Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
15
,
No.
1
,
Febr
uary
20
25
, pp.
303
~
310
IS
S
N:
20
88
-
8708
, DO
I:
10
.11
591/ij
ece.v
15
i
1
.
pp
303
-
310
303
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om
Refinin
g
thyroid f
unctio
n eval
ua
ti
on:
a compar
ative
study
of
prepro
cessing
me
thods
in
diffus
e
reflect
ance
s
pectr
osc
opy
Wince
nt A
nto
Win S
ha
li
ni,
Thul
as
i
Raja
l
ak
sh
mi
, S
el
vana
ya
gam
Vas
an
t
hadev
Su
r
yaka
l
a
Dep
artmen
t of
Ele
ctron
ics an
d
Co
m
m
u
n
icatio
n
E
n
g
in
eering
,
Co
lleg
e of En
g
in
eering
and
T
echn
o
lo
g
y
,
SRM I
n
stitu
te of
S
cien
ce a
n
d
Techn
o
lo
g
y
,
Kattan
k
u
lath
u
r,
I
n
d
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
un
10, 2
024
Re
vised
A
ug 27, 2
024
Accepte
d
Se
p 3, 2
024
Thyroi
d
dysfun
ct
ion
,
co
mpri
sin
g
condi
t
ions
such
as
hyp
ert
hyr
oidi
sm
and
hypothyroi
dism
,
rep
r
ese
nts
a
subs
ta
ntial
globa
l
he
al
th
chall
enge
,
nec
essit
at
ing
timel
y
and
pr
e
ci
se
d
ia
gnosis
for
ef
fecti
ve
the
r
ape
ut
ic
int
erv
ent
ion
an
d
pa
ti
en
t
welf
ar
e.
Conv
ent
ion
al
dia
gnostic
mod
a
li
ties
ofte
n
invol
ve
inv
asive
proc
edur
es,
th
at
coul
d
ca
use
dis
com
fort
and
in
c
onveni
en
ce
for
indi
vidu
al
s.
The
non
-
i
nva
sive
technique
s
li
ke
d
iffuse
ref
lecta
n
ce
spec
troscopy
(
DRS
)
ca
n
off
er
a
prom
isin
g
alter
na
ti
v
e.
Thi
s
study
under
score
s
th
e
cri
t
ical
ro
le
of
pre
proc
essing
me
thods
in
enh
anc
ing
the
ac
cur
ac
y
of
thyr
oid
hor
mone
fun
ct
ion
al
it
y
throug
h
a
non
-
inv
asive
appr
o
ac
h
.
In
the
proposed
study
the
spe
ctral
da
ta
ac
qui
re
d
from
the
DRS
setup
are
subjec
t
ed
to
diff
ere
nt
pr
epr
oc
essing
te
chn
ique
s
t
o
i
mprove
th
e
e
ffic
a
cy
of
the
pre
d
ic
t
ion
mode
l
.
Thirty
i
ndivi
dual
s
wi
th
thyroi
d
dysfun
ct
ion
wer
e
inc
lud
ed
in
th
e
study,
and
pre
proc
essing
me
thods
such
as
base
l
ine
cor
recti
on
,
mu
ltipli
c
at
iv
e
sc
atte
r
cor
re
ction
(M
SC
),
and
st
and
ard
norm
al
var
iate
(
SNV
)
,
were
sys
te
mati
c
al
ly eva
lu
ated.
T
he
study hi
ghl
ig
hts t
hat
SNV
pre
proc
essing
o
utpe
rform
ed
oth
er
methods
with
a
root
mean
square
err
or
(RMS
E)
of
0.
00
5
and
an
R²
of
0
.
99
.
In
con
tra
st
,
MS
C
result
ed
in
an
RMS
E
of
0.
87
and
an
R
²
of
0.
86
,
while
base
li
n
e
cor
re
ct
i
on
show
ed
a
R
MS
E
of
0
.
84
and
an
unusual
R²
of
1.
09,
indi
c
at
ing
pot
ential
issues.
SN
V
prove
d
to
be
th
e
most
eff
ec
t
ive t
e
chni
que
.
Ke
yw
or
d
s
:
Ba
sel
ine
corre
ct
ion
D
i
f
f
u
s
e
r
e
f
l
e
c
t
a
n
c
e
s
p
e
c
t
r
o
s
c
op
y
Mult
ipli
cat
ive
scat
te
r
co
rr
ec
ti
on
Pr
e
processin
g
t
echn
i
qu
e
Stand
a
r
d norm
al
v
ariat
e
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
BY
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Th
ulasi R
aj
al
akshmi
Dep
a
rtme
nt of
Ele
ct
ro
nics
and C
om
m
unic
at
ion
En
gin
ee
rin
g,
C
ollege
of E
ng
i
neer
i
ng and
Tech
nolo
gy
,
SRM In
sti
tute of Sci
ence
and
Tech
no
l
ogy
Katt
ankulat
hur
-
60
3203,
Ta
mil
N
ad
u
,
In
dia
Emai
l:
r
aj
al
akt@srmi
st.ed
u.
in
1.
INTROD
U
CTION
Pr
e
-
proces
sin
g
of sp
ect
ral d
at
a
is
c
ru
ci
al
for ac
hieving
reli
a
ble o
utc
om
es
. P
re
processin
g met
hods
are
cru
ci
al
f
or
m
odel
perf
or
m
an
ce,
as
s
pectra
can
be
af
fecte
d
by
va
rio
us
di
sturb
a
nces
th
a
t
impact
meas
ureme
nt
accurac
y
[
1]
–
[
4]
.
M
aj
or
in
flu
ences
inclu
de
measu
rin
g
ge
ome
try
-
s
uch
as
sample
thick
ne
ss,
detect
or
dis
ta
nce,
con
ta
ct
press
ure,
an
d
li
gh
t
s
ource
a
ng
le
[5],
[6]
.
Eli
minati
ng
scat
te
rin
g
ef
f
ect
s
fr
om
di
ff
e
ren
tl
y
siz
e
d
pa
rtic
le
s
is
al
so
esse
ntial
in
pre
proce
ssing.
This
discuss
io
n
will
c
on
ce
ntrate
on
the
pr
e
-
proces
sing
of
data
obta
ined
from dif
fuse re
flect
ance s
pect
ro
sc
opy f
or no
n
-
i
nv
asi
ve
th
yr
oid
hor
mone
f
un
ct
io
ning
ass
essment.
Diff
e
re
nt
sp
ect
ro
sc
opic
meth
od
s
e
nc
ounter
sp
eci
fic
chall
e
ng
e
s.
Nea
r
-
inf
r
ared
s
pectr
os
c
opy
ty
pical
ly
con
te
nds
with
co
ns
ist
ent
or
li
near
s
hifts
i
n
the
baseli
ne
du
e
to
scat
te
re
d
li
ght,
Ra
ma
n
s
pectr
os
c
opy
oft
en
exh
i
bits
poly
nomial
bac
kgr
ounds
from
fl
uoresce
nce,
an
d
mid
-
inf
rar
e
d
sp
ect
ra
a
re
a
f
f
ect
ed
by
va
riat
ion
s
i
n
sample
thick
ne
ss
[7],
[
8]
.
T
he
pu
rpose
of
preprocessi
ng
is
to
rem
ov
e
t
hes
e
inter
fer
e
nces
w
hile
retai
ning
t
he
crit
ic
al
info
r
m
at
ion
withi
n
th
e
sp
ect
r
um
.
Di
ffuse
re
flect
an
ce
sp
ect
r
os
c
opy
(
DRS)
has
prov
e
n
to
be
a
va
luable
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
303
-
310
304
asset
in
me
dic
al
diag
no
sti
cs.
The
ac
c
urat
e
interp
retat
ion
of
sp
ect
ral
dat
a
from
DRS
is
highly
de
pe
ndent
o
n
eff
ic
ie
nt
pr
e
pr
ocessin
g
te
c
hniqu
es
,
s
uch
as
mu
lt
ipli
cat
ive
scat
te
r
co
rr
ect
i
on
(
M
SC
),
sta
nd
a
r
d
no
rmal
var
ia
te
(S
N
V
),
an
d
ba
sel
ine
c
orrecti
on.
These
methods
are
c
ru
ci
al
f
or
a
ddressi
ng
bas
el
ine
s
hi
fts
an
d
var
ia
ti
on
s
i
n
diffuse
r
e
flect
ance s
pectra,
e
nsuri
ng that th
e
data is
prop
e
rl
y normali
zed
a
nd r
ea
dy fo
r
in
-
de
pth anal
ys
is
.
M
ulti
plica
ti
ve
sign
al
c
orrecti
on
(
M
SC
)
a
ddresses
majo
r
e
ff
ect
s
by
def
i
ni
ng
a
re
fer
e
nc
e
sp
ect
r
um
us
ua
ll
y
t
he
me
an
of
t
he
cal
ib
rati
on
data
a
nd
the
c
orrecti
ng
s
pectra
f
or
ba
sel
ine
a
nd
mu
l
ti
plica
ti
ve
scat
te
ring
eff
ect
s,
al
ign
i
ng
with
the
K
ub
el
ka
–
Munk
t
he
ory
[9],
[
10]
.
S
NV
rem
oves
c
on
sta
nt
offset
te
rms
by
s
ub
t
ra
ct
ing
the
s
pectr
um’s
mean
a
nd
scal
i
ng
by
it
s
sta
ndard
dev
ia
ti
on,
makin
g
it
a
popu
la
r
meth
od
f
or
it
s
simpli
ci
ty
[
11]
.
SNV
a
nd
MSC
of
te
n
pro
du
ce
simi
la
r,
inte
rc
hangea
ble
res
ul
ts
[12],
[
13]
.
The
im
portan
c
e
of
pre
process
ing
i
n
DRS:
pr
e
proc
essing
te
c
hn
i
ques
li
ke
MSC
,
S
N
V,
a
nd
baseli
ne
co
rr
e
ct
ion
pla
y
a
fun
dame
ntal
r
ole
i
n
enh
a
ncin
g
t
he
qual
it
y
of
s
pe
ct
ral
data.
MSC
is
us
e
d
to
correct
f
or
sc
at
te
ring
e
ff
ect
s
that
can
dist
or
t
the
sp
ect
ra.
I
t a
djust
s the spect
ra
by ali
gn
i
ng the
m to
a
refe
re
nc
e sp
ect
r
um
, wh
ic
h
minimi
zes
var
ia
ti
ons ca
use
d by
par
ti
cl
e
siz
e,
s
hap
e
,
a
nd
othe
r
ph
ys
ic
al
pro
per
ti
es
of
t
he
sample.
SNV
i
s
an
oth
e
r
te
ch
nique
that
nor
mali
zes
each
s
pectr
um
by
rem
ovin
g
scat
te
r
eff
ect
s
and
ce
nteri
ng
the
data
ar
ou
nd
zer
o.
It
is
pa
rtic
ularly
us
e
fu
l
f
or
d
eal
ing
with
m
ulti
plica
ti
ve
interfe
ren
ces
.
Ba
sel
ine
co
rr
ect
i
on
ad
dresses
a
ny
sh
i
fts
or
dr
i
fts
in
t
he
basel
ine
of
the
sp
ect
ra
,
w
hich
ca
n
res
ul
t
fr
om
instr
ume
nt
va
riat
ions
or
sa
mp
le
inco
ns
ist
encies
.
By
correct
in
g
these
baseli
ne
iss
ues
, th
e
sp
ect
ra
be
come mo
re
c
ompa
rab
le
a
nd
r
el
ia
ble f
or
furt
her anal
ys
is
[
14]
.
The
SNV
te
c
hn
i
qu
e
wa
s
m
et
ic
ulo
us
l
y
im
plemente
d
t
o
sign
ific
a
ntly
di
minish
t
he
m
ulti
pli
cat
ive
interfe
ren
ce
re
su
lt
ing
from
sc
at
te
r.
This
a
ppr
oach
in
vo
l
ved
su
bt
racti
ng
the
mean
val
ue
of
the
e
ntire
sp
e
c
trum,
eff
ect
ivel
y
re
movin
g
c
onsta
nt
off
set
te
rms.
A
ddit
ion
al
ly
,
it
normali
ze
d
the
scal
e
of
a
ll
sp
ect
ra
by
di
vid
in
g
each
s
pectrum
by
t
he
sta
ndar
d
de
viati
on
of
the
com
plete
s
pectr
um
[
15]
.
M
SC
a
nd
SNV
are
fr
e
quent
ly
us
e
d
intercha
ngeabl
y,
pro
du
ci
ng
resu
lt
s
that
a
r
e
typ
ic
al
ly
si
mil
ar
[16
]
.
S
NV
is
disti
nguish
e
d
as
a
prefe
rr
e
d
pr
e
processi
ng
method,
know
n
f
or
it
s
strai
ghtf
orward
ne
ss
and
e
ff
ic
ac
y
[
17]
.
M
SC
,
an
d
S
NV
e
nh
an
ce
th
e
pr
e
dicti
ve
cap
abili
ti
es
of
sp
e
ct
ro
sc
op
ic
a
nalyses.
T
hese
preprocessi
ng
m
et
hods
en
sure
that
the
sp
ect
r
al
data
us
e
d
in
pr
e
dicti
ve
m
odel
s
a
re
accu
rate
a
nd
r
el
ia
ble,
le
adi
ng
to
bette
r
cl
inic
al
outc
om
e
s
[18]
.
DRS
has
prov
e
n
us
ef
u
l
in
ot
her
medical
a
reas
.
F
or
instance
,
it
has
bee
n
e
mp
lo
ye
d
in
th
e
diag
nosis
of
br
east
le
si
on
s
and
the
assessme
nt
of
tumor
mar
gi
ns
du
rin
g
s
urge
ries
highli
gh
ti
ng
it
s
a
bili
ty
to
pro
vid
e
real
-
ti
me
feedbac
k
duri
ng
su
r
gical
proce
dures
[
19]
.
I
n
summa
ry,
dif
fu
se
re
flect
anc
e
sp
e
ct
rosco
py,
w
he
n
c
oupl
ed
with
ap
pro
pr
ia
te
pr
e
processi
ng
te
chn
iq
ues
li
ke
M
SC,
S
NV,
an
d
ba
sel
ine
correct
ion,
hol
ds
sig
nificant
promise
in
me
dical
diag
nosti
cs, in
cl
ud
in
g
t
hyr
oid assessme
nt.
2.
METHO
D
2.1.
Dataset
A
ra
ndomi
zed
study
was
co
nducte
d
du
rin
g
this
exami
nation
t
o
colle
ct
r
eal
-
ti
me
sp
ect
r
um
si
gn
al
s.
With
fo
li
o
num
ber
84
62
/I
EC/
2022
se
r
ving
as
pr
oof,
the
SR
M
M
e
dical
Col
le
ge
Hospita
l
a
nd
Re
searc
h
Ce
nter’s
Ethic
al
Com
mit
te
e
gr
a
nted
the
neces
sar
y
et
hi
cal
cl
earance
f
or
this
st
udy.
Thirty
vo
l
un
te
ers
(N
=
30)
bot
h
male
and
female,
a
ge
d
ei
ghte
en
an
d
up,
who
hav
e
re
gula
r
cl
inica
l
visit
s
to
maint
ai
n
t
hyro
i
d
hor
mone
imbala
nc
e
we
re
include
d
i
n
t
he
stu
dy.
2.2.
Ex
peri
menta
l
setup o
f
di
ff
u
s
e reflec
t
an
ce
s
pectr
os
co
py
The
ex
per
im
en
ta
l
config
ur
at
i
on
for
DRS
is
sh
ow
n
in
Fig
ure
1.
It
e
ncom
passes
a
T
ungst
en
Halo
ge
n
li
gh
t
sou
rce
(L
S
-
1)
s
pecifica
ll
y
ta
il
or
e
d
for
the
visi
ble
nea
r
-
inf
rar
e
d
(NIR
)
wa
velen
gth
r
ang
e
,
spa
nnin
g
from
360
t
o
2
,
500
nm
.
A
dd
it
io
na
ll
y,
the
set
up
featu
res
a
s
pe
ct
ro
mete
r
(USB
4
0
0
0
)
e
q
u
i
p
p
e
d
w
i
t
h
i
n
t
e
r
f
a
c
e
c
a
p
a
b
i
l
i
t
i
e
s
a
n
d
h
i
g
h
-
s
p
e
e
d
e
l
e
c
t
r
o
n
i
c
s
.
T
h
e
U
S
B
4
0
0
0
s
h
o
w
c
a
s
e
s
r
e
s
p
o
n
s
i
v
e
n
e
s
s
within
the
wa
velen
gth
range
of 36
0
to
1
,
10
0 nm
.
Figure
2
s
how
s
the
r
eal
-
ti
me
DRS
set
up.
T
his
sect
io
n
ou
t
li
nes
the
detai
le
d
set
up,
co
m
pone
nts,
a
nd
proce
dures
i
nvolv
e
d
in
the
D
RS meas
ur
e
me
nts.
T
he prima
r
y
li
ght so
urce
us
e
d
in t
he DR
S setu
p
is a
Tu
ng
ste
n
halo
gen
la
mp
(
LS
-
1).
This
li
ght
sourc
e
is
spe
ci
fical
ly
ch
ose
n
f
or
it
s
abili
ty
to
e
mit
a
bro
ad
s
pe
ct
r
um
of
li
gh
t,
cov
e
rin
g
both
the
visible
a
nd
NI
R
wa
vel
eng
t
h
ra
ng
es
.
The
e
missi
on
sp
ect
r
um
of
t
he
L
S
-
1
sp
a
ns
f
rom
360
t
o
2
,
500
nm,
ma
king
it
i
deal
f
or
capt
uri
ng
a
wi
de
ra
nge
of
opti
cal
pro
per
ti
es
from
the
ti
ssu
e
.
A
US
B
4000
s
pectr
ome
te
r,
e
qu
i
pp
e
d
with
hig
h
-
sp
e
ed
el
ect
r
on
ic
s
and
a
c
ompu
te
r
inter
face,
is
us
e
d
to
capt
ure
the
diffusel
y
ref
le
ct
ed
li
ght
fro
m
t
he
ti
ss
ue.
The
U
SB
4000
is
respo
ns
iv
e
acr
os
s
a
wavel
eng
th
range
of
360
t
o
1
,
100
nm.
T
his
s
pectr
om
et
er
is
sel
ect
ed
f
or
it
s
abili
ty
to
pro
vid
e
hi
gh
-
r
esolutio
n
s
pectral
data
quic
kl
y
a
nd
accuratel
y.
T
he
sp
ect
ral
data
colle
ct
ed
by
t
he
US
B
4000
are
esse
ntial
f
or
a
nalyzin
g
ho
w
li
ght
inte
rac
ts
with
the
ti
ssu
e,
in
cl
ud
in
g
a
bsor
ption,
scat
te
ring,
an
d
re
fle
ct
ion
pro
pe
rtie
s,
w
hich
a
re
ind
ic
at
ive
of
ti
ssu
e
com
posit
ion
a
nd
healt
h.
T
he
li
gh
t
emit
te
d
from
the
T
ungst
en
Halo
ge
n
so
urce
is
tra
nsmi
tt
ed
to
the
t
issue
thr
ough
a
s
pe
ci
al
iz
ed
fib
er
op
ti
c
re
flect
an
ce
pr
ob
e
(R40
0).
T
he
R4
00
pro
be
is
e
qu
i
pped
with
bif
ur
cat
ed
op
ti
cal
fib
ers
arr
a
ng
e
d
i
n
a
sp
eci
fic
c
onf
igurat
ion
t
o
opti
mize
li
gh
t
deliver
y
a
nd
colle
ct
ion
.
Th
e
pro
be
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Ref
inin
g
thy
roi
d
fu
nction ev
al
ua
ti
on:
a co
m
pa
r
ative
stu
dy of p
repr
ocessi
ng
…
(
Wi
nce
nt An
t
o
Wi
n S
hali
ni
)
305
con
ta
in
s
7
op
t
ic
al
fiber
s,
eac
h
with
a
n
in
ne
r
diamet
er
of
400
micr
om
et
ers.
On
e
fiber
is
po
sit
io
ned
at
the
center,
surr
ounded
by
six
othe
r
fi
ber
s
.
T
he
six
s
urrou
nd
i
ng
fi
bers
are
re
sp
onsi
ble
f
or
de
li
ver
in
g
li
ght
to
th
e
ti
ssu
e,
wh
il
e th
e central
fibe
r c
ollec
ts t
he dif
fu
sel
y reflect
e
d
li
ght.
Figure
1.
DRS
c
o
n
f
i
g
u
r
a
t
i
o
n
F
i
g
u
r
e
2
.
R
e
a
l
-
t
i
m
e
D
R
S
s
e
t
up
Accurat
e D
RS
measu
reme
nts
require ca
reful
cal
ibrati
on
of the s
ys
te
m to
e
ns
ure that
th
e s
pectral data
accuratel
y
ref
l
ect
the
ti
ssu
e's
pro
per
ti
es.
B
ari
um
Su
l
phate
(
Ba
SO
₄)
is
us
e
d
as
a
ref
le
ct
an
c
e
sta
nd
a
rd.
Ba
SO
₄
is
chosen
for
it
s
high
re
flect
anc
e
of
a
ppr
ox
im
at
el
y
99%,
ma
king
it
an
idea
l
ref
ere
nce
ma
te
rial
.
The
li
gh
t
fr
o
m
the
Tu
ngste
n
halo
gen
sourc
e
is
directed
on
t
o
the
Ba
S
O₄
surfa
ce.
T
he
re
flect
ed
li
gh
t
from
the
Ba
SO
₄
i
s
captu
red
by
the
s
pectr
om
et
er
to
gen
e
rate
a
ref
e
ren
ce
sp
ect
r
um
.
T
hi
s
re
fer
e
nce
spe
ct
ru
m
se
r
ves
as
a
ben
c
hm
a
r
k
a
ga
inst w
hich
ti
ssu
e s
pectra a
re
com
par
e
d.
To
acc
ount
f
or
ambie
nt
li
ght
and
el
ect
ronic
no
ise
,
a
da
rk
s
pectr
um
is
acq
uire
d
by
bl
oc
kin
g
the
li
ght
so
urce.
T
his
s
te
p
e
nsures
th
at
an
y
non
-
si
gn
al
-
relat
ed
c
ompone
nts
a
re
re
move
d
fro
m
the
s
pectr
a
l
data,
enh
a
ncin
g
the
accu
rac
y
of
t
he
s
ubse
qu
e
nt
ti
ssu
e
meas
urements.
Af
te
r
acq
uiri
ng
the
re
fer
e
nce
an
d
da
r
k
sp
ect
ra,
the
sy
s
te
m is set t
o re
flect
ance m
ode
u
si
ng the
Sp
ec
tra S
uite so
ftw
are.
I
n
t
h
i
s
s
t
u
d
y
,
p
a
r
t
i
c
i
p
a
n
t
s
w
e
r
e
i
n
i
t
i
a
l
l
y
b
r
i
e
f
e
d
o
n
t
h
e
n
o
n
-
in
v
a
s
i
v
e
a
p
p
r
o
a
c
h
a
n
d
s
a
f
e
u
t
i
l
i
z
a
t
i
o
n
o
f
n
e
a
r
-
i
n
f
r
a
r
e
d
l
i
g
h
t
(
N
I
R
)
o
n
t
h
e
n
e
c
k
r
e
g
i
o
n
.
T
h
e
m
e
a
s
u
r
e
m
e
n
t
s
i
t
e
a
n
d
p
r
o
b
e
t
i
p
w
e
r
e
c
l
e
a
n
s
e
d
w
i
t
h
a
n
a
l
c
o
h
o
l
-
b
a
s
e
d
s
o
l
u
t
i
o
n
t
o
e
n
s
u
r
e
a
c
c
u
r
a
t
e
s
p
e
c
t
r
a
l
r
e
a
d
i
n
g
s
a
n
d
p
a
r
t
i
c
i
p
a
n
t
s
w
e
r
e
s
a
f
e
f
r
o
m
i
n
f
e
c
t
i
o
n
s
.
T
h
e
f
i
b
e
r
o
p
t
i
c
p
r
o
b
e
i
s
p
o
s
i
t
i
o
n
e
d
o
n
t
h
e
n
e
c
k
.
T
h
e
l
i
g
h
t
f
r
o
m
t
h
e
T
u
n
g
s
t
e
n
h
a
l
o
g
e
n
s
o
u
r
c
e
i
s
t
r
a
n
s
m
i
t
t
e
d
t
h
r
o
u
g
h
t
h
e
p
r
o
b
e
a
n
d
o
n
t
o
t
h
e
t
i
s
s
u
e
.
T
h
e
r
e
f
l
e
c
t
e
d
l
i
g
h
t
,
c
a
r
r
y
i
n
g
i
n
f
o
r
m
a
t
i
o
n
a
b
o
u
t
t
h
e
t
i
s
s
u
e
’
s
o
p
t
i
c
a
l
p
r
o
p
e
r
t
i
e
s
,
i
s
c
a
p
t
u
r
e
d
b
y
t
h
e
c
e
n
t
r
a
l
f
i
b
e
r
a
n
d
t
r
a
n
s
m
i
t
t
e
d
b
a
c
k
t
o
t
h
e
s
p
e
c
t
r
o
m
e
t
e
r
.
T
h
e
S
p
e
c
t
r
a
S
u
i
t
e
s
o
f
t
w
a
r
e
i
s
u
s
e
d
t
o
c
a
p
t
u
r
e
a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
303
-
310
306
a
n
a
l
y
z
e
s
p
e
c
t
r
a
l
d
a
t
a
i
n
r
e
a
l
t
i
m
e
.
T
h
e
c
o
l
l
e
c
t
e
d
d
a
t
a
a
r
e
i
m
m
e
d
i
a
t
e
l
y
t
r
a
n
s
f
e
r
r
e
d
t
o
a
c
o
m
p
u
t
e
r
c
o
n
n
e
c
t
e
d
t
o
t
h
e
s
p
e
c
t
r
o
m
e
t
e
r
,
e
n
s
u
r
i
n
g
t
h
e
i
n
t
e
g
r
i
t
y
a
n
d
a
c
c
u
r
a
c
y
o
f
t
h
e
m
e
a
s
u
r
e
m
e
n
t
s
.
2.3.
Spec
tr
al
p
r
e
p
r
o
c
e
s
s
i
n
g
Sp
ect
ral
pr
e
pro
cessi
ng
te
c
hniq
ues
are
util
iz
ed
mathemat
ic
al
ly
to
e
nh
a
nce
s
pectral
data.
T
he
obje
ct
ive
is
to
recti
f
y
un
wan
te
d
in
flue
nc
es
li
ke
un
pr
e
di
ct
able
no
ise
,
va
riat
ion
s
i
n
li
ght
pat
h
le
ngth,
and
li
gh
t
scat
te
rin
g
du
e
to
div
e
rse
phys
ic
al
pro
pe
rtie
s
of
sam
ple
s
or
instr
um
e
nt
-
relat
ed
facto
r
s.
T
his
sta
ge
i
s
ty
pical
ly
e
xe
cuted
befor
e
e
mp
l
oying
m
ulti
var
ia
t
e
m
odel
ing,
ai
ming
t
o
mit
igate
,
re
move,
or
sta
nd
a
rd
iz
e
the
se
in
flue
nces
on
t
he
sp
ect
ra,
t
her
e
by
sig
nificant
ly
im
pro
ving
the
reli
abili
ty
of
the
cal
ib
rati
on
m
od
el
[
20]
.
I
n
this
stu
dy,
th
ree
sp
ect
ral
pr
e
pro
cessi
ng app
ro
a
ches a
re c
omp
arati
vely
e
xplo
red
:
SNV, M
S
C, an
d
baseli
ne
correcti
on
.
2.3.1
.
Mul
tipli
cat
i
ve
sc
atter
c
o
r
r
e
c
t
i
o
n
M
ulti
plica
ti
ve
scat
te
r
co
rr
ect
i
on
(
M
SC
)
is
a
r
obus
t
te
c
hniqu
e
us
e
d
t
o
address
scat
te
r
eff
ect
s
in
sp
ect
ral
data,
wh
ic
h
arise
du
e
to
va
riat
ion
s
in
pa
rtic
le
siz
e,
surface
te
xt
ur
e
,
an
d
oth
e
r
phys
ic
al
pr
op
e
rtie
s
of
the
sam
ple.
T
he
se
va
riat
ion
s
can
distor
t
t
he
li
gh
t
pat
h
a
nd
i
ntensity
,
le
adi
ng
to
inacc
ur
aci
es
in
t
he
data.
M
SC
regresses
eac
h
sp
ect
r
um
agai
nst
a
ref
e
ren
ce
s
pectr
um
an
d
c
orrects
us
in
g
the
sl
ope
a
nd
i
nterce
pt
of
the
li
near
fit.
This
mini
mize
s
baseli
ne
offsets
a
nd
mu
lt
ipli
cat
ive
eff
ect
s
[
21]
.
The
proces
s
of
M
SC
begi
ns
with
cal
culat
ing
the
mean
s
pectr
um
from
t
he
e
ntire
cal
ibrati
on
set
.
T
his
me
an
s
pectrum
a
ct
s
as
the
re
fere
nce
sp
ect
r
um
.
F
or
each
s
pectr
um
(
),
a
li
nea
r
re
gr
essi
on
is
pe
r
forme
d
a
gain
st
the
mea
n
s
pe
ct
ru
m
to
dete
r
mine
the slo
pe (
)
an
d
int
e
rcep
t
(
).
The reg
ressi
on model is e
xpre
ssed
as
(
1)
,
=
+
+
,
(
1)
wh
e
re
re
pr
es
ents
t
he
e
rro
r
te
rm
t
hat
i
nclud
e
s
t
he
act
ua
l
inf
ormat
io
n.
On
ce
th
e
i
nter
cept
a
nd
sl
op
e
are
determi
ned, th
e co
rr
ect
e
d
s
pe
ct
ru
m
(
,
)
is ob
ta
i
ned usi
ng
(
2).
(
,
)
=
−
,
(2)
This
c
orrecti
on
proces
s
rem
ov
e
s
both
m
ulti
plica
ti
ve
an
d
ad
diti
ve
scat
te
r
e
ff
ect
s
,
nor
mali
zi
ng
the
s
pectra
t
o
the mea
n
s
pect
rum a
nd ef
fecti
vely re
du
ci
ng
baseli
ne
s
hifts
and m
ulti
plica
ti
ve
va
riat
ions.
2.3.2.
Stand
ar
d
no
rm
al
v
a
r
i
a
t
e
Stand
a
r
d
no
r
mal
var
ia
te
(
SNV)
is
a
n
a
dd
it
io
nal
prep
ro
ces
sin
g
te
ch
nique
that
nor
mali
zes
each
sp
ect
r
um
in
de
pende
ntly
t
o
e
li
minate
mu
lt
ipli
cat
ive
scat
te
r
e
ff
ect
s
a
nd
a
dju
sts
f
or
bas
e
li
ne
va
riat
ion
s
.
SNV
ind
ivi
du
al
l
y
ce
nters
a
nd scale
s each
s
pectr
um b
y
s
ubtract
ing t
he
m
ean a
nd
div
idi
ng
by t
he
sta
nda
rd
de
vi
at
ion
.
This
c
orrects
a
d
d
i
t
i
v
e
and
mul
ti
plica
ti
ve
effe
ct
s
[
22]
.
T
he
SNV
proces
s
i
nvolv
e
s
cal
c
ulati
ng
t
he
mea
n
(
̅
̅
̅
̅
)
and stan
dard
de
viati
on
(
)
f
or
each s
pectr
um
(
).
T
he
mea
n
is
calc
ulate
d
as
(
3)
,
̅
=
1
∑
,
=
1
(
3)
and
the
sta
ndar
d dev
ia
ti
on is c
al
culat
ed
as
(
4)
,
=
√
1
−
1
∑
(
−
̅
)
2
=
1
,
(4)
wh
e
re
n
is
the
numb
e
r
of
dat
a
po
i
nts
in
t
he
sp
ect
r
um.
Eac
h
data
point
in
the
sp
ect
rum
is
then
sta
nd
a
r
dized
us
in
g
(5).
=
−
̅
,
(
5)
This
tra
ns
f
orm
at
ion
re
su
lt
s
in
sp
ect
ra
t
hat
ha
ve
zer
o
mean
an
d
unit
va
ria
nce,
w
hich
reduces
the
in
flue
nce
of
scat
te
r
a
nd
e
nhance
s
t
he
spe
ct
ral
feat
ur
es
.
By
m
aki
ng
t
he
sp
ect
ra
i
ndepende
nt
of
t
he
or
igi
nal
scal
e
a
nd
sample set c
ha
racteri
sti
cs, SN
V
e
ns
ures t
hat
the d
at
a is
m
ore co
ns
ist
ent a
nd easie
r
to
i
nte
rpret.
2.3.3.
B
as
el
ine
c
o
r
r
e
c
t
i
o
n
Ba
sel
ine
co
rr
e
ct
ion
is
a
vital
prep
ro
ce
ssin
g
meth
od
in
sp
e
ct
ro
sc
opy
t
hat
sign
ific
a
ntly
e
nh
a
nces
th
e
qu
al
it
y
a
nd
pr
eci
sion
of
dat
a
analysis
.
It
i
s
necessa
r
y
to
el
imi
nate
sp
e
ct
ral
arti
facts
that
ma
y
res
ult
from
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Ref
inin
g
thy
roi
d
fu
nction ev
al
ua
ti
on:
a co
m
pa
r
ative
stu
dy of p
repr
ocessi
ng
…
(
Wi
nce
nt An
t
o
Wi
n S
hali
ni
)
307
factors
s
uch
a
s
el
ect
ronic
in
te
rf
ere
nce,
ins
uff
ic
ie
nt
di
gital
filt
ering
,
or
incomple
te
di
gital
samplin
g
[23
]
.
Ba
sel
ine
corre
ct
ion
is
a
prep
ro
ces
sin
g
te
ch
nique
use
d
to
rem
ov
e
baseli
ne
dri
fts
a
nd
ba
ckgr
ound
noi
se
from
sp
ect
ral
data.
These
bas
el
ine
dri
fts
ca
n
be
c
au
se
d
by
in
strume
nt
i
ns
ta
bili
ty,
en
vir
onme
ntal
cha
ng
e
s,
or
sample
inco
ns
ist
encies
,
an
d
can
ob
sc
ure
the
sp
ect
ral
featur
e
s
of
interest
. T
he
pr
oce
ss
of
baseli
ne
c
orrecti
on starts
w
it
h
identif
ying
the
baseli
ne
of
t
he
sp
ect
r
um
us
ing
an
ap
pro
pr
i
at
e
meth
od,
s
uc
h
as
poly
nom
ia
l
fitt
ing
,
m
o
v
i
n
g
a
v
e
r
a
g
e
,
o
r
o
t
h
e
r
b
a
s
e
l
i
n
e
f
i
t
t
i
n
g
a
l
g
o
r
i
t
h
m
s
.
T
h
e
d
e
t
e
r
m
i
n
e
d
b
a
s
e
l
i
n
e
i
s
s
u
b
s
e
q
u
e
n
t
l
y
s
u
b
t
r
a
c
t
e
d
f
r
o
m
t
h
e
o
r
i
g
i
n
a
l
s
p
e
c
t
r
um
t
o
d
e
r
i
v
e
t
h
e
c
o
r
r
e
c
t
e
d
s
p
e
c
t
r
u
m
,
r
e
p
r
e
s
e
n
t
e
d
a
s
(6)
,
=
−
(
)
.
(6)
h
ere
,
is
t
h
e
ori
gi
nal
s
pec
trum,
a
nd
is
the
sp
ect
rum
a
fter
baseli
ne
correct
ion.
T
hi
s
correct
ion
proc
ess
e
nsures
th
a
t
the
s
pectral
f
eat
ur
es
relat
e
d
to
the
c
hemic
al
co
mposit
io
n
are
not
obscu
r
ed
by
baseli
ne variat
i
on
s
, lea
ding to
more acc
ur
at
e
sp
ect
ral a
nalys
is.
2.4.
P
e
r
f
o
r
m
a
n
c
e
-
m
e
t
r
i
c
s
R
-
square
d
(
²
)
a
nd
r
oot
mea
n
sq
ua
re
er
ror
(
RMSE
)
a
re
t
he
m
os
t
us
ed
pe
rformance
m
et
rics
f
or
evaluati
ng
t
he
accu
racy
of
pr
e
dicti
ve
m
odel
s
in
s
pectr
os
c
op
ic
a
naly
s
is.
²
in
dicat
es
the
pro
portio
n
of
var
ia
bili
ty
in
t
he
de
pende
nt
var
ia
ble
that
c
an
be
e
xp
la
ine
d
by
t
he
in
de
pende
nt
var
ia
bl
es,
wh
e
reas
R
M
SE
measu
res
th
e
aver
a
ge
siz
e
of
the
predict
io
n
er
r
or
s
i
n
the
same
un
it
s
as
the
dep
e
ndent
var
ia
ble
[
24]
.
²
is
determi
ned
by
squa
rin
g
t
he
co
rr
el
at
io
n
be
tween
the
pr
edict
ed
a
nd
obser
ve
d
value
s,
with
a
valu
e
of
1
represe
nting
a
n
ide
al
fit
.
T
a
ble
1
s
hows
the
e
valuati
on
metri
cs
with
descr
i
ption.
T
he
c
ho
ic
e
of
s
pectral
pr
e
processi
ng
te
chn
iq
ue
can
sign
ific
a
ntly
i
mp
ac
t
t
hese
metri
cs
s
howi
ng
s
up
e
rio
r
pe
rformance
in
certai
n
app
li
cat
io
ns
[25]
.
Table
1
.
E
val
ua
ti
on
met
ric ta
ble
S.NO
Metr
ic
Fo
rm
u
la
Descripti
o
n
1
R
2
2
=
1
−
∑
(
−
̂
)
2
∑
(
−
̅
)
2
wh
ere
is th
e ac
tu
a
l valu
e,
̂
is
th
e predicted
valu
e,
an
d
̅
is
th
e m
ean o
f
the act
u
al valu
e.
2
RMSE
=
√
∑
(
−
̂
)
2
wh
ere
is th
e ac
tu
a
l valu
e,
̂
is
th
e predicted
valu
e and
n is th
e
n
u
m
b
er
o
f
sa
m
p
les
3.
RESU
LT
A
N
D DIS
CUSSI
ON
The
s
pectral
da
ta
acq
uired
th
rou
gh
dif
f
us
e
r
eflect
ance
s
pe
ct
ro
sc
opy
c
ont
ai
ns
nonlinea
riti
es
that
can
impact
the
acc
ur
ac
y
of
predic
ti
ve
models.
T
o
a
ddress
t
his
issue,
th
ree
pre
processi
ng
te
c
hn
i
qu
e
s
we
re
a
pp
li
ed
t
o
t
h
e
d
a
t
a
:
S
N
V
,
M
S
C
,
a
n
d
B
a
s
e
l
i
n
e
C
o
r
r
e
c
t
i
o
n
.
T
h
e
r
a
w
s
p
e
c
t
r
a
l
d
a
t
a
plo
t
f
or
30
pa
rtic
ipants
with
thyr
oid
dy
s
f
un
ct
io
n
is
pr
ese
nted
i
n
Figure
3
(
a
)
.
F
rom
the
fig
ur
e
it
is
inferred
that,
ar
ound
700
t
o
95
0
nm,
the
ref
le
ct
ance
int
ensity
re
mains
r
e
l
a
t
i
v
e
l
y
s
t
a
bl
e
a
n
d
l
o
w
.
B
e
y
o
n
d
9
5
0
n
m
,
t
h
e
r
e
is
an
i
n
c
r
e
a
s
e
in
n
o
i
s
e
a
n
d
v
a
r
i
a
b
i
l
i
t
y
in
the
r
e
f
l
e
c
t
a
n
c
e
i
n
t
e
n
s
i
t
y
amon
g
pa
rtic
ipants
.
So
me
s
pik
e
s
and
a
bru
pt
ch
a
ng
e
s
in
inte
ns
i
ty
are
ob
s
er
ved
ar
ound
1
,
00
0
nm
,
w
hich
may
ind
ic
at
e
inc
rea
sed
se
ns
it
ivit
y
or
va
riabil
it
y
in
that
reg
i
on.
T
he
relat
ively
sta
bl
e
ref
le
ct
ance
inten
sit
y
between
700
an
d
950
nm
sug
gests
that
the
par
ti
ci
pa
nts’
sp
ect
ra
l
respo
ns
es
a
re
c
on
sist
e
nt
in
thi
s
reg
i
on.
This
sta
bili
ty
is
of
te
n
desi
rab
le
in
sp
ect
ral
an
al
ysi
s
as
it
can
indi
cat
e
a
un
i
form res
ponse
to
t
he
li
ght a
cro
ss
parti
ci
pa
nts.
F
i
g
u
r
e
3
(
b
)
s
h
o
w
s
t
h
e
s
p
e
c
t
r
a
l
p
l
o
t
o
f
S
N
V
p
r
e
p
r
o
c
e
s
s
i
n
g
,
w
h
i
c
h
n
o
r
m
a
l
i
z
e
s
t
h
e
d
i
s
t
r
i
b
u
t
i
o
n
a
c
r
o
s
s
t
h
e
w
a
v
e
l
e
n
g
t
h
r
a
n
g
e
.
I
n
t
h
e
S
N
V
p
r
o
c
e
s
s
,
t
h
e
d
a
t
a
a
r
e
m
e
a
n
-
c
e
n
t
e
r
e
d
a
n
d
s
c
a
l
e
d
b
y
t
h
e
i
r
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
,
r
e
s
u
l
t
i
n
g
i
n
s
p
e
c
t
r
a
c
e
n
t
e
r
e
d
a
r
o
u
n
d
z
e
r
o
w
i
t
h
u
n
i
f
o
r
m
v
a
r
i
a
n
c
e
.
T
h
i
s
n
o
r
m
a
l
i
z
a
t
i
o
n
h
e
l
p
s
t
o
r
e
d
u
c
e
m
u
l
t
i
p
l
i
c
a
t
i
v
e
e
f
f
e
c
t
s
,
e
n
h
a
n
c
i
n
g
t
h
e
c
o
m
p
a
r
a
b
i
l
i
t
y
a
n
d
i
n
t
e
r
p
r
e
t
a
b
i
l
i
t
y
o
f
t
h
e
s
p
e
c
t
r
a
l
d
a
t
a
.
T
h
e
S
N
V
p
r
e
p
r
o
c
e
s
s
e
d
p
l
o
t
d
e
m
o
n
s
t
r
a
t
e
s
a
s
t
a
n
d
a
r
d
i
z
e
d
r
e
p
r
e
s
e
n
t
a
t
i
o
n
,
m
i
n
i
m
i
z
i
n
g
b
a
s
e
l
i
n
e
s
h
i
f
t
s
a
n
d
s
c
a
l
i
n
g
i
n
c
o
n
s
i
s
t
e
n
c
i
e
s
.
Figure
3
(
c
)
dis
plays
t
he
sp
ect
ral
plo
t
afte
r
ba
sel
ine
correct
ion
us
in
g
Sa
vi
tz
ky
-
G
olay
filt
erin
g.
This
te
chn
iq
ue
s
ucc
essfu
ll
y
el
imi
na
te
s
baseli
ne
ir
regularit
ie
s,
s
moothin
g
t
he
s
pectral
baseli
ne
for
a
m
ore
co
ns
ist
ent
and
sta
bili
zed
data
represe
ntati
on
.
T
he
c
orre
ct
ed
plo
t
re
veal
s
e
nh
a
nce
d
cl
a
r
it
y
of
s
pectral
f
eat
ur
es,
as
unw
anted
fluctuati
ons
an
d
distor
ti
ons
ar
e
mit
igate
d.
T
his
plo
t
highli
gh
ts
the
e
ff
ect
i
ven
e
ss
of
the
ba
sel
ine
co
rr
ect
i
on
in
impro
ving
the
qu
al
it
y
a
nd
i
nterpreta
bili
ty
of
sp
ect
ral
data.
Figure
3
(
d
)
pr
esents
the
s
pec
tral
plo
t
after
M
SC
pr
e
processi
ng.
M
SC
e
ff
ect
ive
ly
ad
dresses
sc
at
te
ring
e
ff
ect
s
,
res
ulti
ng
in
a
more
un
if
orm
sp
ect
r
um
ac
r
oss
the
entire
wa
velen
gth
ra
nge.
By
reducin
g
the
impact
of
scat
te
ring,
this
no
rmali
zat
ion
process
e
n
ha
nce
s
the
com
par
a
bili
ty
and inter
pr
et
a
bi
li
ty o
f
t
he
s
pe
ct
ral infor
mati
on.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
303
-
310
308
(a)
(b)
(c)
(d)
Figure
3. Pr
e
pr
ocesse
d
s
pectr
al
plo
ts
(a)
ra
w
sp
ect
ral
data
pl
ot
of
30
pa
rtic
ipants
,
(b)
S
NV
,
(c)
baseli
ne
cor
recti
on
,
a
n
d
(
d)
M
SC
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
omp E
ng
IS
S
N:
20
88
-
8708
Ref
inin
g
thy
roi
d
fu
nction ev
al
ua
ti
on:
a co
m
pa
r
ative
stu
dy of p
repr
ocessi
ng
…
(
Wi
nce
nt An
t
o
Wi
n S
hali
ni
)
309
T
h
e
p
r
o
p
o
s
e
d
p
r
e
p
r
o
c
e
s
s
i
n
g
t
e
c
h
n
i
q
u
e
s
w
e
r
e
s
t
a
t
i
s
t
i
c
a
l
l
y
a
n
a
l
y
z
e
d
b
a
s
e
d
o
n
t
h
e
p
e
r
f
o
r
m
a
n
c
e
m
e
t
r
i
c
s
a
s
s
h
o
w
n
i
n
T
a
b
l
e
2
,
i
t
i
s
c
l
e
a
r
t
h
a
t
S
N
V
p
r
e
p
r
o
c
e
s
s
i
n
g
a
p
p
r
o
a
c
h
s
i
g
n
i
f
i
c
a
nt
l
y
o
u
t
p
e
r
f
o
r
m
s
t
h
e
ot
h
e
r
m
e
t
h
o
d
s
i
n
t
e
r
m
s
o
f
m
o
d
e
l
a
c
c
u
r
a
c
y
a
n
d
reli
abili
ty
.
S
N
V
r
e
s
u
l
t
e
d
i
n
a
n
R
M
S
E
o
f
0
.
0
0
5
a
n
d
a
n
²
o
f
0
.
9
9
.
T
h
i
s
h
i
g
h
l
i
g
h
t
s
S
N
V
’
s
s
u
p
e
r
i
o
r
a
b
i
l
i
t
y
t
o
n
o
r
m
a
l
i
z
e
t
h
e
s
p
e
c
t
r
a
l
data
,
e
f
f
e
c
t
i
v
e
l
y
m
i
t
i
g
a
t
i
n
g
m
u
l
t
i
p
l
i
c
a
t
i
v
e
e
f
f
e
c
t
s
.
I
n
c
o
n
t
r
a
s
t
,
M
S
C
s
h
o
w
s
a
n
R
M
S
E
o
f
0
.
8
7
a
n
d
a
n
2
o
f
0
.
8
6
.
W
h
i
l
e
M
S
C
r
e
d
u
c
e
s
s
c
a
t
t
e
r
e
f
f
e
c
t
s
a
n
d
i
m
p
r
o
v
e
s
t
h
e
u
n
i
f
o
r
m
i
t
y
o
f
t
h
e
s
p
e
c
t
r
u
m
,
i
t
s
t
i
l
l
e
x
h
i
b
i
t
s
a
r
e
l
a
t
i
v
e
l
y
hi
g
h
e
r
r
o
r
m
a
r
g
i
n
a
n
d
e
x
p
l
a
i
n
s
o
n
l
y
8
6
p
e
r
c
e
n
t
o
f
t
h
e
v
a
r
i
a
n
c
e
,
m
a
ki
n
g
i
t
l
e
s
s
r
e
l
i
a
b
l
e
c
o
m
p
a
r
e
d
t
o
S
N
V
.
T
h
e
h
i
g
h
e
r
R
M
S
E
i
n
d
i
c
a
t
e
s
m
o
r
e
s
i
g
n
i
f
i
c
a
n
t
e
r
r
o
r
s
i
n
t
he
p
r
e
d
i
c
t
i
o
n
s
,
s
u
g
g
e
s
t
i
n
g
t
h
a
t
w
h
i
l
e
M
S
C
i
s
b
e
n
e
f
i
c
i
a
l
,
i
t
d
o
e
s
n
o
t
a
c
h
i
e
v
e
t
h
e
s
a
m
e
l
e
v
e
l
o
f
p
r
e
c
i
s
i
o
n
a
s
S
N
V
.
Ba
sel
ine
c
orre
ct
ion
us
in
g
Sa
vitzky
-
Go
la
y
f
il
te
ring
pr
ese
nt
s
an
R
M
SE
of
0.84
an
d
a
n
R²
of
1.0
9.
Althou
gh
the
RMSE
is
sli
g
ht
ly
lowe
r
tha
n
M
SC,
the
²
val
ue
e
xceed
i
ng
1
is
unusual
a
nd
points
to
pote
ntial
ov
e
rf
it
ti
ng
or
anomal
ie
s
in
the
m
od
el
’s
e
va
luati
on
proce
ss.
This
c
ou
l
d
imply
t
hat
w
hi
le
baseli
ne
co
rr
ect
io
n
eff
ect
ivel
y
sm
oo
t
hs
the
sp
ect
ral
data,
it
mi
ght
intr
oduce
a
r
ti
fact
s
or
i
ncon
sist
encies,
th
us
aff
ect
in
g
t
he
overall
model reli
abili
ty a
nd inter
pr
et
abili
ty.
Table
2
.
E
valu
at
ion
-
metri
c
of
pr
e
processi
ng
t
e
c
h
n
i
q
u
e
Prepro
cess
in
g
t
e
c
h
n
i
q
u
e
s
2
R
M
S
E
M
S
C
0
.
8
6
0
.
8
7
SNV
0
.
9
9
0
.
0
0
5
Bas
elin
e
c
o
r
r
e
c
t
i
o
n
1
.
0
9
0
.
8
4
4.
CONCL
US
I
O
N
The
stu
dy
sig
ni
fies
the
imp
or
t
ance
of p
re
pro
cessi
ng
i
n
s
pec
trosc
op
ic
a
naly
sis.
In
the p
r
opos
e
d
stu
dy,
the
sel
ect
io
n
of
e
ff
ic
ie
nt
pre
processi
ng
te
c
hn
i
qu
e
s
t
o
ob
t
ai
n
im
pro
ved
model
acc
ur
ac
y
has
bee
n
ev
al
uated.
The
st
udy
in
volve
d
var
io
us
pr
e
process
i
ng
te
chn
iq
ues
su
c
h
as
S
N
V,
MSC
an
d
ba
sel
ine
c
orrecti
on
ou
t
of
wh
ic
h
SNV
outpe
rformed
a
s
the
m
os
t
r
ob
us
t
te
ch
nique
,
sign
ific
a
ntly
i
mpro
ving
t
he
cl
arit
y,
c
ompa
rab
il
it
y,
and
i
nter
pr
et
a
bili
ty
of
sp
ect
r
al
data,
ther
eb
y
en
ha
ncin
g
diagnostic
accu
r
acy
f
or
th
yroid
dys
functi
on.
Wh
il
e
M
SC
a
nd
base
li
ne
correct
ion
hav
e
t
heir
me
rits,
they
do
not
matc
h
the
pe
rformance
of
SNV
in
this
c
on
te
xt.
The
fin
dings
e
mphasiz
e
the
ne
cessi
ty
f
or
me
ti
culou
s
sel
ect
ion
an
d
a
ppli
cat
ion
of
prep
ro
c
essing
te
ch
niques
to
ens
ur
e
high
-
qu
al
it
y
sp
ect
ral
data,
ulti
mate
ly
le
adi
ng
to
more
acc
ur
at
e
non
-
in
vasive
diag
nosti
cs
an
d
bette
r
patie
nt outc
ome
s.
A
C
K
N
O
W
L
E
D
G
E
M
E
N
T
T
h
e
a
u
t
h
o
r
s
a
r
e
t
h
a
n
k
f
u
l
t
o
O
p
t
i
c
a
l
W
i
r
e
l
e
s
s
C
o
m
m
u
n
i
c
a
t
i
o
n
L
a
b
,
D
e
pa
r
t
m
e
n
t
o
f
E
l
e
c
t
r
o
n
i
c
s
a
n
d
C
o
m
m
u
n
i
c
a
t
i
on
E
n
g
i
n
e
e
r
i
n
g
,
S
R
M
I
n
s
t
i
t
u
t
e
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
K
a
t
t
a
n
k
u
l
a
t
h
u
r
c
a
m
p
u
s
f
o
r
p
r
o
v
i
d
i
n
g
D
R
S
s
e
t
u
p
f
o
r
c
o
l
l
e
c
t
i
n
g
r
e
a
l
-
t
i
m
e
d
a
t
a
.
T
h
e
a
u
t
h
o
r
s
a
r
e
t
h
a
n
k
f
u
l
t
o
t
h
e
D
e
p
a
r
t
m
e
n
t
o
f
G
e
n
e
r
a
l
M
e
d
i
c
i
n
e
,
S
R
M
M
e
d
i
c
a
l
C
o
l
l
e
ge
H
o
s
p
i
t
a
l
a
n
d
R
e
s
e
a
r
c
h
C
e
n
t
e
r
,
K
a
t
t
a
n
k
u
l
a
t
hu
r
,
I
n
d
i
a
f
o
r
s
u
p
p
o
r
t
i
n
g
u
s
i
n
c
o
l
l
e
c
t
i
n
g
t
h
e
d
a
t
a
s
e
t
.
REFERE
NCE
S
[1]
D.
Dah
m
an
d
K
.
Da
h
m
,
Inter
p
retin
g
d
iffu
se
reflecta
n
ce
a
n
d
tra
n
smitta
n
ce:
a
th
eo
retica
l
i
n
tro
d
u
ctio
n
to
a
b
so
rp
tio
n
sp
ectro
scopy
o
f sca
tterin
g
mater
ia
ls
.
IM
Pub
licatio
n
s, 20
0
7
.
[2]
Å
.
Rin
n
an
,
F
.
v
an
d
en
Berg
,
an
d
S.
B.
Eng
elsen
,
“Re
v
iew
o
f
th
e
m
o
st
co
m
m
o
n
p
re
-
p
roc
ess
in
g
tech
n
iq
u
es
for
n
ear
-
in
fr
ared
sp
ectra
,”
TrAC Tr
e
n
d
s in
A
n
a
lytica
l Ch
emistr
y
,
v
o
l.
2
8
,
n
o
.
1
0
,
p
p
.
1
2
0
1
–
1
2
2
2
,
No
v
.
2
0
0
9
,
d
o
i: 10
.1016
/j.tr
ac.
2
0
0
9
.0
7
.00
7
.
[3]
E
.
A
r
e
n
d
s
e
,
O
.
A
.
F
a
w
o
l
e
,
L
.
S
.
M
a
g
w
a
z
a
,
H
.
N
i
e
u
w
o
u
d
t
,
a
n
d
U
.
L
.
O
p
a
r
a
,
“
C
o
m
p
a
r
i
n
g
t
h
e
a
n
a
l
y
t
i
c
a
l
p
e
r
f
o
r
m
a
n
c
e
o
f
n
e
a
r
a
n
d
m
i
d
i
n
f
r
a
r
e
d
s
p
e
c
t
r
o
m
e
t
e
r
s
f
o
r
e
v
a
l
u
a
t
i
n
g
p
o
m
e
g
r
a
n
a
t
e
j
u
i
c
e
q
u
a
l
i
t
y
,
”
L
W
T
,
v
o
l
.
9
1
,
p
p
.
1
8
0
–
1
9
0
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
l
w
t
.
2
0
1
8
.
0
1
.
0
3
5
.
[4]
J.
C.
Ma
ch
ad
o
,
M
.
A.
Fa
ria,
I.
M.
P.
L.
V.
O.
Fer
rei
ra,
R.
N.
M.
J.
P
ásco
a
,
an
d
J.
A.
Lop
es,
“Var
ie
ta
l
d
iscri
m
i
n
atio
n
o
f
h
o
p
p
ellet
s
b
y
near a
n
d
m
id
in
fr
ar
ed
sp
ectrosco
p
y
,”
Ta
la
n
ta
,
v
o
l.
1
8
0
,
p
p
.
6
9
–
7
5
,
Ap
r.
20
1
8
,
d
o
i: 10
.101
6
/j.talanta.2
0
1
7
.12.
0
3
0
.
[5]
S.
Bas
sin
i
et
a
l.
,
“
Mater
i
al
p
er
form
a
n
ce
in
lead
an
d
le
ad
-
b
ismuth
allo
y
,”
in
Co
mp
reh
en
sive
Nu
clea
r
Ma
teria
ls
,
Elsev
ie
r,
2
0
2
0
,
p
p
.
2
1
8
–
2
4
1
.
[6]
A.
P
.
C
raig,
B.
G
.
Bo
telh
o
,
L.
S.
Oliv
eira,
an
d
A
.
S.
F
ranca,
“
Mid
i
n
fr
ared
sp
ectrosco
p
y
an
d
ch
em
o
m
et
rics
as
to
o
ls
for
t
h
e
class
i
fication
o
f
roas
ted
co
f
fees
b
y
cu
p
q
u
ality
,”
Foo
d
Ch
emi
str
y
,
v
o
l.
2
4
5
,
p
p
.
1
0
5
2
–
1
0
6
1
,
Ap
r.
2
0
1
8
,
d
o
i:
1
0
.10
1
6
/j.f
o
o
d
ch
e
m
.20
1
7
.11
.06
6
.
[7]
S.
R.
Kh
an
d
asam
m
y
et
a
l.
,
“Blo
o
d
s
tain
s,
p
ain
tin
g
s,
an
d
d
rug
s:
Raman
sp
ectrosco
p
y
ap
p
li
catio
n
s
in
forens
ic
scien
ce,”
Fo
ren
sic
Ch
emistr
y
,
v
o
l.
8
,
p
p
.
1
1
1
–
1
3
3
,
May 2
0
1
8
,
d
o
i: 1
0
.10
1
6
/j.f
o
rc.
2
0
1
8
.02
.00
2
.
[8]
J.
Eng
el
et
a
l.
,
“Br
eaki
n
g
with
trend
s
in
p
re
-
p
rocess
in
g
?,”
TrAC
Tren
d
s
in
Ana
lytica
l
Ch
em
istr
y
,
v
o
l.
5
0
,
p
p
.
9
6
–
1
0
6
,
Oct.
2
0
1
3
,
d
o
i: 10
.1016
/j.tr
ac.
2
0
1
3
.0
4
.01
5
.
[9]
H.
M
a
rtens
,
S.
A
.
Jen
sen
,
an
d
P.
Geladi
,
“Multiv
ar
i
ate
lin
earity
trans
form
atio
n
for
n
ea
r
-
in
fr
a
red
re
flecta
n
ce
sp
ectrometr
y
,
”
in
Pro
ceedin
g
s o
f the No
rd
ic symp
o
si
u
m on
ap
p
lied
sta
t
istics
,
1
9
8
3
,
p
p
.
2
0
5
–
2
3
4
.
[10
]
P.
Geladi
,
D.
Ma
c
Do
u
g
all,
an
d
H.
Mar
ten
s,
“Linea
riz
atio
n
an
d
scatter
-
c
o
rr
ectio
n
for
n
ea
r
-
in
fr
ared
ref
le
ctan
c
e
sp
ectra
o
f
m
e
at,
”
App
lied
Sp
ectro
sc
o
p
y
,
v
o
l.
3
9
,
n
o
.
3
,
p
p
.
4
9
1
–
5
0
0
,
May 1
9
8
5
,
d
o
i: 10
.1
3
6
6
/0
0
0
3
7
0
2
8
5
4
2
4
8
6
5
6
.
[11
]
R.
J.
Ba
rnes
,
M.
S.
Dh
an
o
a,
an
d
S
.
J.
Liste
r,
“Stan
d
a
rd
n
o
rm
al
v
ar
iate
trans
for
m
atio
n
an
d
d
e
-
trend
in
g
o
f
n
ear
-
in
f
rar
ed
d
iff
u
s
e
ref
lect
an
ce sp
ectra
,”
App
lied
Sp
ectro
sco
p
y
,
v
o
l.
4
3
,
n
o
.
5
,
p
p
.
7
7
2
–
7
7
7
,
1
9
8
9
,
d
o
i: 10
.1366
/0
0
0
3
7
0
2
8
9
4
2
0
2
2
0
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
omp E
ng,
V
ol.
15
, No
.
1
,
Febr
uary
20
25
:
303
-
310
310
[12
]
M.
Zeaite
r,
J.
-
M.
Ro
g
er,
an
d
V.
B
ello
n
-
Maurel,
“Ro
b
u
stn
ess
o
f
m
o
d
els
d
ev
elo
p
ed
b
y
m
u
ltiv
ariat
e
calib
rat
io
n
.
Part
II
:
The
in
fluen
ce
o
f
p
re
-
p
rocess
in
g
m
eth
o
d
s,”
TrAC
Tren
d
s
in
Ana
lytica
l
Ch
emistr
y
,
v
o
l.
2
4
,
n
o
.
5
,
p
p
.
4
3
7
–
4
4
5
,
May
2
0
0
5
,
d
o
i: 10
.1016
/j.tr
ac.
2
0
0
4
.1
1
.02
3
.
[13
]
T.
F
earn,
C.
Ricci
o
li,
A.
Ga
rr
id
o
-
V
aro,
an
d
J.
E
.
Gu
err
e
ro
-
Gin
el,
“On
th
e
g
eo
m
et
ry
o
f
SNV
an
d
MSC,”
Ch
emo
metrics
a
n
d
Intellig
en
t Lab
o
ra
t
o
ry S
ystems
,
v
o
l.
9
6
,
n
o
.
1
,
p
p
.
2
2
–
2
6
,
Ma
r.
20
0
9
,
d
o
i: 10
.10
1
6
/j.chemolab.
2
0
0
8
.11
.0
0
6
.
[14
]
T.
Gen
k
awa
et
a
l.
,
“Bas
elin
e
co
rr
e
cti
o
n
o
f
d
iff
u
se
ref
le
ctio
n
n
ear
-
in
fr
a
red
sp
ectra
u
sin
g
sear
ch
in
g
regio
n
stan
d
ard
n
o
rm
al
v
ar
iate
(SRSNV
),
”
App
lied
Sp
ectro
sco
p
y
,
v
o
l.
6
9
,
n
o
.
1
2
,
p
p
.
1
4
3
2
–
1
4
4
1
,
Dec.
2
0
1
5
,
d
o
i: 10
.1366
/1
5
-
0
7
9
0
5
.
[15
]
S.
Ab
ati
,
C
.
B
ra
m
ati,
S
.
Bo
n
d
i,
A.
L
iss
o
n
i,
an
d
M
.
T
ri
m
archi,
“
Oral
canc
er
an
d
p
reca
n
cer
:
A
n
ar
rative
review
o
n
th
e
relevan
ce
o
f
early
d
iag
n
o
sis
,”
Inter
n
a
tio
n
a
l
Jo
u
rn
a
l
o
f
Envir
o
n
men
ta
l
Resear
ch
a
n
d
Pub
lic
Hea
lth
,
v
o
l.
1
7
,
n
o
.
2
4
,
Dec.
2
0
2
0
,
d
o
i: 10
.3390
/ijerph17
2
4
9
1
6
0
.
[16
]
B.
Yu
,
A.
Sh
ah
,
V
.
K.
Nag
ar
a
jan
,
an
d
D.
G.
Fer
ris,
“Di
ff
u
se
ref
l
ectance
s
p
ectrosco
p
y
o
f
ep
i
th
elial
tiss
u
e
with
a
smar
t
fiber
-
o
p
tic
p
rob
e,”
Biomed
ica
l Optics
E
xp
res
s
,
vo
l.
5
,
n
o
.
3
,
Mar
.
20
1
4
,
d
o
i: 10
.1
3
6
4
/
BOE.5.0
0
0
6
7
5
.
[17
]
B.
S.
Id
rees
et
a
l.
,
“In
-
v
itro
stu
d
y
o
n
th
e
id
en
tification
o
f
g
astro
in
testinal
stro
m
al
tu
m
o
r
t
iss
u
es
u
sin
g
laser
-
in
d
u
ced
b
reakd
o
w
n
sp
ectrosco
p
y
with ch
em
o
m
et
ric
m
eth
o
d
s,”
Biomed
ica
l
Op
tics Expr
ess
,
v
o
l.
1
3
,
n
o
.
1
,
Jan
.
2
0
2
2
,
d
o
i: 10
.1364
/B
OE.
4
4
2
4
8
9
.
[18
]
S.
V
.
Su
ryak
ala
an
d
S.
Prince
,
“
Inv
estig
atio
n
o
f
g
o
o
d
n
e
ss
o
f
m
o
d
el
d
ata
fi
t
u
sin
g
PLSR
an
d
PCR
regr
ess
io
n
m
o
d
els
to
d
eter
m
in
e
in
form
ativ
e
wav
el
en
g
th
b
an
d
in
NIR
regio
n
for
n
o
n
-
in
v
asiv
e
b
lo
o
d
g
lu
co
se
p
redictio
n
,”
Op
tica
l
a
n
d
Qu
a
n
tu
m
Electro
n
ics
,
v
o
l.
5
1
,
n
o
.
8
,
Au
g
.
2
0
1
9
,
d
o
i:
1
0
.100
7
/s1
1
0
8
2
-
0
1
9
-
1
9
8
5
-
7.
[19
]
A.
S.
Hak
a,
K.
E.
Sh
afer
-
Peltie
r,
M.
Fitzm
au
rice,
J.
C
rowe,
R.
R.
D
asar
i,
an
d
M.
S.
Fe
ld
,
“Diagn
o
sin
g
b
reast
cancer
b
y
u
sin
g
Raman
sp
ect
ros
co
p
y
,”
Pro
ceedin
g
s
o
f
th
e
Na
tio
n
a
l
Acad
emy
o
f
S
cien
ces
,
v
o
l.
1
0
2
,
n
o
.
3
5
,
p
p
.
1
2
3
7
1
–
1
2
3
7
6
,
Au
g
.
2
0
0
5
,
d
o
i: 10
.1073
/p
n
as.0
5
0
1
3
9
0
1
0
2
.
[20
]
G.
R
EI
CH,
“N
ear
-
in
fr
ared
sp
ect
ros
co
p
y
an
d
im
ag
in
g
:
b
asic
p
rincip
les
an
d
p
h
arm
ac
eu
tical
ap
p
licatio
n
s,”
Adva
n
ced
Dru
g
Deliver
y
Reviews
,
v
o
l.
57
,
n
o
.
8
,
p
p
.
1
1
0
9
–
1
1
4
3
,
Ju
n
.
2
0
0
5
,
d
o
i:
1
0
.10
1
6
/j.add
r.
2
0
0
5
.01
.02
0
.
[21
]
Ju
n
Hu
an
g
,
Saly
Ro
m
ero
-
Tor
res,
a
n
d
Mojg
an
Mos
h
g
b
ar,
“Practi
cal
c
o
n
sid
eration
s
in
d
ata
p
retr
eat
m
en
t
f
o
r
NIR
an
d
Rama
n
sp
ectrosco
p
y
,”
Am
erica
n
P
h
a
rma
ceut
ica
l R
eview
,
2
0
1
0
.
[22
]
K.
H
eil
an
d
U.
Sc
h
m
id
h
alter,
“An
e
v
alu
atio
n
o
f
d
iff
er
en
t
NIR
-
sp
ect
ral
p
re
-
tr
eat
m
en
ts
to
d
e
rive
th
e
so
il
p
ara
m
eters
C
an
d
N
o
f
a
h
u
m
u
s
-
clay
-
rich s
o
il,”
S
en
so
rs
,
v
o
l.
2
1
,
n
o
.
4
,
Feb
.
2
0
2
1
,
d
o
i: 10
.33
9
0
/s210
4
1
4
2
3
.
[23
]
A.
-
H
.
E
m
was
et
a
l
.
,
“Reco
m
m
en
d
ed
strateg
ies
for
sp
ect
ral
p
rocess
in
g
an
d
p
o
st
-
p
rocess
in
g
o
f
1
D
1
H
-
NMR
d
ata
o
f
b
io
fluid
s
wit
h
a particular
f
o
cu
s
o
n
urin
e,”
Meta
b
o
lo
mics
,
v
o
l.
1
4
,
n
o
.
3
,
Mar
.
2
0
1
8
,
d
o
i: 10
.10
0
7
/s1
1
3
0
6
-
0
1
8
-
1
3
2
1
-
4.
[24
]
G.
Li
an
d
S.
Den
g
,
“Qu
an
titativ
e
an
a
ly
sis
o
f
n
ear
-
in
fr
a
r
ed
sp
ectrosco
p
y
u
sin
g
th
e
BEST
-
1
D
Co
n
v
Net
m
o
d
el,”
Pro
cess
es
,
v
o
l.
1
2
,
n
o
.
2
,
Ja
n
.
2
0
2
4
,
d
o
i: 10
.33
9
0
/p
r120202
7
2
.
[25
]
E.
B.
Silv
a
et
a
l.
,
“
A
regio
n
al
leg
acy
so
il
d
ataset
for
p
redictio
n
o
f
sand
an
d
clay
co
n
ten
t
with
VI
S
-
NIR
-
SW
IR,
i
n
so
u
t
h
ern
Brazil,”
Revista
B
ra
sileir
a
d
e Ciên
cia
do
So
lo
,
v
o
l.
4
3
,
2
0
1
9
,
d
o
i
: 10
.15
9
0
/
1
8
0
6
9
6
5
7
rbcs
2
0
1
8
0
1
74.
BIOGR
AP
HI
ES OF
A
UTH
ORS
Wince
nt
An
to
Win
Sh
ali
ni
re
ce
iv
ed
h
er
B
.
E.
degr
ee
in
e
lectr
oni
cs
an
d
com
munica
ti
on
engi
ne
eri
ng
fro
m
Anna
Univ
er
sity,
Chenn
ai
in
201
6
.
and
M.E
degr
e
e
in
com
munica
ti
on
sys
te
ms
from
Anna
Univer
sity
,
Chenna
i
in
201
8
.
She
is
cur
re
ntl
y
a
Ph.D.
student
at
SR
M
Instit
ute
of
Science
and
Technol
ogy,
Chenn
ai
.
Her
rese
ar
ch
in
t
ere
sts
include
opti
c
al
co
mm
un
ic
a
ti
on,
spe
ct
ros
copy,
and
b
io
-
p
hotoni
cs
.
She
c
an
be
con
ta
c
te
d
at
e
ma
i
l:
aw2720@srmist.edu.
in
.
Th
ulas
i
Rajalak
sh
mi
is
cu
rre
ntl
y
an
associ
at
e
profe
ss
or
in
the
Depa
r
tm
en
t
of
El
e
ct
roni
cs
and
Comm
unicati
on
Engi
ne
eri
ng,
Fa
cul
ty
of
Engi
ne
e
ring
and
T
ec
hno
logy,
at
th
e
Katt
anku
la
thur
Cam
pus
of
SR
M Insti
tute
of
Sci
e
nce
and
Technol
ogy
(form
erl
y
k
nown a
s SR
M
Univer
sity).
He
r
rese
arc
h
int
er
e
sts
include
m
edica
l
i
ma
ge
pro
ces
sing,
bio
signa
l
proc
essing,
biosensors,
and
biom
edica
l
instr
ume
nt
at
ion
.
She
has
guide
d
sev
e
ral
postgradu
ate
proje
c
ts
and
is
cur
r
ent
ly
supe
rvising
Ph.D.
st
udent
s.
She
h
as
patent
gra
n
ts.
She
has
al
so
r
ecei
ved
proj
ec
t
funding
fro
m
S
RMIS
T
for
th
e
dev
el
opm
ent
o
f
an
au
tom
a
te
d
audi
o
meter
.
She
h
as
a
lso
cont
ributed
ch
ap
te
rs
to
seve
r
al
b
ooks
and
publi
shed
num
ero
us
pa
per
s
in
interna
t
io
nal
journ
al
s.
She
ca
n
be
con
tacte
d
via email
at
rajala
k
t@srmi
st
.
edu.in.
Selvanayagam
Vas
anth
a
dev
Su
ryak
ala
serv
es
as
a
d
edi
c
ated
assistant
profe
s
so
r
in
the
Depa
rtme
nt
of
El
e
ct
roni
cs
and
Co
mm
uni
c
at
ion
Eng
ineeri
n
g
a
t
SR
M
Instit
u
te
of
Scie
n
ce
and
Technol
ogy
,
at
the
Ka
tt
ank
ula
thur
C
am
pus.
Her
rese
arc
h
i
nte
rests
en
com
p
ass
a
wide
spec
trum
of
topics,
including
image
proc
essing,
diffuse
ref
l
ecta
n
ce
spec
troscopy,
and
optica
l
com
mun
i
cation. She
has
guide
d
nume
rous
postgr
adua
t
e
student
s
.
She
has
publi
she
d
nume
rous
pape
rs i
n
in
te
rna
ti
onal j
ourn
al
s.
She
ca
n
be
con
tacte
d
via email
at
suryaka
s@
srmi
st
.
edu.in.
Evaluation Warning : The document was created with Spire.PDF for Python.