SIGNALS 15

2.3.2.3 MOTION ARTIFACTS 16

2.3.2.4 ANATOMICAL GUIDANCE 16
2.4 APPLICATIONS
2.4.1 BRAIN DEVELOPMENT 17
2.4.2 APPLICATION TO THE COGNITIVE AND PSYCHOLOGICAL SCIENCE 18
2.4.3 BRAIN COMPUTER INTERFACE 19
2.4.4 HYPERSCANNING 20
2.5 FNRIs PUBLICATION PER YEAR 21
3. CONCLUSION 22

v
 LIST OF FIGURES

PAGE
FIGURE NO. NAME OF FIGURE
NO
2.1 FNRIS MACHINE 2

2.3 BANANA-SHAPED PHOTON PATH 5

2.4 PORTABLE FNIR SYSTEM DESIGN 6

2.5 THE WIRELESS FNIR SYSTEM DESIGN 7

2.6 (A) FLEXIBLE SENSOR DESIGN (B) A PARTICIPANT 7

WEARING THE FLEXIBLE SENSOR

2.7 HIGH-DENSITY FNIRS 12

2.8 WEARABLE FNIRS 13

2.9 (A) EXAMPLE OF FNIRS CAP ON THE HEAD OF A 17

7-MONTH OLD INFANT

(B) FNIRS HEADGEAR ON A 13-MONTH-OLD INFANT. 17

2.10 BATTERY OPERATED AND WIRELESS UNIT 18

2.11 HYPERSCANNING FNIRS 20

2.12 VOLUME OF FNIRS PUBLICATIONS PER YEAR 21

vi
 CHAPTER-1

INTRODUCTION

Functional Near-Infrared Spectroscopy (fNIRS) is a non-invasive, non-ionizing

method for measuring and imaging the functional hemodynamic response to brain activity.
Near-infrared light can propagate several centimeters through the scalp and skull, and
spectroscopically interrogate the concentrations of oxygenated (HbO), deoxygenated (HbR),
and total (HbT) hemoglobin within the brain. By shining near-infrared light on the scalp and
placing a detector a few centimeters away, changes in the amount of diffuse light reaching the
detector provide a measure of changes in cerebral hemoglobin concentrations. Since its first
implementation over 20 years ago, fNIRS has proven to be an effective tool to study normal
brain function and its alteration in disease. Similar to EEG, fNIRS' safety, low-cost,
portability, and high temporal resolution give it the potential for widespread implementation.
An important difference is that while EEG measures the fast, electrical responses associated
with neuronal activity, fNIRS relies on neurovascular coupling and measures the
hemodynamic response "just as is done with fMRI. fNIRS is particularly suited for
populations and studies for which other imaging modalities are limited (e.g. fMRI), including
infants and children, procedures involving mobility and inter-activity, and clinical
environments. As recently reviewed in a Neurolmage special issue, dominant application
areas include behavioral and cognitive neurodevelopment, perception and cognition,
psychiatric conditions, and neurological applications including epilepsy, stroke and brain
injury. The exponential growth of fNIRS applications over the past 25 years and what we
believe are major milestones in the field. Beyond the traditional continuous-wave approach.
Notably, time-domain (TD) NIRS enables "null-distance" depth resolution, and diffuse
correlation spectroscopy (DCS) is a NIRS cousin technique sensitive to the motion of red
blood cells that provides an index of blood flow with enhanced brain sensitivity compared
with NIRS.

1
 CHAPTER-2

FUNCTIONAL NEAR-INFRARED SPECTROSCOPY

2.1 FNRI MACHINE

2.1.1 PRINCIPLES OF FUNCTIONAL NEAR-INFRARED SPECTROSCOPY

It is well known that the functional state of tissue can influence its optical properties
(for instance, cyanosis in hypoxia and pallor in anemia). The human brain undergoes a
number of physiological changes as it responds to environmental stimuli; these changes in
blood levels and electrochemical activity also affect its optical properties. Functional optical
imaging capitalizes on the changing optical properties of these tissues by using light in the
near-infrared range (700-900 nm) to measure physiological changes.

Brain activity is associated with a number of physiological events; two of these events
can be assessed using optical techniques. During neural activity, ionic fluxes across the cell's
membrane (e.g., shifts in sodium and potassium ions) result in a change in the membrane
potential. The ionic fluxes also cause changes in the magnetic and electrical fields, which,
when summed across a large number of synchronously activated neurons, can be assessed
using EEG or MEG. Neuronal activity is fueled by glucose metabolism, so increases in neural

2
activity result in increased glucose and oxygen consumption from the local capillary bed. A
reduction in local glucose and oxygen stimulates the brain to increase local arteriolar vasodi-
lation, which increases local cerebral blood flow (CBF) and cerebral blood volume (CBV), a
mechanism known as neu-rovascular coupling. Over a period of several seconds, the
increased CBF carries both glucose and oxygen to the area, the latter of which is transported
via oxygenated hemoglobin in the blood. The increased oxygen transported to the area
typically exceeds the local neuronal rate of oxygen utilization, resulting in an overabundance
of cerebral blood oxygenation in active areas, Although the initial increase in neural activity
is thought to result in a focal increase in deoxygenatediiemoglobin in the capillary bed as
oxygen is withdrawn from the hemoglobin for use in the metabolization of glucose, this
feature of the vascular response has been much more difficult to measure, and more
controversial, than hyperoxygenation.

Because oxygenated and deoxygenated hemoglobin (oxy-Hb, deoxy-Hb) have

characteristic optical properties in the visible and near-infrared light range, the change in
concentration of these molecules during neurovascular coupling can be measured using
optical methods. The most commonly used method of near-infrared spectroscopy measures
changes in the ratio of oxy-Hb"to blood volume. Most biological tissues are relatively
transparent to light in the near-infrared range between 700-900 nm, largely because water, a
major component of most tissues, absorbs very little energy at these wavelengths. However,
the chromophores oxy-Hb and deoxy-Hb do absorb a fair amount of energy in this range. As
such, this spectral band is often referred to as the optical window for the noninvasive
assessment of brain activation £6J. Photons introduced at the scalp pass through most of the
tissue and are either scattered or absorbed by it. Because a relatively predictable quantity of
photons follows a banana-shaped path back to the surface of the skin, these photons can be
measured at the scalp using photodetectors. If wavelengths are chosen to maximize the
amount of absorption by oxy-Hb and deoxy-Hb, changes in the chromophore concentrations
cause changes in the number of photons that are absorbed and the number of photons that are
scattered back to the surface of the scalp. These changes in light intensity measured at the
surface of the scalp are quantified using a modified Beer-Lambert law, which is, essentially,
an empirical description of optical attenuation in a highly scattering medium. By measuring
absorption changes at two (or more) wavelengths, one of which is more sensitive to oxy-Hb,
the other to deoxy-Hb, changes in the relative concentrations of these chromophores can be
calculated.

3
4
Procedure

Typically, a functional near-infrared (fNIR) apparatus is comprised of a light source

that is coupled to the participant's head via either light-emitting diodes (LEDs) or though
fiber-optical bundles (the optode) and with a light detector that receives the light after it has
interacted with the tissue. Because light is scattered after entering the tissue, a photode-tector
placed 2-7 cm away from the optode can collect light after it has passed through the tissue.
When the distance between the source and photodetector is set at 4 cm, the fNIR signal
becomes sensitive to hemodynamic changes within the top 2-3 mm of the cortex and extends
laterally 1 cm to either side, perpendicular to the axis of source-detector spacing. Studies
have shown that at interoptode distances as short as 2-2.5 cm, gray matter is part of the
sample volume, Using this technique, several types of brain activity have been assessed,
including motor activity, visual activation, auditory stimulation, and the performance of
cognitive tasks.

2.1.2 EVENT-RELATED OPTICAL SIGNAL

Figure 2.3

A second, much more controversial method, labeled the event-related optical signal,
or EROS, capitalizes on the changes in the optical properties of the cell membranes
themselves that occur as a function of the ionic' fluxes during firing. Using invasive
techniques, it has been well established that the optical properties of cell membranes change
in the depolarized state relative to the resting state, and that optical methods can be used to
detect these changes. The ability to measure the actual depolarization state of neuronal tissue
provides obvious advantages in that it is a direct measure of neural activity, with millisecond-
level time resolution but with the superior spatial resolution lacking in EEG/MEG. However,
there are also a number of limitations to the noninvasive use of the EROS signal in humans.

5
A primary disadvantage of the fast optical signal is the low signal-to-noise ratio (SNR)
resulting from the need to image through skin, skull, and cerebral-spinal fluid. Basic sensory
and motor movements such as tactile stimulation and finger-tapping require between 500-
1,000 trials to establish a reliable signal. The low SNR may play a role in current difficulties
with experimental replication; however, more cross-validation work is warranted. The final
constraint is that these methods require a more expensive and cumbersome laser-based light
source (versus an LED-based light source), they are not portable, and the potential risk of
inadvertent damage to the eyes is increased relative to the systems available for measuring
hemodynamic responses. (LED-based near-infrared sources pose very little, if any, risk upon
eye exposure). In spite of these current limitations, the fast optical signal continues to be an
important area of investigation because it offers glimpses of the "holy grail" of neuroimaging:
the direct measurement of neuronal activity with millisecond time resolution and superior
spatial resolution.

2.1.3 FNIRS SYSTEM TYPES

A wide variety of both commercial and custom-built fNIR instruments are currently in
use. These systems differ with respect to their use and system engineering, with tradeoffs
between light sources, detectors, and instrument electronics that result in tradeoffs in the
information available for analysis, safety, and cost. Three distinct types of fNIR
implementation have been developed: time-resolved systems, frequency-domain systems, and
continuous wave spectroscopy systems, each with its own strengths and limitations. Time-
resolved and frequency-domain systems provide information on shifts in both phase and
amplitude of the light and are necessary for more precise quantification of fNIR signals.
Continuous wave (CW) systems apply either continuous or a slow-pulsed light to tissue and
measure the attenuation of amplitude of the incident light. These systems utilize less
sophisticated detectors than time-resolved and frequency-domain systems, and, therefore,
they cannot resolve the time-changing component of the light. As such, CW systems provide
somewhat less information than time- or frequency-domain systems, but this tradeoff results
in the capacity to design more compact and inexpensive hardware, making it advantageous
for specified applications. They can be laser-based, but light-emitting diodes (LEDs) can also
be used in CW designs to increase safety (particularly with respect to eye exposure) and to
again decrease both instrument size and cost, making it possible to deploy these systems in
clinical or educational settings.

6
2.1.4 FNIR SYSTEM DESIGNS
PORTABLE FNIR SYSTEM

Figure 2.4
The current generation is comprised of three primary components: 1) an LED-based
sensor that covers the entire forehead of the participant, 2) a control module (approximately 2
in x 6 in x 8 in) with integrated power supply for sensor control and data acquisition, and 3) a
laptop computer for the data analysis software.

WIRELESS FNIR SYSTEM

Figure 2.5

The wireless fNIR is a miniaturized CW system designed to monitor hemodynamic

responses of the cortex under ambulant conditions. The system consists of two primary
components: a wearable device to collect and transmit the data and a data analysis and
display computer. The wearable device comprises three elements: 1) a personal digital
assistant (PDA), 2) a control circuit and battery holder, and 3) the sensor. The PDA (currently
an HP iPAQ Pocket PC) supports both integrated Wi-Fi (IEEE 802.11b standard) and
Bluetooth for wireless communication. The customized PDA software controls the sensor

7
circuitry, reading, saving, and sending the data via a wireless network. The data acquisition
rate of the system is 2 Hz (two data points per second).
2.1.5 FNIR SENSOR DESIGNS FLEXIBLE SENSOR DESIGN

Figure 2.6
The current flexible sensor, developed in the Drexel Biomedical Engineering laboratory,
consists of four LED light sources and ten detectors, which cover the forehead using 16
voxels, with a source-detector separation of 2.5 cm,. The light sources contain three built-in
LEDs having peak wavelengths at 730, 805, and 850 nm, with an overall outer diameter of
9.2 ± 0.2 mm. The photodetectors are monolithic photodiodes with a single supply
transimpedance amplifier having the size of 0.90 in x 0.90 in. Communication between the
data analysis computer and the task presentation computer is established via a serial port
connection to time-lock fNIR measurement to computer-generated task events.

The flexible sensor design consists of three parts: a reusable, flexible circuit board
that carries the necessary infrared sources and detectors, a replaceable cushioning material,
and a disposable, single-use medical-grade adhesive tape that serves to attach the sensor to
the participant. The flexible circuit provides a reliable and integrated wiring solution as well
as consistent and reproducible component spacing and alignment. Because the circuit board
and cushioning materials are flexible, the components move and adapt to the various contours
of the participant's head, allowing the sensor elements to maintain an orthogonal orientation
to the skin surface, dramatically improving light coupling efficiency and signal strength.

MODULAR SENSOR DESIGN

Currently, the optical team is working on a modular sensor design with adjustable
channel placements. The purpose of the modular design is to provide a low-cost, scalable
sensor for any forehead application. The modular design will allow the investigator to place
individual voxels according to the International 10/20 system, a measurement system widely
used in EEG applications that allows the placement of sensors over consistent brain regions
across different head sizes and shapes.

8
SOFTWARE APPLICATIONS

The expanding scope of fNIRs has resulted in an increasing demand for new signal
processing algorithms, both to increase the SNR of the fNIR system in the extraction of the
hemodynamic response during cognitive tasks (to identify, eliminate, and/or compensate for
noise and other signal distortions, such as electronic drift) and to extract physiologically
relevant information from the measurements. In addition, algorithms are necessary to
compensate for movement artifacts under ambulant conditions. The Drexel team has
developed a number of data-processing algorithms to address these issues. These algorithms
are currently being refined and integrated into an end-user software package that controls the
fNIR system and allows post experimental processing of the data and the production of
images for data presentation.

2.2 COMPARISON OF FNIRS WITH OTHER NEUROIMAGING MODALITIES

2.2.1. Comparison of fNIRs and EEG, ERP

Electroencephalographic (EEG) or ERP measures of physiological function. These

measures have several advantages for research, in that they are relatively inexpensive,
noninvasive, and have nearly instantaneous time resolution. They can be used with infants
and children, as well as adults, and can be used repeatedly with no adverse effects. As a
result, EEG and ERP paradigms have contributed important data for developing models of
cognitive and emotional processing. However, EEG measures are limited in their ability to
provide the precise location of an electrical source. EEG does yield spatial information, but
this spatial information must be reconstructed by probabilistic models. Although
mathematical models are improving, even when applied in very-high-density fields, EEG can
only provide a relative approximation of current sources.

2.2.2 Comparison of fNIRs and fMRI, PET

The introduction of neuroimaging modalities such as fMRI and PET has made it
possible to examine much more precisely the anatomical location of the neural circuitry
underlying various mental events among human participants. Many clinicians are familiar
with the basic principles of fMRI and utilize the result of fMRI research to inform their
clinical practice and research. fMRI is currently considered the "gold standard" for measuring
functional brain activation because it offers safe, noninvasive neuroimaging with high spatial
resolution. It is therefore useful to briefly compare and contrast fNIRs with the well-known
technology of fMRI. One of the primary measures used for fMRI is the blood-oxygen-level-

9
dependent (BOLD) signal that accompanies neuronal activation in the brain secondary to, for
instance, the presentation of a stimulus. As previously discussed, increased CBF to an active
area exceeds the additional neuronal metabolic demand, resulting in a decrease of deoxy-Hb
concentration in the local tissue. The magnetic susceptibility of blood containing oxy-Hb
differs very little from water or other tissues, which have low paramagnetic properties.
However, deoxy-Hb is highly paramagnetic and, therefore, has very different magnetic
properties than surrounding tissues and can act as a naturally occurring contrast agent, The
presence of deoxygenated blood in a given area results in a less uniform magnetic field.
Because the MRI signal depends on the uniformity of the magnetic field experienced by
water molecules, less uniformity fr.c, whoiijnoix^co^Sb-^s-prpsent) results in a greater
mixture of signal frequencies and, therefore, a more rapid decay of the overall signal. In
contrast, as the deoxygenated blood in a given area is replaced by oxygenated blood, the local
magnetic environment becomes more uniform, and the MRI signal lasts longer and is
consequently stronger during image acquisition. The signal change is typically around 1% or
less, depending on the strength of the magnetic field. Therefore, fMRI, like fNIR, is an
indirect measure of neuronal activity, assessing changes in the relative concentration of
deoxy-Hb in local tissue. There is no simple relationship between the magnitude of the signal
change and any single physiological parameter because it relies on changes in blood flow,
blood volume, and local oxygen tension. There is also a time delay between when the local
neurons are activated and begin to use oxygen and when vasodilation occurs, which allows an
increased blood flow and the transport of oxy-Hb to the area. Labeled the hemodynamic
response, this process occurs over several seconds, following the initiation of neuronal
activity.

The more commonly used fNIRs shares much in common with the BOLD-based
signal, as it measures relative changes in concentrations of deoxy-Hb that are dependent on
the hemodynamic response. As such, both are indirect measures of neural activity, with
temporal resolution on the order of seconds because they are limited by the hemodynamic
response. In addition, just as arterial spin labeling can be used to calculate blood flow with
fMRI, because fNIRs measures relative changes in oxy-Hb as well as deoxy-Hb, total blood
flow can be calculated from the differential equation. Both technologies also provide a level
of spatial resolution, are safe and noninvasive, and can be used repeatedly with the same
individuals. Due to their SNRs, both technologies typically require some level of repeated
stimulation, either in a block design or an event-related design. From there, however, there

10
are a numbe of important differences between the two technologies.

fNIR is unlikely to supplant fMRI for basic research on the neurophysiological

underpinnings of various cognitive, emotional, and motivational processes for two important
reasons. First, fMRI has better spatial resolution, on the order of 1 mm2, although the fast
imaging of fMRI reduces its spatial resolution somewhat to a few millimeters relative to
conventional MRI. By contrast, due to the scatter of photons in a diverse medium, current
fNIRs systems have a spatial resolution on the order of 1 cm2. Second, fMRI has the capacity
to image the entire brain, whereas fNIR is limited to the outer cortex. Although a large
hemorrhage might be able to be imaged as deeply as the thalamus with fNIRs, more subtle
signals, such as those induced by a cognitive or emotional event, are limited to a depth of
approximately 2-A mm of the cortex.
Conversely, fNIR has a number of advantageous properties that hold enormous
potential for research studies and clinical applications that require the quantitative
measurement of hemodynamic changes in the cortex under a variety of conditions not
amenable to fMRI. The limitations of fMRI relative to fNERs include the fact that
participants must lie within the confines of the magnet bore, which limits its use for many
applications, including the imaging of many patients with significant symptoms. The
refrigerant systems used to supercool the magnets also produce loud noises, which can
interfere with certain protocols. fMRI is also highly sensitive to movement artifact; subject
movements on the order of a few millimeters can invalidate the data. The intense strength of
the magnets necessary to create the MRI signal precludes the use of any ferrous metals in or
around the magnet. Finally, fMRI systems are quite expensive, with an initial cost of a few
million U.S. dollars, depending on the strength of the magnet, and individual participant runs
can cost several hundred U.S. dollars each.
In comparison to fMRI, with fNIRs participants can sit upright and work on a
computer, watch television or movies, and even walk on a treadmill. fNIRs systems are not as
susceptible to movement artifact as fMRI, and algorithms for the removal of motion artifact
during desktop use, as well as ambulatory use, are currently being refined, These attributes
also allow fNIRs to be used with children and with patient populations that may find
confinement to an fMRI magnet overwhelming or painful. A number of sensor applications
exist, depending on their use, including caps, tension straps, and medical-grade adhesive
applications. fNIR is quiet and comfortable and is therefore amenable to sensitive protocols
such as the induction of positive moods. It is readily amenable to integration with a number

11
of other technologies, including EEG. Portable systems exist that operate from a laptop
computer and a control box approximately 2 in x 6 in x 8 in.

2.3 RECENTLY ADVANCE IN FNIRS

2.3.1 HARDWARE ADVANCES

2.3.1.1 HIGH-DENSITY FNIRS

12
2.3.1.2 WEARABLE FNIRS

Wearable fNIRS is the next advance that will transform the technology, enabling
studies of brain activity associated with natural behaviors in ecologically valid settings and
dramatically reducing the cost of fNIRS systems. Taking full advantage of the portability of
fNIRS will dramatically increase the spectrum of applications. It will allow studies of normal
and pathological brain function in more natural environments (social interactions, psychiatric
diseases, stroke recovery), and more efficient monitoring of patients with brain injury or
neurodegenerative disease (stroke, Alzheimer's, concussion) and of normal and abnormal
brain development (autism, language development). In addition, the lower cost of wearable
devices and their flexibility will result in higher accessibility to the technology, compared to
expensive conventional brain imaging technologies.

13
 Dispensing with optical fibers and placing light emitters and receivers directly on the
head, wearable devices offer lower weight and higher flexibility than traditional NIRS
systems. The majority of wearable devices take advantage of light-emitting diodes (LEDs)
and photodiodes. Early portable systems used electrical cables, more prone to RF
interference, to transfer analog signals to a controller module, typically held in a backpack.
More recent wearable systems implement digital conversion directly in the optode. These
systems can be extremely light, but generally afford only low dynamic range on the detected
signal, and restricted measurements to fixed source-detector separations and to forehead only.
Choi developed a dedicated integrated circuit (IC) for multi-channel, high-dynamic range
detection. In 2016, seeing a need for probes that can be adjusted to users' specific needs, two
research groups proposed modular designs for scalable fNIRS probes. Luhmann shared their
design online for easy replication. Chitnis showed very promising in vivo results with good
signal-to-noise ratio at larger than 5-cm source-detector separations on hairy regions of the
head. Given this impressive performance achieved with inexpensive photodiodes and LEDs,
and continued miniaturization of digital converters and FPGAs, we anticipate significant
design advances in wearable fNIRS systems over the coming five years that will have a
profound impact on the growth and utility of fNIRS applications. There are no technical
challenges to realizing these advances other than optimizing the circuit designs based on
available components to achieve highly multiplexed measurements of brain activity. These
wearable and low-cost systems will greatly aid with the adoption of high density fNIRS
measurements that are presently constrained by the bulk of fiber optics which would now be
replaced by much smaller and lighter electrical cables.

2.3.2 SIGNAL PROCESSING ADVANCES

2.3.2.1 STATISTICS

Similar to the analysis of other neuroimaging modalities, specifically functional MRI,

the contrast in fNIRS is based on changes in the level of oxy- and deoxy-hemoglobin
between two or more task or rest conditions. During a typical functional brain study, a
participant performs repeated trials of a specific task(s). A statistical model is used to detect
differences in the level of hemoglobin between a pair of specific tasks or between task and
baseline. The most common statistical model used is a linear regression model, where
measured hemoglobin changes are modeled as linear combinations of regressors derived from
the timing of the stimulus events, and the evoked hemodynamic response is solved by
deconvolution or using a canonical model. As reviewed by Huppert, there is continued debate

14
about the optimal model to maximize sensitivity of the fNIRS analysis. This is a particular
problem in infant and pediatric studies where the shape of the appropriate canonical model is
even less certain and evolving with age.

There are several important distinctions between fNIRS and fMRI analyses in terms
of noise structure and types of artifacts. Of particular concern is that the "noise" (i.e. the un-
modelled signal) is generally correlated and not normally-distributed as it arises from
systemic physiology and motion artifacts. Pre-coloring or pre-whitening general linear
models are used to correct these problems, however approaches commonly used in IMRI-
based GLM analysis do not necessarily directly translate to fNIRS. In particular, the most
appropriate model for physiological noise in fNIRS is still being debated with considerations
being given to band-pass filtering, auto-regressive filters, adaptive or moving-averaging
filters, and regression techniques including the use of polynomials, wavelets, or externally
measured physiological time-courses such as finger-tip pulse-oximetry. As these methods
continue to be proposed, there is an increasing need for rigorous comparisons of the tradeoffs
in the sensitivity and specificity of these methods. We discuss further some of the promising
approaches to address the contamination by systemic physiology and motion artifacts in the
following sections.

2.3.2.2 SHORT-SEPARATION REGRESSION OF INTERFERING

PHYSIOLOGICAL SIGNALS

As the majority of the optical signal comes from the scalp and not the brain, any
hemodynamic fluctuations in the scalp interferes with accurate estimation of the stimulus-
evoked responses in the brain and correction methods should be implemented to improve
estimation accuracy. Methods that use standard averaged long-distance fNIRS signals to
derive an estimate of the global physiological interference to use as a regressor, run the risk
of removing the actual brain response as this does not provide a true independent measure of
global physiology. Methods that use auxiliary measurements such as blood pressure,
respiration, and heart rate variability, on the other hand, cannot resolve spatial heterogeneity
in the scalp hemodynamic fluctuations. Moreover, different tasks result in different global
physiological changes which can either mask the brain response or create false positive. For
low-density probe designs, a short-separation channel that measures only scalp
hemodynamics has been shown to be a robust solution to filtering the contaminating signals .
However, most current fNIRS systems and probe assemblies do not permit the inclusion of
short-separation channels. Because of the robustness of this method, we anticipate that

15
commercial instrumentation will be modified to permit short-separation measurements and
that such measurements will become standard in the next five years.

2.3.2.3 MOTION ARTIFACTS

As fNIRS application area broadens to different age and disease groups, motion
artifacts in the fNIRS signal is becoming an important challenge. The fNIRS signal is
susceptible to motion artifacts due to the subject movements that cause uncoupling of the
optodes and the scalp, resulting in either high-frequency spikes or baseline shifts. These
artifacts are commonly detected and removed or corrected before further processing. While
smoothing methods such as wavelet filtering are excellent in removing the sharp spikes, the
baseline shifts in the signal will remain after this type of filtering. Methods such as spline
interpolation or targeted principal component analysis (tPCA) are better at correcting baseline
shifts, but they, on the other hand, leave some residual after correcting high-frequency spikes.
There is clearly an opportunity to use hybrid methods that take advantage of different types
of correction algorithms. For instance, a hybrid method can be implemented that first corrects
for the baseline shifts and then uses a smoothing method to correct for the remaining spikes.
Moreover, more objective detection methods that rely only on the natural physiological
variations in the signal to set a threshold for motion artifact such as heart beat would be ideal.
It is expected that motion correction algorithms will mature and become routine over the next
five years.

2.3.2.4 ANATOMICAL GUIDANCE

fNIRS provides a measure of hemodynamic changes and does not provide structural
images that can be used to guide interpretation of the signals. The simplest way to obtain
structural guidance to permit, for instance, comparison of results across subjects and across
studies, is to localize the fNIRS measurement channels with respect to the 10-20 reference
points that were originally defined for electroencephalography. The relation between these
10-20 locations and the underlying cortical structure and the standard Montreal Neurological
Institute (MNI) stereotactic coordinates was originally worked out in the context of fNIRS by
Jurcak and further developed as recently reviewed in. These tools make it possible to
determine the MNI coordinates of fNIRS measurements to facilitate not only comparisons
across fNIRS studies, but also to compare fNIRS results with IMRI and PET studies that
routinely use MNI coordinates for reporting the locations of brain activation.

An alternative approach is to reconstruct an image of brain activation from the fNIRS

16
data using a head model as a spatial prior. This spatial prior has the dual advantage of
improving the image reconstruction, and of providing a functional image on the brain
structure, thus localizing activations to known anatomical regions. Importantly, these images
permit comparison across subjects as the images are in a consistent brain space. Software
analysis packages are becoming standardized , making these analyses available to the broader
community. What is needed in the coming years is extension of the head models from
"adults" to children and infants , as well as elderly adults.

2.4 APPLICATIONS
2.4.1 BRAIN DEVELOPMENT

Figure 2.9

fNIRS is ideal for neuroimaging in infants and children (Figure 3 a,b), as it does not require
subjects to be still, asleep or sedated, but instead allows them to interact freely with their
environment. Also, because of the thinner scalp and skull compared to adults, the fNIRS
cerebral sensitivity is great in infants. These advantages have led to the adoption of fNIRS for
a vast range of studies of both typical and atypical neurodevelopment), including the
development of object and face processing, number processing, language acquisition, social
communication, and neuromotor development. Studies of atypical functional development
have focused mainly on attention deficit hyperactivity disorder and autism spectrum disorder.

17
Most studies have demonstrated discrimination between groups, but individual risk
assessment and diagnosis remain challenging, even more so than in adults because
established standards in children at different ages are still lacking.

These studies foreshadow the crucial role fNIRS will play to deepen our
understanding of the developing brain. But practical challenges specific to infants and
children exist, including limited time to place and localize the probe, and the prevalence of
motion artifacts. These issues are being addressed by the development of improved infant-
specific probes. Optimal approaches and parameters for motion artifact identification and
correction cannot be directly exported from adult to infant studies, and need to be established
empirically in this population.

Another challenge resides in the data interpretation. For instance, brain atlases
developed for adults cannot be simply scaled down to infants' heads, as the brain undergoes
structural changes early in life. Fortunately, reference brain atlas for infants and children at
different ages are starting to appear. Finally, neurovascular coupling is still developing in the
young brain, and conflicting observations in infants have been reported with either
hemodynamic responses resembling those observed in adults, or instead displaying a negative
response. fNIRS signals in infants need to be interpreted carefully, as they may not directly
reflect neuronal activity, but instead the development of neurometabolic and neurovascular
systems.

2.4.2 APPLICATION TO THE COGNITIVE AND PSYCHOLOGICAL SCIENCE

Figure 2.10

18
 fNIRS is proving to be an invaluable tool in the cognitive and psychological sciences
because of the ease with which neural activity can be measured in natural settings with
portable and low-cost systems. Measurements of the prefrontal cortex are particularly easy
because of reduced interference from hair, but studies are and should measure from more
brain regions. Recent reviews have addressed the application to emotional processing,
mobility and aging, clinical psychology, psychiatry, integration with neuromodulation, and
brain computer interfaces. fNIRS complements fMRI because it permits rapid studies of
larger numbers of subjects and more frequent longitudinal measurements of particular
relevance for learning and treatment studies. We anticipate that ongoing development of
highly capable wearable fNIRS systems will greatly expand the repertoire of cognitive and
psychological studies.

2.4.3 BRAIN COMPUTER INTERFACE

Brain computer interface (BCI) is a growing field of research that aims to facilitate
human being's communication and interaction with the environment by directly measuring
and self-regulating the neuronal and/or hemodynamic activity in the brain. While EEG is the
most studied non-invasive BCI, we see more and more promising studies with fNIRS. In a
recent work, Chaudhary and colleagues have succeeded to reach above chance level accuracy
to a yes/no paradigm with patients who are in a complete locked-in state. Another interesting
study attempted to control a robot through mere motor imagery comparable to what has been
done with EEG [54]. This type of research can have a high impact on prosthetics such as
controlling a robot-arm to perform daily functions, controlling a wheelchair or the
environment without the need of a chip implant. Another future application would be to use
fNIRS-BCI in personalized augmented reality applications. These can span from subjective
preferences during daily life to personalized medicine. EEG and fNIRS can potentially
complement each other through combining EEG's ability to directly capture neural activity
with millisecond time scale, fNIRS's better spatial localization and its ability to measure slow
and integrated hemodynamic changes which are more representative of the brain states.

19
2.4.4 HYPERSCANNING

Figure 2.11

Studying the social brain ideally involves imaging socially interacting people in a
naturalistic environment. Hyperscanning is a technique that allows this type of research by
measuring brain activity simultaneously from two or more people during realtime interactions
(-Figure Jd,e). So far, various imaging modalities such as fMRI, MEG, EEG and fNIRS have
been used in hyperscanning studies. Among these, fNIRS and EEG are the most suitable
modalities as they provide the naturalistic environment that social interactions require, and
EEG-fNIRS combination during hyperscanning can give invaluable insights into the nature of
social interactions. With wearable fNIRS systems, the study of diseases strongly linked with
problems in social interactions as autism or depression; interactions such as student-teacher,
parent-child or patient-clinician; the brain correlates of changes in social interactions during
development; and differences due to gender or certain traits.

Social interactions typically involve both neuronal and systemic physiological

processes. Moreover, the type of signal processing methods used to analyze fNIRS
hyperscanning data such as correlation, coherence, Granger causality or transfer entropy are
more sensitive to superficial contamination in fNIRS data. It will be essential to filter
superficial contamination with, for instance, the short separation regression approach
described above.

20
2.5 fNRI Publication per year

Figure 2.12

(a) Schematic and (b) photograph of a high-density probe with 96 sources and 92
detectors yielding over 1200 overlapping measurement channels, (c) The HD-DOT
measurements, combined with anatomical light propagation modeling and reconstruction,
enable mapping with high resolution on the cortical surface of the hemodynamic response to
different cognitive brain tasks. The HD-DOT spatial response (blue) and the independently
measured fMRI BOLD response (yellow) show very good spatial overlap (green). Figure
panels a, b and c were modified with permission from (d-e-f) Wearable probe, (d) Photograph
of a modular wearable device consisting of 4 independent DOT modules each constructed
from 30 x 30 mm printed circuit board, 4 photodiodes and 2 dual-wavelength sources, (e)
Schematic representation of the wearable device positioned on the scalp over the primary
somatomotor cortices (f) Group-average mapping of the hemodynamic response to a finger
tapping task using the wearable probe presented above and image reconstruction based on a
subject-registered atlas. Figure panels d, e and f were modified with permission from.

21
 CHAPTER 3

CONCLUSION

fNIR spectroscopy is an emerging technology that uses near-infrared light to measure

changes in the concentration of oxygenated and deoxygenated hemoglobin in the cortex.
Although fNIRs imaging is limited to the outer cortex, it provides neuroimaging that is safe,
portable, and very affordable relative to other neuroimaging technologies. It is also relatively
robust to movement artifacts and can readily be integrated with other technologies such as
EEG.

fNIRs has been demonstrated to have adequate validity in the measurement of

functional brain activity during a variety of cognitive, emotional, and motor tasks and has the
potential to provide a flexible neuroimaging tool for clinicians and researchers alike. The
portable and wireless instrumentations, combined with robust analytic algorithms and end-
user software, make fNIR a viable option for the study of cognition- and emotion-related
hemodynamic changes in both adults and children, under either stationary or ambulant
conditions.

22