!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA
ESCUELA DE DOCTORADO
PROGRAMA DE DOCTORADO:
INVESTIGACIÓN APLICADA A LAS CIENCIAS SANITARIAS
INTEGRACIÓN DE PRINCIPIOS FÍSICOS EN UNA UNIDAD DE HIPOACUSIA
INTEGRATION OF PHYSICAL PRINCIPLES IN A HEARING LOSS UNIT
TESIS DOCTORAL
PRESENTADA POR: Dª MARÍA TERESA PÉREZ ZABALLOS
DIRIGIDA POR:
PROF. DR. ÁNGEL M. RAMOS MACÍAS
PROF. DR. SANTIAGO RODRÍGUEZ FEIJOÓ
DIRECTOR CODIRECTOR
Ángel M. Ramos Macías
San iago Rod íguez Feijoó
Ma ía Te esa Pé ez Zaballos
DOCTORANDO
!
!
i
a la o illa. G acias a ellos ha sido posible que odas mis emp esas lleguen a
buen pue o, incluida es a esis. Yo no soy nada sin mi amilia. Ellos son mi mo o .
Po eso, quie o da les las g acias a odos y quie o elici a nos, po que somos la
mejo amilia del mundo, po que aguan amos has a en los momen os mas du os.
Clemen e; Vicen e, Ma ina, Vicen y Dani; Balbina, José, Jose, Consu y Juan; la
Abuela y el Abuelo y Do a, g acias. En especial, es oy ag adecida a mis pad es y
mis he manos, po que son lo que más quie o en el mundo. G acias po
le an a me cuando caigo y po aga a me cuando me ambaleo. Jun os podemos
con odo.
¿Y que puedo deci le a Ca los, que ha sido el su ido úl imo de los es agos y el
es és de esc ibi es a esis?. Es el homb o i me donde me apoyo y mi aleg ía de
cada día. Él es la azón de que quie a se mejo y que in en e da siemp e lo
mejo de mí. Además, po azones que desconozco, sien e especial ca iño po
mis momen os EUREKA, y me anima siemp e a lanza me y p oba mis ideas. Con
alguien así, ¿quien no iba a se eliz? Soy muy a o unada de ene lo a mi lado.
G acias po que e me al y como soy.
Y pa a inaliza , quie o ag adece a a ias pe sonas que me han ayudado de
di e en es mane as: a Be na do y a Mabel , po o ganiza las sesiones con los
con olado es; a Es he y a Ri a, po apoya me y escucha me an o es e año; a
Miguel, po es imula me in elec ualmen e desde que enía 16 años (no iene ni
idea de lo mucho que ha in luido en mi ida) y a Guacima a, po pasa de
olun a ia a g an amiga y conseje a.
A odos, g acias.

Índice
!
!
ÍNDICE
Índice
!
!
Índice
!
!
Í N D I C E
I.
INTRODUCTION
1.- Mo i a ion and o e iew o his disse a ion
2.- The physics o sound
2.1.- Ana omy o a sound wa e
2.2.- In ensi y o a sound wa e
2.3.- Fou ie Analysis o a sound wa e
3.- The physics and ana omy o he ea
3.1.- The ou e ea
3.2.- The middle ea
3.3.- The inne ea
4.- Psychophysics
4.1.- Pi ch pe cep ion
4.2.- Loudness pe cep ion
5.- The cochlea implan .
5.1.- Signal p ocessing
5.2.- Signal p ocessing s a egies
5.3.- Pi ch pe cep ion wi h cochlea implan s
5.4.- Loudness pe cep ion wi h cochlea implan s
3
5
5
8
11
14
15
16
18
24
25
28
33
34
36
41
42
II.
EXPERIMENTAL WORK
6.- Psychoacophysics: Speech pe cep ion in noise s udies
6.1.- E ec s o High-F equency Supp ession o Speech Recogni ion
in Noise in Spanish No mal-Hea ing Subjec s
6.1.1- In oduc ion
47
47
48
Índice
!
!
6.1.2- Objec i es and hypo heses
6.1.3- Subjec s and Me hods
6.1.4- Resul s
6.2.- E ec s o speech-in-noise aining and high equency
supp ession in Ai T a ic Con olle s. A con ol g oup s udy
6.2.1- In oduc ion
6.2.2- Objec i es and hypo heses
6.2.3- Subjec s and Me hods
6.2.4- Resul s
7.- Elec ically e oked psychoacous ics in cochlea implan ecipien s
7.1.- PsychoLAB: A Psychophysics expe imen al se up o measu e
elec ode disc imina ion in Nucleus cochlea implan ecipien s
7.1.1- In oduc ion
7.1.2- Pla o m design
7.2.- E ec o he elec ode-inne wall dis ance on pi ch pe cep ion in
cochlea implan ecipien s using s aigh and pe imodiola
a ays
7.2.1- In oduc ion
7.2.2- Objec i es and hypo heses
7.2.3- Subjec s and Me hods
7.2.4- Resul s
8.- Elec omagne ism
8.1.- The e ec o e e ence elec ode posi ion on powe
consump ion in Cochlea Implan s
8.1.1- In oduc ion
8.1.2- Objec i es and hypo heses
8.1.3- Ma e ial and Me hods
50
50
56
59
59
60
62
65
70
71
71
73
82
82
85
86
92
106
106
106
109
110
82
Índice
!
!
8.1.4- Resul s
8.1.4- Resul s
116
III.
DISCUSSION
9.- Psychophysics
9.1.- How he heal hy ea unde s ands speech in noise and he
e ec s o aining on i
10.- Elec ically e oked psychoacous ics in cochlea implan ecipien s
10.1.- How physical and psychophysical a iables a ec elec ode
disc imina ion in cochlea implan ecipien s
11.- Elec omagne ism
11.1.-The ole o he e e ence elec ode in cochlea implan s powe
consump ion
125
125
131
131
137
137
0
0
0
IV.
CONCLUSIONS
143
V.
RESUMEN EN ESPAÑOL
147
VI.
REFERENCES
155
VIII.
TABLE INDEX
177
IX.
FIGURE INDEX
179
X.
APPENDIX
193
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!

!
!
!
In oduc ion
!
!
!
1!
I. INTRODUCTION
In oduc ion
!
!
!
2!
In oduc ion
!
!
!
3!
1.- MOTIVATION AND OVERVIEW OF THIS DISSERTATION
In oday's wo ld, whe e in e disciplina i y plays a key ole in esea ch, i is i al
o change he adi ional schemes ha sepa a ed p o essionals acco ding o
hei deg ees. To ge a ue scien i ic e olu ion, di e en ields o knowledge
need o be mixed and hyb idized, so ha old p oblems a e looked a om new
pe spec i es. This is how inno a i e solu ions eme ge.
This disse a ion a emp s o illus a e his concep . The addi ion o a physicis
o a Hea ing Loss Uni has yielded di e en solu ions o clinical p oblems. The
p esence o a physicis in a clinical session ha lis ens o he eal needs o he
heal h pe sonnel can d ama ically speed up inno a ion ega ding su ge y,
cochlea implan p og amming, ehabili a ion and pa ien e alua ion. In his
hesis, wo k has been done ying o co e all o hese aspec s.
The h ead o he s udies p esen ed he e is he imp o emen o sound
pe cep ion o cochlea -implan ed pa ien s unde di e en condi ions.
The wo k has been di ided in o he di e en physical ields ha we e applied
o sol e each ini ial ques ion. Th ee sound condi ions we e e alua ed. Fi s ,
speech-in-noise pe cep ion mechanisms a e s udied. Then, how pi ch
disc imina ion is a ec ed by elec ode p oximi y was e alua ed. Finally, he
imp o emen o sound in pa ien s wi h high impedances was looked a .
The i s pa ocuses on psychoacous ics, a hyb id ield be ween physics and
psychology. Two s udies we e conduc ed. The i s is a s udy on he ele ance
o high equencies o he pe cep ion o speech in noise in no mal hea ing
indi iduals. This showed ha high speech componen s, abo e 8 kHz, help o
unde s and wo ds subjec ed o ambien noise. The second s udy is a
con inua ion o he i s , and explo es he e ec o long- e m speech in noise
aining in ai a ic con olle s. This job sec o de elops speech pe cep ion
abili ies du ing he cou se o hei wo k. The special acous ic condi ions o
which hey a e subjec ed ain hei b ains o ex ac he maximum amoun o
audi o y cues o be able o es ablish adio communica ions. These ha e he
In oduc ion
!
!
!
10!
Ano he a iable ha can be used o desc ibe sound in ensi y is he sound
p essu e le el (SPL). In ensi y and p essu e can be ela ed by he equa ion:
!=!!
!" [2.4]
Whe e P is sound p essu e and is measu e in Pascals (Pa), ρ is he densi y o
ai (1.2 kg/m3 a 20oC) and is he speed o sound (331 m/s in ai ). Thus, SPL
can be desc ibed as: (5) (6)
!"#(!")=20 log!"
!
!
!
[2.5]
While he in ensi y o a sound is a e y objec i e quan i y ha can be
measu ed wi h sensi i e ins umen a ion, he loudness o a sound is a
subjec i e esponse ha will a y wi h a numbe o ac o s. The sound
in ensi y mus be ac o ed by he ea 's sensi i i y o he pa icula equencies
con ained in he sound and will a y sligh ly depending on he indi idual. Bu
his will be discussed in mo e de ail in u he chap e s. Fo now, i su ices o
say ha , since humans do no pe cei e low and high equency sounds as
well as hey do pe cei e middle equencies a ound 2 kHz, h ee weighing
sys ems o SPL ha e been es ablished: A, B and C. The g aph in igu e 2.4
desc ibes he sensi i i y o each weighing sys em as a unc ion o equency.
Using he dB(A) measu ing sys em, he sound le el me e will be less
sensi i e o e y high and e y low equencies. I app oxima es he ea when
hea ing a 1kHz sound a 40 dB. Measu emen s made wi h his scale a e
exp essed as dB(A), which is e y use ul o elimina ing inaudible low
equencies. The in e media e B-con ou app oxima es he ea o medium
loud sounds and i is a ely used. Finally, he dB (C) measu ing sys em does
no il e ou he lows and highs o such an ex en as he o he con ou s. I
app oxima es he ea a e y high sound le els and has been used o a ic
noise su eys in noisy a eas. (6)

In oduc ion
!
!
!
11!
Figu e 2.4. Rela i e esponse o each weighing SPL sys em as a unc ion o equency. I can
be seen how dBA is mos sensi i e o equencies up o 1000 Hz, while dBB and dBC end o
la en ou soone . (Own p oduc ion)
2.3.- Fou ie Analysis o a sound wa e
All sounds can be decomposed in o supe posi ions o sinusoids o di e en
equencies, ampli udes and phases o , seen om he opposi e pe spec i e,
simple sinusoids can combine o o m any kind o complex sound wa e.
Figu e 2.5 illus a es his p inciple. The decomposi ion is known as Fou ie
analysis and gi es back he composi ion o he sound wa e ( ) in e m o i s
pu e ones o equency composi ion and he in ensi y wi h which hey appea .
This ansla es ma hema ically in o a sum o cosines and sines o di e en
equencies:
!!=!!+!!sin(2!∙!(2!"#))
!
!!!+!!cos(2!∙!(2!"#))
!
!!! [2.6]
Whe e he exp essions in he sums a e called he Fou ie coe icien s, A0 is a
cons an ha desc ibes he a e age alue o he unc ion (i.e. he poin a ound
which i oscilla es up and down) and An and Bn a e he ampli udes o each
equency o one componen . (7) (8)
In oduc ion
!
!
!
12!
Figu e 2.5. When wo single equency wa es a e combined, he esul is a wa e o m ha is
he sum o hem. (Own p oduc ion)
The ans o ma ion o a complex wa e pa e n in ime in o i s equency
componen dis ibu ion is known as Fou ie ans o ma ion and i cons i u es a
basic ool in he ield o acous ics, because i allows he scien is o
cha ac e ise sounds. A Fou ie ans o m shows ha any wa e can be
decomposed in o a sum o sinusoids.
I is no he aim o his sec ion o desc ibe he ma hema ical backg ound
behind a Fou ie ans o m, which is complex and a ely used analy ically.
Mos eal li e wa e o ms a e so complex ha Fou ie ans o ms a e ca ied
ou by compu e s. These no mally use an algo i hm called he Fas Fou ie
T ans o m o FFT. Gi en a wa e o m unc ion in he ime domain (like ha in
igu e 2.5, blue line) an FFT would gi e i s p o ile in he equency domain
( igu e 2.5 ed lines). (8)
These illus a ions show he essen ial na u e o he FFT. Fo a sine wa e wi h
a single equency, he FFT consis s o a single peak. Combining wo sound
wa es p oduces a complex pa e n in he ime domain, bu he FFT clea ly
shows i as consis ing almos en i ely o wo equencies. Thus, Fou ie
In oduc ion
!
!
!
13!
analysis can g ea ly simpli y acous ic p oblems and e eal aspec s ha a e
no clea ly isible in he ime domain. (8)
In oduc ion
!
!
!
14!
3.- THE PHYSICS AND ANATOMY OF THE EAR
The ea is he mos complex ansduce e e c ea ed. I is able o con e
ene gy om sound p essu e in o mechanical mo emen and his, in u n, in o
elec ical ne e impulses ha a el all he way o he b ain. The ea ’s abili y o
ca y ou his p ocess allows us o pe cei e he de ails o a complex sound.
The ea de ec s he equencies ha make hem up; each one o hei
indi idual in ensi ies and, h ough he in eg a ion o bo h aspec s, i can
ecognise ea u es such as who is alking and wha a pe son is saying e en
when noise is p esen .
The ea has h ee main pa s: ou e , middle and inne ea ( igu e 3.1). Each
pa has a speci ic unc ion in he p ocess o de ec ion and in e p e a ion o
sound. The i s se es o pick up sound and p epa es i o ans e he ene gy
o he middle ea in he mos e icien way, by ma ching ai and bone
impedance. The second pa is esponsible o ans o ming he ene gy o he
sound wa e (p essu e a ia ions) in o mechanical displacemen s o he
ossicles, which c ea es a comp essi e wa e wi hin he inne ea . The inne
ea is illed wi h luid, which bends hai cells cilia as i mo es and in u n
ans o ms his mechanical displacemen in o elec ical impulses ha a e
iden i iable by he b ain.
In oduc ion
!
!
!
15!
Figu e 3.1. Ana omy o he ea . The pinna and ex e nal audi o y canal o m he ou e ea ,
which is sepa a ed om he middle ea by he ympanic memb ane. The middle ea houses
h ee ossicles, he malleus, incus and s apes. Toge he hey o m he sound conduc ing
mechanism. The inne ea consis s o he cochlea, which ansduces ib a ion o a ne ous
impulse and he es ibula laby in h ha houses he o gan o balance. (Own p oduc ion using
a ee license image om B uceBlaus)
3.1.- The ou e ea
The ou e ea consis s o he ea and ea canal. I s o e all unc ion is o ac as
a p e-ampli ie o imp o e he sensi i i y o he sys em and i s ac ion
mechanisms a e a di ec esul o he physical cha ac e is ics o sound. (9)
The sound ene gy sp eads sphe ically om he sou ce. Fo a poin sou ce o
sound, as he adii o he sphe es o sound p opaga ion ge u he away, hey
also become la ge . Gi en ha he su ace o a sphe e is p opo ional o he
squa e o i s adius (!=4!!!), he ene gy in he sound wa e a i ing a a
small a ea (e.g. an ea d um) declines wi h he squa e o he dis ance om
he sou ce (!∝1/!!)). This ela ionship be ween sound in ensi y o ene gy
and dis ance om he sou ce is known as he in e se squa e law. (9)

In oduc ion
!
!
!
16!
The e o e, o a gi en sound in ensi y, a la ge pinna will ha e a b oade a ea
and hus will cap u e sound mo e e icien ly. The ea canal, on he o he hand,
ac s as a b oadband esona o ube, closed a one end. I has an a e age
diame e o 0.7 cm and is 2.7 cm long (L). By ac ing as a ube closed a one
end as in igu e 3.2, i gi es ise o s anding wa es a a undamen al esonan
equency ha is ou imes i s leng h and a odd mul iples o i , called
ha monics. The closed end is es ic ed o be a wa e node and he open end
an an inode. As seen in igu e 3.2, his condi ion yields a wa eleng h o he
undamen al mode ha is ou imes he leng h o he ai column inside he
canal (10.8 cm). The es ic ion o he closed end p e en s i om p oducing
e en ha monics. Since (!=!/! and !=4!), i s esonan equencies a e
gi en by he equa ion:
!
!=!"
!!;!=1,3,5… [3.1]
So he i s esonan equency will be a ound 3 kHz. (9) (10) (11)
!
Figu e 3.2. The ea canal ac s as a ube open a one end, which o ces he closed end o be a
node and he open end and an inode. This causes he minimum esonan equency (1s
ha monic) o be ou imes he leng h o he ea canal as. The nex ha monic ha ollows
hese bounda y condi ions is an odd mul iple o his esonan equency. (Own p oduc ion)
3.2.- The middle ea
The main pa s o he middle ea a e he ympanic memb ane and he h ee
ossicles (malleus, incus and s apes). Toge he hey ampli y he sound wa e
In oduc ion
!
!
!
17!
and pe o m he impo an unc ion o impedance ma ching. This imp o es he
e iciency o he ans e o he mechanical ib a ions o he ympanic
memb ane (~ low mechanical impedance) o he o al window (~ 20 imes
highe impedance). (11)
The ympanic memb ane is a ci cula memb ane o abou 8-9 mm in diame e ,
0.1 mm in hickness, wi h an a ea o 65 o 80 mm2 and a weigh o 14mg. I
has a conical shape and is a ached o he inne pa o he s apes. I
es ablishes he bounda y be ween he ou e and middle ea , ac ing jus like
he memb ane o a mic ophone. I con e s he ai p essu e wa es in o
mechanical ib a ions ha a e ansmi ed o he ossicles. The ympanic
memb ane has he peculia p ope y o being a damped esona o ( he sys em
goes back e y as o i s es ing posi ion) due o he high ension o ce o
which i is subjec ed. This b oadens he numbe o equencies ha can cause
he sys em o esona e and i he e o e has a la ge ange ha goes om 20 o
20,000 Hz. Because i is 20 imes he size o he o al window ( he memb ane
ha connec s he middle ea wi h he inne ea luid ca i y), i ampli ies he
incoming sound by a ac o o 20, compa ed wi h he di ec ecep ion o he
sound by he o al window alone. (12)
The audi o y ecep o s o he inne ea ope a e in a luid en i onmen , and he
inne ea is eally an "unde wa e " sound ecei e . When sound in ai s ikes a
luid bounda y (a bounda y be ween media wi h di e en acous ic
impedances) he e is a heo e ical loss o 99.9% o he ene gy in a sound
wa e in ai due o e lec ion. This 99.9% loss is equi alen o 30 dB. In o de
o o e come his misma ch in he impedance be ween ai and luid, he middle
ea is in e posed be ween he ympanic memb ane and he o al window. (13)
Ossicles wo k like a composi e le e ha achie es a mul iplica ion o o ce
be ween he ympanic memb ane and he o al window. Al hough hey a e
among he smalles bones in he body hey a e capable o sp eading o ce
wi h g ea p essu e because o hei physical s uc u e and layou wi hin he
inne ea . When hese h ee bones wo k oge he in combina ion wi h wo
In oduc ion
!
!
!
18!
small muscles, hey o m he ossicle chain. This chain ans e s he o ce
coming om he ympanic memb ane (e ec i e a ea o o 65 mm2) o he o al
window (e ec i e a ea o 3.2 mm2), hus inc easing he p essu e wi h which
his o ce is applied. Since he p essu e is he o ce pe uni a ea (!=!/!),
he ela ionship be ween he p essu e in he ympanic memb ane and he o al
window will be:
!=!
!∙!! (Tympanic p essu e)
!
!=!!∙!!
!!
[3.2]
!=!
!∙!! (O al window p essu e)
The ac o A /Ao is app oxima ely 20. I also happens ha he le e a m
o med by he malleus in o a ing abou i s pi o is somewha longe han ha
o he incus, gi ing ano he ac o o abou 1.3 in p essu e inc ease. These
wo ac o s mul iplied, yield abou a 26x inc ease in p essu e, which is abou
29 dB, hus jus abou o e coming he heo e ical 30 dB loss due o he
ai /liquid in e ace. The e o e, he middle ea ma ches he acous ic impedance
be ween ai and luid, maximizing he low o ene gy om he ai o he luid o
he inne ea . (9)
3.3.- The inne ea
The inne ea consis s o a bonny laby in h, di ided in o he es ibula sys em
and he cochlea. I is a sealed ca i y illed wi h a luid called lymph. The
cochlea is he o gan in cha ge o sound ansduc ion in o elec ical signals.
The cochlea is spi ally wounded up, o ming wo and a hal u ns, wi h a
leng h o abou 30 mm and a a iable sec ional a ea ha anges om 4 mm2
a i s basal end, o 1 mm2 a i s apical end. The cochlea ca i y is di ided by
wo memb anes: he Basila and Reissne memb anes. Bo h di ide i
longi udinally in o h ee channels. Figu e 3.3 shows i s main ea u es. They
connec a he apex o he cochlea in a egion known as Helico ema. The
Reissne memb ane sepa a es he Scala Ves ibuli om a smalle ca i y
In oduc ion
!
!
!
19!
known as he cochlea duc . This, in u n, is sepa a ed by he basila
memb ane (BM) om a hi d ca i y called Scala Timpani. The BM is
composed o a g ea numbe o au , adially pa allel ib es sealed be ween a
gela inous ma e ial o e y weak shea s eng h. These ib es a e esonan a
p og essi ely lowe equencies as one p og esses om he basal o he
apical ends o he cochlea. Fou ows o hai cells lie on op o he BM,
oge he wi h suppo ing cells. A single inne ow is medial, closes o he
cen al co e o he cochlea. I has an abundan ne e supply ca ying
messages o he b ain. The h ee ou e ows ecei e mainly an a e en ne e
supply. The assembly o med by he BM, he ec o ial memb ane, he hai
cells and he ne e e minals is called he o gan o Co i. Any na u al
displacemen o he cochlea pa i ion esul s in a ocking mo ion o hese
s uc u es and consequen ly a la e al displacemen o he inne hai cells. This
mo emen causes depola iza ion and hipe pola iza ion o he hai cells and
ansmi s he ne e impulse o he neu al ends ha go along he cen e o he
cochlea o modiolus o ming a s uc u e known as he spi al ganglion. This
complex is he hea o he ea , as i con e s mechanical ib a ions in o
elec ical cu en s han can be unde s ood by he b ain. (10)
In oduc ion
!
!
!
26!
mechanics o na u e expands, so does he p esence o physics in e e y ield
o science.
Wi h he excep ion o pu e ones, pi ch is no a simple unc ion o he spec al
con en o a sound. Ra he , pi ch is ela ed mo e closely o he epe i ion a e,
o en elop epe i ion a e, o he sound wi h a ange om 30 Hz o 5 kHz.
Sounds wi h he same epe i ion a e and e y di e en spec a o en ha e he
same pi ch (e.g. a pu e one wi h a equency o 500 Hz and a complex one
wi h high ha monics wi h a undamen al o i s esonan equency, F0 o 500
Hz). On he con a y, sounds wi h simila spec a can ha e e y di e en
pi ches (e.g whi e noise wi h ampli ude modula ed a 100 o 200 Hz). Thus,
he audi o y sys em combines in o ma ion ac oss he cochlea loca ion
( equencies and in ensi y), o de i e he pi ch o he sound. Mo eo e , pi ch is
also ep esen ed in e ms o he p ecise iming o neu al impulses in he
audi o y ne e and a highe cen es in he audi o y sys ems.
Many sounds ha e acous ic wa e o ms ha epea hemsel es o e ime.
These sounds a e o en pe cei ed as ha ing a pi ch ha co esponds o he
epe i ion a e o he sound. This a ibu e o sound is one o he mos ele an
pe cep ual dimensions o speech communica ion, ca ying p osodic
in o ma ion in wes e n languages, bu also seman ic in o ma ion in onal
languages such as Manda in. Mo eo e , i enhances ou abili y o pe cep ually
seg ega e sound sou ces, based on di e ences in undamen al equency
(F0). Con e sely, i can be used o g oup oge he he indi idual sound
componen s (ha monics) ha a ise om he same ib a ing sou ce. Pi ch,
he e o e, de ines and di e en ia es ou acous ic en i onmen .
Theo ies o pi ch pe cep ion
The e a e cu en ly wo heo ies o pi ch pe cep ion (see igu e 4.1). The
spa ial heo y is based on wo main ene s. The i s is ha he s imulus is
subjec ed o spec al analysis in he cochlea; so each equency exci es
di e en loca ions along he basila memb ane and, consequen ly, neu ons

In oduc ion
!
!
!
27!
wi h di e en cen al equencies. This p ocess is called ono opic
o ganiza ion. The second ene s a es ha he pe cep ion o a s imulus is
ela ed o he pa e n o exci a ion ha i p oduces. Fo a pu e one, i is
assumed ha his co esponds o he place o maximum exci a ion. The i s o
hese p inciples has been con i med in a se ies o independen expe imen s,
including di ec obse a ion o he mo emen o he MB. The second is s ill in
dispu e. (19) (15)
Figu e 4.1. Tempo al and place coding o sound. In empo al coding (le ), neu ons i e ac ion
po en ials in phase wi h he sound wa es. Place coding ( igh ) e e s o he pe cep ion o pi ch
depending on he si e o s imula ion. This is possible because he audi o y co ex ha
esponds o sound is a anged ono opically. (Own p oduc ion)
An al e na i e heo y is he empo al heo y, based on he assump ion ha he
one o a s imulus is ela ed o i s e oked neu onal i ing ime pa e n. The
peaks o neu al i ing end o occu a a pa icula phase o he wa e o m in
he MB ( he cells phase-lock o s imuli). This causes he ime in e als
be ween successi e peaks o app oach in eg al mul iples o he wa e o m
pe iod. Phase-locking becomes e y weak o sinusoids wi h equencies
abo e app oxima ely 5 kHz, as s a ed in he p e ious chap e , al hough he
In oduc ion
!
!
!
28!
p ecise uppe limi in humans is no known. Howe e , he ones p oduced by
musical ins umen s, he human oice, and he mos common sou ces o
sound ha e undamen al equencies below 5 kHz. (Moo e 2013)
Physiological measu emen s and simula ions using compu a ional models
ha e demons a ed ha he epe i ion a es o s imuli a e e y well
ep esen ed by he pa e n o phase locking in he audi o y ne e. Fo many
esea che s, he ime e sus place a gumen has been eplaced by he
ques ion o how and whe e in he audi o y pa hway he phase-locked ac i i y
is analysed. The maximum equency o which a ib e will phase lock declines
om he audi o y ne e o he audi o y co ex and i is hough ha
somewhe e in he b ains em, synch ony ep esen a ion is con e ed in o a
a e-place ep esen a ion in which di e en neu ons code o di e en pi ches
in e ms o o e all i ing a e. (26) (27)
Howe e , he e emain some undamen al ques ions o be answe ed
conclusi ely, such as how phase-locked neu al ac i i y is ans o med in o a
a e-place ep esen a ion o pi ch; whe e his ans o ma ion akes place and
how his in o ma ion is used in objec and pa e n iden i ica ion.
4.2.- Loudness pe cep ion
Loudness is he subjec i e in ensi y o a sound. I is subjec i e because i
depends en i ely on he indi idual. The e m in ensi y is used he e because
he esponse indica es how s ong he sound seems o he lis ene . This
de ini ion is ague, bu in expe imen s on loudness i usually su ices o elici
consis en esponses. These esponses can hen be s udied o es ablish a
ela ion be ween loudness and he expe imen ally manipula ed physical
a iables. Sound in ensi y is he mos impo an o hem in de e mining
loudness, bu spec al a iables such as signal equency and bandwid h also
ha e an e ec on i . The du a ion and in e mi ency o a sound a e among he
signi ican empo al a iables. Mo eo e , backg ound sounds can also a ec
he pe cep ion o loudness. The e o e, sound in ensi y mus be ac o ed by he
In oduc ion
!
!
!
29!
ea 's sensi i i y o he pa icula equencies, and hei du a ions, con ained in
he sound. As a comple e e iew o all he physical pa ame e s ha a ec
loudness pe cep ion would equi e a le el o de ail ha is beyond he scope o
his disse a ion, jus he main a iables will be desc ibed in de ail he e. Fo a
mo e de ailed desc ip ion, he au ho e e s he eade o. (19) (28)
Loudness le el
Loudness le el indica es how loud a 1 kHz one mus be in o de o sound as
loud as a es one. To de e mine he le el o a gi en sound, he subjec is
asked o adjus he le el o a 1 kHz one un il i has he same loudness as he
es sound. They a e p esen ed al e na i ely a he han simul aneously. The
esul an le el o he 1 kHz one is he loudness le el o ha one and is gi en
in phons. I ins ead he 1 kHz one is ixed in le el, and es sounds a
di e en equencies a e adjus ed o gi e a loudness ma ch, an equal-
loudness con ou is gene a ed ( igu e 4.2). These s anda d equal loudness
con ou s (ISO 226, 2003), a e based on ex ensi e measu emen s o se e al
labo a o ies, as desc ibed by Suzuki and Takeshima (29) . This cu e shows
how s ong each equency sound mus sound o equal he co esponding
phon line. Fo example, i he line is a 30 dB SPL, a 125 Hz sound mus ha e
an SPL o 50 dB o sound as loud as a 1 kHz sound a 30 dB SPL. These
con ou s end o la en a high loudness le els. This means ha he a e o
g ow h di e s o ones o di e en equency. The a e o g ow h is g ea e o
low and e y high equencies han o middle ones. (30) (29)
In oduc ion
!
!
!
30!
Figu e 4.2. Equal-loudness con ou s o loudness le els om 10 o 120
phones o sounds p esen ed binau ally om he on al di ec ion. The
absolu e h eshold cu e (MAF) is also shown. (19)
This is he basis behind he weigh ing sys ems seen in p e ious chap e s. The
A weigh ing is based oughly on he 30-phon equal-loudness con ou . A high
le els, he weigh ing becomes mo e linea , so he C weigh ing is used. The B
weigh ing is used o in e media e le els, and i is based on he 70-phon
equal-loudness con ou . (19)
Loudness models
S.S S e ens was he pionee in he de elopmen o scales o loudness and
o he senso y dimensions. He sugges ed ha pe cei ed loudness, L, was a
powe unc ion o physical in ensi y I:
!=!!!.![4.1]
Whe e k is a cons an ha depends on he subjec and he uni s used.
Acco ding o his equa ion, gi en wo sounds whe e one is 10 imes mo e
In oduc ion
!
!
!
31!
in ense han he o he in he physical domain, he pe cei ed loudness is only
wo imes highe (10!.!≈2) which implies he exis ence o comp ession. (31)
The powe law ela ionship be ween in ensi y and loudness has been
con i med in a la ge numbe o expe imen s using a a ie y o echniques.
Howe e , he e ha e also been c i icisms. The echniques a e e y
suscep ible o bias e ec s like: 1) he ange o s imuli p esen ed; 2) he i s
s imulus p esen ed; 3) he ins uc ions o he subjec ; 4) he ange o
pe missible esponses; 5) he symme y o he esponse ange and 6) o he
ac o s ela ed o expe ience, mo i a ion, aining and a en ion. (32)
Un o una ely, he unde lying mechanisms o loudness pe cep ion a e no ye
ully unde s ood. A common assump ion is ha he in ensi y is in some way
ela ed o he o e all neu onal ac i i y e oked by a sound. In his scena io, he
in ensi y o a sinusoidal one would be de e mined no only by he ac i i y o
neu ons wi h cha ac e is ic equencies (CF) nea he one equency, bu also
by he deg ee o ac i i y o adjacen CF neu ons. In o he wo ds, he olume
may depend on a sum o neu onal ac i i y h ough di e en equency
channels (known as c i ical bands).
Models inco po a ing his basic concep ha e been p oposed by Fle che and
Munson, Zwicke and Scha and, mo e ecen ly, by Moo e e al. The models
p oposed by Moo e a e illus a ed in igu e 4.3. The i s s age is a ixed il e
o simula e he ansmission o sound h ough he ou e and middle ea . The
nex s ep is o calcula e a pa e n o exci a ion o sound s udio. Then, he e is
a ans o ma ion om he exci a ion le el (in dB) o he speci ic loudness,
which is a kind o "sound densi y" ha ep esen s he olume pe c i ical band.
This ans o ma ion in ol es a non-linea comp ession; o example, a 10- old
inc ease in he sound in ensi y, which co esponds o an inc ease o 10 dB,
causes less han a 10- old change in speci ic loudness. Al hough he models
a e based on psychoacous ic da a, his ans o ma ion can be unde s ood as
ep esen ing he way he physical exci a ion is ans o med in o neu al ac i i y.
The speci ic loudness is p obably ela ed o he amoun o neu al ac i i y in

In oduc ion
!
!
!
32!
he co esponding CF. The comp ession in he ans o ma ion pa ly e lec s
wha occu s in he basila memb ane. The o e all olume o a gi en sound, in
sones, is assumed o be p opo ional o he o al a ea unde he speci ic
pa e n o loudness. One could hink o his a ea as app oxima ely
p opo ional o he o al neu onal ac i i y e oked by a sound. Sound models o
his ype ha e been qui e success ul a explaining he expe imen al da a on
he sono i y o simple and complex sounds. (30) (33) (34)
Figu e 4.3. Basic s uc u e o Moo e e al. loudness model. (Own p oduc ion)
S imulus!
Fixed!=il e !
o ! ans e !
o !ou e /
middle!ea !
T ans o ma i
on!o !
spec um! o!
exci a ion!
pa e n!
T ans o ma i
on!o !
exci a ion! o!
speci=ic!
loudness!
Calcula ion!o !
A ea!unde !
speci=ic!
loudness!
pa e n!
In oduc ion
!
!
!
33!
5.- COCHLEAR IMPLANTS
Cochlea implan s (CI) a e de ices designed o es o e hea ing o p o oundly
dea people. Hea ing aids a e a common solu ion o people wi h mode a e
hea ing h esholds; howe e , o highe h esholds (~ 90 dB), hese do no
wo k as success ully. In hese si ua ions, a cochlea implan is used o bypass
he middle and inne ea and di ec ly s imula e he audi o y ne e om wi hin
he cochlea.
Figu e 5.1. Main componen s o a cochlea implan : a) speech p ocesso ; b) ansmi e ; c)
ecei e /s imula o ; d) elec ode a ay. (Own p oduc ion using ee image om Na ional
Ins i u e on Dea ness and O he Communica ion Diso de s a he Na ional Ins i u es o
Heal h)
Figu e 5.1 shows he main pa s o a cochlea implan . Mos mode n sys ems
include ou componen s: a po able ex e nal sound p ocesso , a ansmi e ,
an implan ed ecei e /s imula o and a se o in acochlea elec odes. The
sound p ocesso consis s o a mic ophone ha collec s he sound signal; an
a!
b!
c!
d!
In oduc ion
!
!
!
34!
elec onic p ocesso , gene ally in he o m o a digi al signal p ocesso (DSP),
ha encodes i in o elec ical pulses. The signal hen goes o he adio
equency ansmi e , which sends i o he anscu aneous ecei e /
s imula o h ough a ansmi e coil. This ansmission is done h ough
induc ance (magne ic ield a ia ions), which induces al e na ing cu en
pulses in he ecei e coil. The ecei e / s imula o is a su gically implan ed
elec onic de ice. I ecei es and decodes signals om he sound p ocesso
and gene a es elec ical signals o selec i ely ac i a e he in acochlea
elec odes. Small elec ic cu en pulses a e sen o hese elec odes o
s imula e ne e cells in he pe iphe al audi o y sys em, he eby causing
audi o y sensa ions. The elec odes a e placed in he o m o an a ay, which
is made up o 16 o 24 con ac s (depending on he de ice). They a e 12.1 o
27 mm away om each o he depending on he manu ac u e . Each elec ode
s imula es a di e en si e in he cochlea, which co esponds o a equency
band ma ching he cochlea ono opy. Thus, mainly a limi ed popula ion o
ne e ib es pe cei es he s imulus. This allows di e en sec ions o he
cochlea o be independen ly s imula ed, mimicking and bypassing he basila
memb ane equency il e ing unc ion. (35) (36)
Sound is il e ed in he speech p ocesso by a bank o bandpass il e s ha do
no o e lap, c ea ing a se o channels ha con ain in o ma ion om di e en
egions o he sound equency spec um. Each o hese channels is able o
send in o ma ion o one elec ode o he a ay. The esolu ion o a heal hy
human audi o y sys em is 1 Hz in he 100 Hz egion. This would be he
equi alen o ha ing 100 channels om 100 o 200 Hz. Ins ead, cochlea
implan s ha e 16 o 24 channels o he whole audible equency ange. This
numbe depends on he p ocessing algo i hm o each speci ic cochlea
implan manu ac u e . (37)
5.1- Signal p ocessing
The mos impo an CI uni is he signal p ocessing s a egy used o he
ans o ma ion o he elec ical s imula ion signal. The di e en signal
In oduc ion
!
!
!
35!
p ocessing echniques ha ha e been de eloped can be classi ied acco ding
o he sound s imuli in o ma ion ha hey use: 1) he wa e o m, 2) he
en elope, and 3) he spec al cha ac e is ics ( ine s uc u e) o he signal.
The la es speech p ocesso s a e based on he channel ocode p inciple,
which is used in elephone communica ion wi h much less bandwid h han ha
equi ed o ansmi he aw speech signal. A channel ocode consis s o an
analyse and a oice syn hesize . CI speech p ocesso s use he signal
analysing block o he ocode . As shown in igu e 5.2, a ocode analyse
il e s he incoming speech signal in o a numbe o con iguous equency
channels using a bank o bandpass il e s. The ou pu o each il e is hen
passed h ough an en elope de ec o , which consis s o a ull-wa e ec i ie
and low-pass il e . This way, an es ima ion o he ene gy o each band is
ob ained as well as i s e olu ion in ime. In addi ion, he analyse makes he
decision o alloca e sound o ha channel o no , and i also es ima es i s one
(F0). The dynamic ange (DR) adap a ion block ans o ms he acous ic DR
o each channel in o he elec ical DR (de ined as he span o in ensi ies
going om h eshold o maximum com o able le el) necessa y o each
elec ode. This ans o ma ion is speci ic o each pa ien and is di e en o
each elec ode. Finally, acco ding o he s imula ion a e (i.e. he numbe o
pulses deli e ed o elec odes pe second), he p ocesso gene a es
s imula ion pulses ep esen ing he cu en le el o be sen o each elec ode
a each ime ins an . In he majo i y o s a egies, he s imula ion pulses a e
gene a ed in a way such ha , a each momen , he e is only one channel
ac i e, o a oid o e lapping sensa ions caused by adjacen elec odes
wo king a he same ime. The s imula ion pa e n compu ed by he p ocesso
is ansmi ed o he CI ecei e and he cu en pulses a e hen sen o he
elec odes.
The pe o mance o cochlea implan s depends g ea ly on se e al ac o s
such as he pa ien 's age, du a ion and causes o dea ness, he numbe o
su i ing spi al ganglion cells, he numbe o unc ional elec odes, he speech
p ocessing s a egy and so on.
In oduc ion
!
!
!
42!
5.4- Loudness pe cep ion wi h Cochlea Implan s
In designing a speech p ocesso o an implan lis ene , one o he mos
impo an ac o s lies in he p ope ans o ma ion o acous ic ampli ude in o
elec ic ampli ude. No mal acous ic hea ing can p ocess sounds o e a ange
o 120 dB. Speech sounds in a no mal con e sa ion can ange om 40 o 60
dB. Howe e , implan lis ene s ypically ha e dynamic anges o only 6 o 30
dB. Thus, he no mal acous ic hea ing ange mus be conside ably
comp essed o elec ic s imula ion, in o de o p ese e he acous ic
ampli ude a ia ions ha ansmi impo an speech in o ma ion. (53)
Loudness in elec ically s imula ed human lis ene s has been measu ed
quan i a i ely using a ing me hods and magni ude es ima ion echniques.
Howe e , he exis ing da a a e no consis en ac oss s udies. P ings ’s g oup
showed ha loudness unc ion depends on s imulus equency o simple
sinusoidal and pulsa ile s imuli in cochlea implan s. They ound ha loudness
was an exponen ial unc ion o s imulus ampli ude o high equencies (> 300
Hz) and a powe unc ion o low equencies (< 300 Hz). Some in es iga o s
ha e ound ha log loudness g ows linea ly as a unc ion o log cu en ,
al hough some da a show wo s ages, wi h loudness ha g ows g adually a
low s imula ion le els and hen g ows s eeply a highe s imula ion le els
( igu e 5.5). O he g oups ound ha loudness g ows linea ly as a unc ion o
cha ge. The e has no ye been a widely accep ed quan i a i e desc ip ion in
implan esea ch o he ela ion be ween loudness and elec ic s imulus le el.
I is no clea whe he he disc epancies among he abo e-men ioned
loudness unc ions we e due o p ocedu al di e ences o due o indi idual
implan lis ene di e ences. (54) (55) (56) (57) (58)

In oduc ion
!
!
!
43!
Figu e 5.5. G ow h unc ion o discha ge a e e sus elec ical pulse in ensi y. [19]
!!!
44!
!!!
45!
II. EXPERIMENTAL WORK
!!!
46!
Expe imen al Wo k
!
!!!
47!
6.- PSYCHOACOPHYSICS: SPEECH PERCEPTION IN NOISE
STUDIES
In e e yday lis ening si ua ions, accu a e speech pe cep ion elies on he
capaci y o he audi o y sys em o p ocess complex sounds in he p esence o
backg ound noise. The e is ample e idence o audi o y aining esul ing in
pe cep ual enhancemen s. Howe e , he e ha e been su p isingly ew
in es iga ions o how aining impac s speech-in-noise pe cep ion. E en less
has been done in he ield o how high equencies a ec his abili y.
In his chap e , wo s udies a e desc ibed. The i s ocuses on he physical
composi ion o sound, as a way o enhance speech- ecogni ion in noise. Mo e
speci ically, i in es iga es how speech’s high equency con en a ec s i s
unde s anding in noisy en i onmen s. The second s udy deals wi h how
aining can enhance his abili y e en when sound has been band- equency
limi ed.
6.1.- E ec s o High-F equency Supp ession o Speech Recogni ion in
Noise in Spanish No mal-Hea ing Subjec s
(Published as: de Miguel, ÁR, Zaballos, MTP, Macías, ÁR, Ba ei o, SAB,
González, JCF, Plasencia, DP (2015). E ec s o High-F equency Supp ession
o Speech Recogni ion in Noise in Spanish No mal-Hea ing Subjec s. O ol
Neu o ol, 36(4), 720-726.)
I is common o pa ien s o ask why hey hea bu ail o unde s and o why
hey ha e ouble unde s anding oices. Pa o he answe has o do wi h
one’s hea ing and pa wi h one’s cogni i e p ocessing o he wo d. In his
s udy, a en ion is ocused on he i s one. Hea ing loss can a ec he abili y
o unde s and he oices, especially when compe ing noise is p esen . Wi h
hea ing loss, i isn’ abou no hea ing sounds a all. I is abou hea ing jus
pa s o he sounds, maybe e en jus he owels, which ha e lowe

Expe imen al Wo k
!
!!!
48!
equencies. To illus a e his, le ’s ake he sen ence: high equencies a e
impo an o unde s and speech in noise. Fo someone wi h a mode a e
hea ing loss in he highe equencies, his migh sound app oxima ely like
his: _i _ __e_ue__ie_ a_e i_o_a__ _o u__e__a__ __ee_ i_ oi_e. One would
p obably hea he wo ds bu no unde s and he meaning. Pa ien s can
some imes complain abou hese p oblems bu ha e an appa en ly no mal
audiome y. Thus, he aim o his s udy is o in es iga e whe he equencies
abo e 8 kHz,-which a e beyond he scope o a clinical audiome y and ou side
he bandwid h o ope a ion o hea ing p os hesis-, can ha e a signi ican
e ec on speech pe cep ion in noise.
6.1.1- In oduc ion
A heal hy human ea can pe cei e equencies om 20 o 20000 Hz. This
ange a ies be ween indi iduals and deg ades g adually wi h age, especially
o high equencies. Since he peak ene gy o ica i es spoken by emale
and child alke s ends o occu be ween 6.3-9 kHz, subjec s wi h hea ing loss
(HL) in his ange will ha e a limi ed audibili y o such phonemes. Mo eo e , i
he e is impo an in o ma ion con ained in he high equency end o he
human hea ing spec um, hen hey will su e om an impo an loss o
in o ma ion. In mild, mode a e and high equency hea ing losses he only
symp om may be sub le di icul y wi h wo d unde s anding, especially in
si ua ions whe e he e is compe ing noise. (59) (60) (61) (62) (63)
A se ies o s udies by S elmachowicz e al. epo ed ha when he high
equency componen s o a sound s imulus a e emo ed, he pe cep ion o he
phonemes /s/ and /z/ a e se iously comp omised in child en wi h No mal
Hea ing (NH) and wi h hea ing loss (HL). They in es iga ed he e ec s o low-
pass il e ing on he pe cep ion o /s/ in ou g oups o lis ene s: child en and
adul s wi h NH and child en and adul s wi h HL. Tes s imuli we e p oduced by
an adul emale, amale, and a child and we e low-pass il e ed a a ious
equencies om 2 o 9 kHz. In gene al, child en wi h NH pe o med mo e
poo ly han he adul s, andsubjec s wi h HL pe o med mo e poo ly han
Expe imen al Wo k
!
!!!
49!
hei coun e pa s wi h NH. Impo an ly, in bo h cases mean pe o mance o
he emale and child alke s imp o ed up o a bandwid h o 9 kHz. These
esul s sugges ha he pe cep ion o ica i es may be di icul o lis ene s
wi h HL, due o he limi ed bandwid h o cu en hea ing p os heses.
None heless, he au ho s sugges ha he e ec s o bandwid h may be ask
dependen . Since adul s wi h pos -lingual hea ing loss a e gene ally able o
use seman ic and syn ac ic cues in he pe cep ion o speech, he acous ic
e ec s o emo ing high equencies in speech may be masked in his age
g oup. Howe e , young child en who a e s ill lea ning speech and language
may be mo e a ec ed. Thus, i is impo an o es ablish how complex audi o y
condi ions such as backg ound noise can a ec his abili y o adul s o
compensa e o a sho e audible equency bandwid h. (63) (64) (65)
A numbe o au ho s ha e also s udied he in luences o high equency on
sound quali y pe cep ion in a ious audi o y si ua ions. Gab ielsson e al.
showed ha he pe cei ed sound quali y o a emale oice depended on he
audible equency bandwid h and on he smoo hness in equency esponse
o he ep oduc ion o ampli ica ion sys em. Moo e and Tan also epo ed
signi ican deg ada ion in speech sound quali y when he cu -o equency
dec eased below 10.8 kHz. (66) (67)
These indings a e also o i al impo ance in he a ea o Cochlea Implan
(CI) de elopmen . Because such de ices ha e a limi ed bandwid h o a ound
8kHz, hei use s canno pe cei e audi o y in o ma ion om equencies
abo e his alue. In spi e o his, he majo i y o CI use s ha e e y good
esul s in speech ecogni ion es s in quie . Howe e , he same si ua ion does
no apply in mo e complex audi o y condi ions, such as in he p esence o
backg ound noise. To ge a ound his issue, CIs use a a ie y o s a egies
and noise il e s, bu e en hough some bene i s ha e been epo ed, hei
pe o mance is s ill no sa is ac o y. (68) (69) (70) (71) (72) (73) (74)
Expe imen al Wo k
!
!!!
50!
6.1.2- Objec i es, hypo hesis and a ionale o he s udy
In o de o unde s and why subjec s wi h HL ha e di icul y in unde s anding
speech in noise, i is i s necessa y o unde s and he p ocesses allowing
subjec s wi h NH o di ec a en ion o a desi ed sou ce and comp ehend i .
Howe e , he neu al mechanisms ha suppo speech disc imina ion in noise
a e no well unde s ood. The e o e, his s udy aims o e alua e how high
equency pe cep ion con ibu es o speech pe cep ion in noise.
To his end, an expe imen was designed o s udy speech unde s anding
deg ada ion as a unc ion o maximum s imulus equency in subjec s wi h
NH.
The hypo hesis o his s udy is ha high equency il e ing o he sound
en i onmen (wo ds and noise) abo e 8 kHz causes a signi ican
dec ease in wo d ecogni ion in noise.
The mo i a ion o his s udy has come mainly om he ealiza ion ha
hea ing p os heses only each 8 kHz. Which means ha any sound abo e i is
no ampli ied (in he case o hea ing aids) o hea d a all (in he case o CI).
Despi e he ine spec al and e en empo al esolu ion ob ained o low
equencies in he ield o CI, which co espond o he undamen al
componen s o speech, pa ien s s ill s uggle wi h unde s anding in noise
si ua ions. The e o e, he cause migh no be es ic ed o he low equency,
bu also o he high equency ange. I NH subjec s do make use o his
in o ma ion, hen i would be in e es ing o al e CI design o y and ind a way
o b oaden hei equency ange.
6.1.3- Subjec s and Me hods
This obse a ional c oss-sec ional s udy was unde aken a he Hea ing Loss
Uni in he O ola yngology Head Neck Depa men , Uni e si y Hospi al o
Expe imen al Wo k
!
!!!
51!
G an Cana ia (Insula -Ma e no In an il) (Uni e sidad de Las Palmas de G an
Cana ia). The co esponding E hical Commi ee app o ed his s udy.
Subjec s
A o al o 29 subjec s we e selec ed, aged 19-55, wi h an a e age age o 30.
All we e na i e Spanish speake s. The subjec s had no mal hea ing wi h pu e
one h esholds be e han 20 dB HL in all equencies s udied and wi h no
o he known pa hology. None o hem had p io expe ience wi h acous ic
simula ions o had pa icipa ed in any hea ing esea ch expe imen . All
subjec s came olun a ily and ga e hei in o med consen .
Speech ma e ial and backg ound noise
To s udy he e ec o high equency emo al in speech ecogni ion in noise,
di e en amendmen s o he Spanish disyllabic speech lis s we e made.
F equency modi ica ions and noise iles we e gene a ed using MATLAB and
S a is ics Toolbox Release 2012b (The Ma hWo ks, Inc., Na ick,
Massachuse s, Uni ed S a es).
Six alida ed lis s o 25 wo ds each and spoken by a emale speake wi h a
Spanish accen we e used (Appendix II). They belong o he s anda d Spanish
Disyllabic es , commonly used in hea ing-loss assessmen s. Two g oups
we e made o h ee lis s each. In he i s g oup, equency componen s we e
unal e ed and in he second g oup he wo ds we e band-pass il e ed. Fo
u u e e e ence, we will e e o hem as g oups A and B espec i ely. The
audio iles we e sa ed in Wa e o m Audio File Fo ma (.Wa ) o ma wi h a
sampling equency (Fs) o 44 kHz. Fs de e mines how many imes pe
second he analogue signal is digi ally sampled by he eco ding sys em.
Acco ding o he Sampling Theo em, o a band-limi ed signal o be
econs uc ed ully, his mus be done a wice he a e o he highes
equency ha i is o in e es o eco d. The e o e, he maximum sound
equency ha could be ep oduced wi h his sampling a e was 22 kHz. (75)
Expe imen al Wo k
!
!!!
58!
indi idual wo ds. To examine hese issues, co ela ions be ween he amoun
o al e a ion in a wo d and i s mean sco e we e analysed o g oup B lis s ( he
il e ed ones). The ollowing hypo hesis was e alua ed: wo ds whose high
equency spec um had su e ed he highes changes ga e lowe sco es.
Table 6.1 summa izes he esul s o he Pea son co ela ion o each lis . The
only s a is ically signi ican co ela ion was ound o SNR = 5 ( p = 0.037). In
he emaining cases, no co ela ion was ound be ween hese wo a iables (p
= 0.487 o 0 dB SNR and p = 0.171 o -5 dB SNR).
PEARSON CORRELATIONS
5 SNR
Pea son Coe icien
Sig
0.363
0.037
0 SNR
Pea son Coe icien
Sig
-0.007
0.487
-5 SNR
Pea son Coe icien
Sig
-0.198
0.171
Table 6.1. Pea son co ela ion esul s o he compa ison o au oco ela ion o wo ds p e and
pos il e ing and hei o al sco e. The only signi ican co ela ion was ound o he case 5 dB
SNR. (Own p oduc ion)

Expe imen al Wo k
!
!!!
59!
6.2.- E ec s o speech-in-noise aining and high equency supp ession
in Ai T a ic Con olle s. A con ol g oup s udy
(Published as: Zaballos, MTP, de Miguel, ÁR, Plasencia, DP, González, MLZ,
Macías, ÁR (2015). E ec s o long- e m speech-in-noise aining in ai a ic
con olle s and high equency supp ession. A con ol g oup s udy. J In Ad
O ol, 11(3), 212-7.)
Lea ning elec ically s imula ed speech pa e ns is a new and di icul
expe ience o many CI use s. I does no su ice o ha e been gi en he ools
o hea . These pa ien s equi e ex ensi e ehabili a ion and audi o y aining o
be able o make use o he in o ma ion p o ided by he de ice, iden i y
di e ences in sound pa ame e s and o make sense o hem. Speci ically,
speech-in-noise-unde s anding s ill p esen s a g ea challenge o mos CI
use s. Gi en ha CI and hea ing aids alone do no es o e hese abili ies,
e o s a e cu en ly ocused on he ole o he b ain in such si ua ions and
whe he aining can be an e icien he apy. In his s udy, ai a ic con olle s
ha e been s udied because hey seem o ha e de eloped a g ea e abili y o
unde s and speech in noise han he a e age indi idual. This abili y is a esul
o hei con inuous and long- e m audi o y aining in noise unde es ic ed
equency bands. This is an encou aging disco e y o he ield o cogni i e
sciences and audi o y ehabili a ion. Mo eo e , he e ec o high equency
supp ession o speech-in-noise unde s anding obse ed in he p e ious
s udy has been u he assessed o his job sec o o e alua e possible
lea ning e ec s.
6.2.1- In oduc ion
The abili y o unde s and speech in noise equi es bo h senso y and cogni i e
skills (78). The senso y pa consis s o he audi o y sys em locking on o he
a ge speech signal while excluding ambien noise and compe ing oices.
This is achie ed by o ganising audi o y inpu s in o di e en g oups by
iden i ying sha ed cha ac e is ics, such as loca ion and acous ical simila i y.
The iden i ica ion o hese g oups is gi en by he ela i e s abili y o oice
Expe imen al Wo k
!
!!!
60!
pi ch and helps in g ouping i sepa a ely om o he oices. In addi ion, o he
signal-based cues like ha monics, loca ion and iming, aid in g oup o ma ion
o speech. On he cogni i e side, a en ion and wo king memo y skills a e key
o a good speech-in-noise (SIN) abili y. (79) (80) (81)
The e is e idence ha such skills can be imp o ed du ing an adul ’s li e,
causing pe cep ual enhancemen s, plas ici y in single neu ons and in neu al
popula ions (82) (83). S udies o audi o y pe cep ual lea ning e eal long- e m
neu al changes in he adul audi o y co ex (AC) o bo h animals and humans
a e in ensi e audi o y aining (84) (85). Cu en models p opose ha
pe cep ual lea ning in adul s depends s ongly on op-down in luences such
as a en ion, ewa d, and ask ele ance. This demons a es ha he adul AC
is a dynamic and adap i e p ocessing cen e. (86) (87) (88) (89) (90) (91) (92)
(93) (94) (95)
Na u al audi o y aining occu s in job sec o s wi h equen daily adio
communica ions. This is he case o Ai T a ic Con olle s (ATC). People ha
wo k in his ield a e cons an ly exposed o SIN, which is ex emely ha d o
unde s and o an ou side . Du ing hei wo king li e, ATC lea n o iden i y,
ex ac and comp ehend con e sa ions embedded in whi e noise. These
con e sa ions a e no mally es ic ed o he ae onau ics ield, which makes
hem an ideal popula ion o s udy he e ec s and ans e abili y o long- e m
audi o y aining.
Un o una ely, he majo i y o s udies on how aining a ec s SIN pe cep ion
used small s imulus se s and e ealed ha , while SIN pe cep ion can imp o e
in a i icial lis ening condi ions, he bene i can be ex ended o un ained
ma e ial is no clea . (96) (97) (98) (99)
6.2.2- Objec i es, hypo heses and a ionale o he s udy
As opposed o p e ious wo k whe e he pa ien s a e ained in si u and hen
es ed, his s udy looks a how someone's job based lis ening expe ience can
Expe imen al Wo k
!
!!!
61!
lead o enhanced pe o mance on a labo a o y-based es . Mo eo e , he ole
o high equencies has also been e alua ed because adio communica ions
a e limi ed, jus like hea ing p os hesis, o 8 kHz.
Thus, his s udy has wo main objec i es. The i s is o e alua e whe he Ai
a ic Con olle s enhanced abili y o unde s and speech in noise could
be es ed in a speech-in-noise labo a o y es . To his end, a con ol g oup
o no mal hea ing lis ene s and a a ge g oup o ATC we e assessed. The
second is o compa e hei speech in noise pe o mance in high
equency (>8kHz) il e ed and non- il e ed condi ions o ha o no mal
hea ing indi iduals om he p e ious s udy (con ol g oup).
The hypo heses a e: 1) ha ATCs unde s and speech in noise
signi ican ly be e han an o dina y no mal hea ing indi idual; 2) he
di e ences be ween il e ed and non- il e ed wo d ecogni ion a e lowe
han hose o he con ol g oup. The i s hypo hesis is p oposed on he
basis ha hese indi iduals ha e unde gone speech-in-noise aining du ing
adul hood. The second is cons uc ed on he ac ha hei aining ma e ial
in ol ed high equency il e ed adio communica ions (>8 kHz).
The eason behind his s udy is ha ou aging socie y is bound o su e
speech-in-noise-unde s anding impai men s. As indi iduals g ow olde , his
abili y deg ades e en be o e audi o y losses become clinically ele an .
The e o e, i is capi al o come up wi h s a egies o y and slow his p ocess
down as much as possible. Gi en he ecen e idence ha lea ning du ing
adul hood is a g ea e han p e iously hough , he demons a ion ha
audi o y aining unde deg aded sound quali y can lead o be e speech in
noise unde s anding can ha e impo an implica ions in he way we ea
hea ing losses oday. An e ec i e ehabili a ion p og am can imp o e
people’s quali y o li e by easing hei in eg a ion back in o he sound
complexi y o socie y.
Expe imen al Wo k
!
!!!
62!
6.2.3- Subjec s and Me hods
This s udy was done a he Ai Base o Gando Ai po o G an Cana ia and
he Psychoacous ics Labo a o y o he Hea ing Loss Uni , O ola yngology
Head and Neck Depa men , Complejo Hospi ala io Uni e si a io Insula
Ma e no In an il de G an Cana ia (CHUIMI). The E hical Commi ee app o ed
his s udy.
The me hods o his s udy a e analogous o hose o he p e ious s udy. So
only he adds-on will be desc ibed.
Subjec s
A o al o 29 no mal-hea ing subjec s (aged 19-55) om he p e ious s udy
we e used as con ol g oup and 48 ATC (aged 34-56), as he a ge g oup. All
subjec s we e na i e Spanish speake s. The subjec s had NH wi h pu e one
h esholds be e han 20 dB HL as measu ed by a alida ed clinical
audiome y ( om 0.5 o 8 kHz in s eps o 0.5 kHz) and wi h no o he known
pa hology. None o hem had p e iously aken pa in any audi o y
expe imen s be o e and hey all signed an in o med consen be o e
commencemen o he s udy.
Age his og ams o he ATC and con ol g oups we e illus a ed in igu es 6.6
and 6.7, and yea s wo king in he sec o o ATC in igu e 6.8.
Expe imen al Wo k
!
!!!
63!
Figu e 6.6. ATC age his og am.
Figu e 6.7. Con ol g oup age his og am.

Expe imen al Wo k
!
!!!
64!
Figu e 6.8. Yea s wo king as ATC his og am.
Speech ma e ial and backg ound noise
The speech ma e ial was he same as o he p e ious expe imen .
P ocedu es
The expe imen p ocedu e is analogous o ha o he p e ious chap e .
The da a was eco ded in an .xls ile, Mic oso Excel 2011 (Mic oso .
Mic oso Excel (Redmond, Washing on, Uni ed S a es) and IBM SPSS
S a is ics o Macin osh, V. 21.0. (IBM Co po a ion, New Yo k, Uni ed S a es)
was used o he s a is ical ea men . The s a is ical analysis s a s wi h a
no mali y check o each speci ic condi ion using he Kolmogo o -Smi no es
o one sample. Acco ding o he esul , a non pa ame ic o a wo ailed -
s uden es was pe o med. Finally, co ela ions be ween age, yea s o
Expe imen al Wo k
!
!!!
65!
expe ience in he Ae onau ic ield and es sco es we e analysed using a
Pea son co ela ion es . The le el o signi icance chosen was 5%.
6.2.4- Resul s
Fi s , o e all pe o mance di e ence be ween he wo g oups was analysed.
Figu e 6.9 shows how he ATC g oup ou pe o ms he con ol g oup in bo h
condi ions, - il e ed and non- il e ed-, and o all SNR condi ions. Mo eo e ,
he di e ence be ween hem becomes la ge as noise inc eases. No mali y
check ga e Gaussian dis ibu ions. The e alua ion o signi icance e u ned p-
alues < 0.01 o all cases, hus e ealing signi ican mean di e ences
be ween he wo g oups. Fo SNR 5 dB, his di e ence was 7.27% o lis A
and 13.25% o lis B. Fo SNR 0 dB, hese alues ise o 16.95% and
11.45%. The mos in e es ing di e ence is obse ed a SNR -5 dB. In his
case, ATC ou pe o m he con ol g oup by 17.03% and 22.52%.
Figu e 6.9. Success o ATC and con ol g oups as a unc ion o he es . The ATC g oup
ou pe o ms he con ol g oup in all SNR condi ions and bo h in non- il e ed (A=22 kHz) and
il e ed (B=8 kHz) condi ions.
Expe imen al Wo k
!
!!!
66!
Figu es 6.10 and 6.11 illus a e he ela ion be ween yea s wo king as ATC
and es pe o mance and be ween age and es pe o mance. The co ela ion
s udy be ween age, yea s wo king in his p o ession and o e all es esul s
o each condi ion is shown in Table 6.2. No signi ican co ela ion was ound
be ween wo king expe ience and o e all es esul s. A de ailed analysis o
each case e ealed signi ican (p<0.05) in e se co ela ions be ween 8 kHz 5
SNR es esul s and bo h a iables. 22 kHz 5 SNR also ga e a signi ican ,
in e se link be ween age and es esul s (Table 6.3). Howe e , hese esul s
do no ha e a physical jus i ica ion.
Figu e 6.10. Success a e as a unc ion o yea s wo king as ATC.
Expe imen al Wo k
!
!!!
67!
Figu e 6.11. Success a e as a unc ion o age o he ATC g oup.
Age
Yea s wo king as an ATC
Tes esul s
Age
Pea son Co ela ion
Sigma (bila e al)
N
-
0.759
0.000
288
-0.068
0.249
288
Yea s wo king as an ATC
Pea son Co ela ion
Sigma (bila e al)
N
-
0.002
0.973
288
Tes esul s
Pea son Co ela ion
Sigma (bila e al)
N
-
Table 6.2. Pea son co ela ion o age, yea s wo king as ATC and o e all es esul s. No
co ela ion was ound be ween es s esul s and ei he age o expe ience. Ob iously, in he
as majo i y o cases, he olde he ATC, he longe hey ha e wo ked, so he e is a
co ela ion be ween hese wo a iables.
Expe imen al Wo k
!
!!!
74!
Figu e 7.1. Case o Use Diag am. The pa ien in e ac s wi h he esea che by p o iding
answe s o each o he es s eps: T and C le el calcula ion (MCL and THL, espec i ely),
loudness balance and elec ode disc imina ion es .
Communica ion be ween he use in e ace in Visual S udio and he NIC
Py hon lib a y
Because he NIC lib a y was designed o Py hon, bu he use in e ace was
buil in Visual S udio, i is necessa y o de elop a communica ion and
depu a ion me hod be ween hese wo pla o ms. To his end, a ile-based
communica ion sys em was deployed. The da a low om Visual S udio o
Py hon indica es he s imulus pa ame e s: pulse gap, pulse wid h, a e, phase
gap, in ensi y and du a ion o s imulus, and he es in p og ess. The low om
Visual S udio o Py hon gi es in o ma ion abou he connec ion s a e o he
de ice o he compu e , when i is s imula ing, e c.
The (NIC) lib a y o Py hon, de eloped by Cochlea LTD. was used o c ea e
and send elec ical s imuli o he in acochlea elec odes. This is a speci ic

Expe imen al Wo k
!
!!!
75!
lib a y buil o allow esea che s o design cus omized s imuli by changing i s
pa ame e s: pulse wid h, pulse a e, phase gap and in e -pulse gap.
To elimina e he need o p og amming, a use in e ace was buil using Visual
S udio, which communica es wi h Py hon. In u n, Py hon sc ip s con ol he
implan ecei e / ansmi e by sending ins uc ions o he supplied p ocesso .
Thus, he pla o m de elopmen has wo main pa s: 1) Py hon unc ions and
2) use in e ace de elopmen and i s communica ion wi h Py hon.
Da a Base
The da a base has been a anged in such a way ha a pa ien can be
assigned o a ange o psychoacous ic es s ha a e con olled by he
in es iga o and es se up and ou comes alloca ed o an indi idual pa ien .
Tha is, he es s a e nes ed wi hin each pa ien . In igu e 7.2, his s uc u e is
depic ed. Tes ing gene ally s a s wi h ob aining Th eshold and Com o le els
and loudness balance es ing and ends wi h al e na i e o ced choice (AFC)
expe imen s. I can be seen how o each pa ien , he use can de ine as
many s imuli as desi ed, each o which can be associa ed wi h di e en C and
T le els. In u n, each C and T le el con igu a ion can be ma ched wi h many
loudness balance es s. Finally, o each o he loudness balance
con igu a ions, an unlimi ed numbe o al e na i e o ced choice (AFC)
expe imen s can be se up.
Expe imen al Wo k
!
!!!
76!
Figu e 7.2. Da a Base schema ics. Tes s a e nes ed wi hin hei p e ious equi ed s ep, he
a iable pa ien being he main so ing a iable.
Use In e ace
The use in e ace was de eloped using Visual S udio and is hus compa ible
wi h Windows 7 and Vis a. The compu e equi emen s a e: 800 MHz 32/64
bi s p ocesso , 1GB o RAM memo y, Mic oso . NET F amewo k 4.5 and
Mic oso SQL Se e 2008 R2. The applica ion allows he design o
psychoacous ic es s whe e a ious ypes o s imula ion modes and
elec odes can be e alua ed wi hou he need o sc ip o de commands in
any p og amming language. The iles can be expo ed o an excel ile o la e
analysis.
How i wo ks
Inse Pa ien
The op ion Inse Pa ien on he S a page opens he window in igu e 7.3,
whe e demog aphic and clinical da a can be inse ed. The Com o (C) and
Th eshold (T) le el alues om he pa ien ’s s anda d map can also be
Pa ien !!
S imulus!A!
C!and!T!
le el!map!1!
Loudness!
Balance!1!
AFC!1! AFC!n!
Loudness!
Balance!n!
C!and!T!
le el!map!n!
S imulus!B!
Expe imen al Wo k
!
!!!
77!
inse ed. Finally, he loca ion o he elec odes wi h espec o he modiolus
can also be ca alogued as pe imodiola , mid-modiola and la e al wall.
On his window, and all subsequen ones, he e is an indica o o he
communica ion s a e be ween he applica ion and he p ocesso
(connec ed/no connec ed) o sa e y easons. I will show whe he he
compu e co ec ly de ec s he p ocesso o no .
Figu e 7.3. Inse Pa ien . He e demog aphic, clinical da a and elec ode posi ion wi h espec
o he modiolus can be egis e ed.
S imulus De ini ion
Once he pa ien has been sa ed, he Exis en Pa ien window eplaces he
New Pa ien , whe e he same in o ma ion is displayed, bu now he use can
selec he di e en es s o pe o m he s udy.
A s imulus mus be de ined be o e he expe imen can p oceed. The De ine
S imulus bu on opens a window, igu e 7.4, ha allows cus omiza ion o he
ollowing pa ame e s: s imula ion mode, s imulus du a ion, in e -s imulus
in e al, pulse phase and gap du a ions and pulse equency. The s imulus
Expe imen al Wo k
!
!!!
78!
consis s o a ain o elec ical biphasic pulses a a equency speci ied by he
pulse a e op ion. S imula ion mode e e s o he ype o g ound elec ode
used: he e e ence elec ode loca ed in he mas oid egion (1), he in e nal
p ocesso case (2) o a combina ion o he wo (1+2). The in e -s imulus
in e al e e s o he es ing ime be ween wo s imuli. These se ings will
emain ixed o all es s o he s udy.
Figu e 7.4. S imulus de ini ion window. PPS: pulses pe second; S imula ion mode: g ound
elec ode loca ion (1 o g ound elec ode in he mas oid, 2 o in e nal p ocesso case and
1+2 o a combina ion o bo h); # PPS: numbe o pulses pe second (de ines s imulus
du a ion); silence: de ines ime lapse be ween pulse ains.
Dynamic Range Map
The Dynamic Range (DR) o an elec ode de ines he cu en in ensi y span
ha p oduces audi o y sensa ion and is de e mined by he T and C le els o
each elec ode. The window can be seen in Figu e 7.5. The T le el is
calcula ed using he classical ascending and descending me hod o limi s
(116): a s imulus is p esen ed a p og essi ely inc easing in ensi y un il he
pa ien epo s hea ing sensa ion and hen i is lowe ed down again using
smalle in ensi y s eps un il i is no longe hea d. To ind he C le el, he
s imulus is p esen ed in p og essi ely smalle s eps as he in ensi y ises un il
Expe imen al Wo k
!
!!!
79!
he pa ien epo s ha he sound is no longe com o able because i is oo
loud, bu ne e pain ul.
In his pa o he es , a new map can be c ea ed o a p e ious session can
be loaded by selec ing one om he Load Session sc oll. When wo di e en
s imuli a e used in an expe imen , wo independen maps mus be c ea ed
using one s imulus a a ime.
Figu e 7.5. Dynamic Range Map. C and T le els o each elec ode a e de e mined using he
up-down me hod.
Loudness Balance
The second s ep is o loudness balance all elec odes so ha he s imuli
deli e ed o all elec odes a e pe cei ed as equally loud. In he Loudness
Balance window shown in igu e 7.6, he use can compa e an elec ode wi h
21 o he elec odes. In addi ion, one can selec he s imulus o be used (A o
B). The use mus selec he pe cen age o DR in ensi y ha he ini ial
e e ence elec ode will ha e. This will be he elec ode ha will be objec i ely
ixed a he desi ed pe cen age, and all o he s will be pe cep ually balanced

Expe imen al Wo k
!
!!!
80!
wi h espec o i . The numbe o imes one elec ode will be compa ed o
ano he ( epe i ions) can also be selec ed.
The me hod used is he con luence me hod. The pa ien is p esen ed wi h wo
s imuli, he i s co esponding o a e e ence elec ode, and he second o a
es elec ode. The la e begins andomly abo e o below 10% o he DR
pe cen age selec ed p io o es commencemen . The pa ien selec s which
one o he wo s imuli sounds loude using he window in igu e 7.7 and he
in ensi y o he es elec ode will be lowe ed o aised un il con luence is
achie ed. Then he pa ien has o s a e ha hey sound equal using he
app op ia e bu on. The s imuli will be epea ed wi h he same se ings o
con i m ha he pa ien pe cei es hem as equal and hen he sys em will
au oma ically s a again bu wi h he es elec ode being 10% di e en in he
opposi e sense. The es elec ode inal alue is he a e age o he wo
app oaches.
Figu e 7.6. Loudness Balance window. Adjacen elec odes a e loudness balanced using he
con luence me hod.
Expe imen al Wo k
!
!!!
81!
Figu e 7.7. Pa ien iew o elec ode loudness balance. The pa ien has o s a e which
s imulus sounds loude .
Al e na i e Fo ced Choice Expe imen o Elec ode Disc imina ion
The chosen me hod o elec ode disc imina ion is a h ee-in e al, o ced-
choice p ocedu e (3FC) ( igu e 7.8). Two o he s imuli come om a selec ed
e e ence elec ode and a hi d one om a signal elec ode. They a e
p esen ed in andom o de and each es elec ode is p esen ed he numbe
o imes s a ed in he Numbe o Repe i ions sc oll. The pa ien has a use
in e ace wi h h ee bu ons (s imuli 1 o 3) and has o selec he one ha
sounds di e en . The esea che will selec a e e ence elec ode and a
numbe o elec odes o compa e i wi h. Resul s a e gi en as pe cen age
co ec sco es.
Figu e 7.8. Al e na i e o ced choice expe imen o elec ode disc imina ion.
Expe imen al Wo k
!
!!!
82!
7.2.- Pi ch disc imina ion compa ison in CI use s using pe imodiola s
s aigh elec ode a ays
The e is g ea in e es in placing he elec ode a ays as close as possible o
he modiolus, in an a emp o achie e lowe h esholds and hence educe
excess dissemina ion o cu en ha can esul in in e ac ion be ween
channels. Physiological s udies and ma hema ical models sugges ha such
placemen educes he s imula ion cu en equi ed o ac i a e a neu onal
popula ion, inc eases he dynamic ange and esul s in mo e localized egions
o neu onal exci a ion. (117) (104) (118) (119)
Fo his eason, he aim o his s udy was o in es iga e he di e ence in
elec ical pi ch pe cep ion be ween hese wo ypes o elec ode and es i on
subjec s ca ying bo h ypes o a ays. Using he so wa e applica ion
desc ibed in he p e ious chap e , pa ien s i ed wi h he wo ypes o
elec ode a ays we e es ed o hei abili y o disc imina e elec ical pulses
om adjacen and nea -adjacen elec odes a low in ensi y le el (ma ching
25% o he DR o elec ode 11).
7.2.1- In oduc ion
Pi ch pe cep ion is known o a y egula ly and essen ially mono onically wi h
he longi udinal posi ion o he elec ode s imula ed in mos pa ien s i ed wi h
s aigh a ays. Howe e , he p edic i e model de eloped by F ijns sugges s
ha in some ci cums ances, he e is a "c oss-a ousal", whe e axons om
mo e apical a eas a e also exci ed. Thus, besides he local a ge popula ion,
hey also s imula ed i s uppe and lowe spi al egions. This would esul in
abno mally low pi ch pe cep ions, especially a high in ensi y le els. The s udy
sugges s ha his e ec would be mo e likely in a se o elec odes medially
posi ioned and beyond he basal u n. I is he e o e impo an o in es iga e
how elec ical pi ch pe cep ions a y wi h ans e se and longi udinal
s imula ion si es. (120) (121) (107)
Expe imen al Wo k
!
!!!
83!
The esul s o se e al expe imen s in which elec ic ield pa e ns o di e en
elec ode con igu a ions ha e been in es iga ed wi hin he cochlea, e eal
ha he spa ial g adien inc eases as he dis ance be ween he s imula ing
elec odes and neu onal issue dec eases. The e o e, due o he s eepe
slope o he elec ic ield, an elec ode ha is close o he neu onal endings
equi es less cu en injec ion o each he h eshold cu en discha ge. The
esul would be lowe o se s o elec odes placed p oximally h esholds. In
addi ion, elec ic ields wi h s eepe slopes sp ead less o adjacen neu onal
popula ions, which dec eases he p obabili y o in e ac ion be ween channels.
This can esul in a be e disc imina ion be ween he elec odes. (122) (123)
(118) (124) (119) (108) (125)
A numbe o s udies ha e examined he e ec s o he p oximi y o he
elec ode pacing h esholds. Recen s udies compa ing he posi ion o
pe imodiola and la e al wall elec ode a ays ha e shown lowe h esholds in
he o me . Cohen, Saunde s and Cla k also measu ed he abili y o
disc imina e in h ee subjec s wi h pe imodiola a ays. In wo o he h ee
subjec s, a po ion o he elec ode was loca ed nea he modiolus. They
showed be e pi ch disc imina ion o elec odes in his egion. These esul s
sugges a be e spa ial selec i i y and lowe h esholds o elec odes loca ed
close o he neu onal issue. (126) (127) (128) (129)
Shephe d e al. eco ded lowe elec ical audi o y b ains em esponses
(EABR) and a smalle slope o he g ow h unc ion’s ampli ude o elec odes
placed close o he modiolus and o he spi al ganglion, as opposed o
elec odes placed nea he la e al wall o he middle po ion o he scala
ympani. Liang e al. showed ha lowe h esholds we e associa ed wi h be e
spa ial selec i i y o he elec ically s imula ed audi o y ne e ibe s in ca s.
They also ound g ea a iabili y in he esponse o audi o y neu ons o
elec ical exci a ion and sugges ed ha he ne e ibe s close o he
s imula ing elec ode a e he mos selec i e. Finally, Cohen e al. desc ibed
na owe wid h exci a ion p o iles o implan ed subjec s wi h pe imodiola
a ays han o he subjec s implan ed wi h s aigh a ays, as measu ed by
Expe imen al Wo k
!
!!!
90!
dynamic ange a andom. The pa ien lis ens o he wo sounds sequen ially
and mus answe he ques ion: which sound is loude ? ", The answe being
ei he 1s o 2nd. The esponse is eco ded using he bu ons on he in e ace
o lowe o aise he signal elec ode by 1 CL un il he pa ien pe cei es hem
as equally s ong. A his poin , he same sound is eplayed o e i y ha his
is he case and hen he es is ese bu s a ing wi h he opposi e si ua ion.
Tha is, i he a ge elec ode s a ed 10% abo e 25%, his ime i will begin
below 10%. The equal loudness poin is se as he a e age o bo h
app oaches. Subsequen ly, he a ge elec ode will se e as he e e ence
elec ode o he nex adjacen elec ode. Hence, he same p ocedu e is
epea ed un il he apical end is eached and hen he p ocess s a s again
om elec ode 11 owa ds he basal end o he a ay.
The balance is se be ween he closes elec ode pai o wo easons: 1)
Disc imina ion is mos di icul a his spa ial dis ance han any o he , so hei
s imuli need o be as simila ly in ense as possible and 2) because he pa ien s
epo ed g ea di icul ies and e en inabili y o es ablish equal loudness
be ween wo e y di e en elec ical pi ches.
S ep 3: Tes elec ode disc imina ion
Elec ode disc imina ion was measu ed using he h ee-al e na i e- o ced-
choice me hod. A e selec ing he cen al ( e e ence) elec ode and i s wo
nea es neighbo s on ei he side (signals), he numbe o imes ha each
elec ode will be es ed wi h he cen al elec ode was se o h ee. Thus, h ee
elec ical s imuli we e sen o he implan , wo o which came om he same
elec ode ( e e ence) and one om he signal elec ode. They we e p esen ed
in andom o de . The pa ien had o indica e he s imulus ha sounded
di e en om he o he wo.
A e all elec odes we e es ed wi h he e e ence, he es inalizes and he
esea che had o selec he apical, mos adjacen elec ode as e e ence and
i s wo nea es neighbou s on ei he side as signals. The p ocess was

Expe imen al Wo k
!
!!!
91!
epea ed un il all elec odes had ac ed as e e ences. This way all elec odes
we e es ed agains 4 nea es neighbou s, excep o he apical and basal
mos elec odes, which could only be es ed o one side. A simila si ua ion
holds ue o he ones be o e hese, because hey ha e wo elec odes on
one side bu only one on he o he . The esul s we e gi en as pe cen age o
co ec answe s.
Imaging
Plain X- ay S en e s p ojec ions ha e been used in he pas as a s anda d o
elec ode a ay loca ion wi hin he cochlea, bu nowadays i has been
abandoned (137). Many ecen s udies conside he bene i s o high- esolu ion
Compu e Tomog aphy (CT) scanning as a mo e exac me hod o isualize
he s uc u es o he empo al bone. The ad an ages o CT scanning include
he elimina ion o espi a o y and o he mo ion a e ac s and an ob ious
imp o emen in he di e si y o issue ha can be obse ed (138). The main
disad an age o CT in he pos ope a i e assessmen o a cochlea implan is
image deg ada ion, due o pa ial olumes and he me allic a e ac s ha may
in e e e wi h he isibili y o indi idual elec odes (139). In his s udy, a
empo al bone CT scan wi h a z esolu ion o 0.062 mm was used.
To o e come his limi a ion, a compu e -based sys em was de eloped o
ob ain he dis ance be ween each elec ode in he a ay and he inne wall o
he cochlea. A MATLAB sc ip was made o ge an es ima e using wo images
o a CT scan. High-densi y (i.e me allic) objec s we e highligh ed in he i s
image, so ha he con ou s o he implan became isible. In he second one,
an a e age densi y wi h a high con as was used o see he inne wall.
Once he images we e impo ed, he i s image was used o ma k a se ies o
poin s (as many as necessa y a e c ea ed) ha ollow ei he : 1) he inne
con ou o he pe imodiola a ay, because he con ou ad ance implan has
hal -banded elec odes ha a e only loca ed in he inne side o he a ay; o
2) he middle po ion o he s aigh a ay, because in his a ay he elec odes
Expe imen al Wo k
!
!!!
92!
a e a ull ing. Once he poin s we e se , he sys em gene a ed a pa ame ic,
in e pola ing unc ion o calcula e he cu ed shape o he implan . Then he
same p ocedu e was epea ed o he second image, bu ma king he inne
wall ins ead.
Once gene a ed, he in e -elec ode dis ance speci ica ion o he manu ac u e
is used o de e mine indi idual elec ode loca ion. Then he dis ance o he
inne wall was calcula ed as he pe pendicula dis ance o he angen o he
cu e a he elec ode loca ion. The inal measu es a e displayed on he
second image (Appendix IV).
Da a analysis
Analysis o he a iables unde s udy was done using IBM SPSS S a is ics o
Macin osh, V. 21.0. (IBM Co po a ion, New Yo k, Uni ed S a es) and eViews .
Fi s , desc ip i e s a is ics o he endogenous and all exogenous a iables
we e done. Then a linea eg ession model, a logi and a binomial logi model
we e calcula ed o s udy he co ela ion be ween he a iable and i s
p edic o s.
7.2.4.- Resul s
Fou pe imodiola elec ode a ay ca ie s (pa ien s 2-4 and 9) and 4 s aigh
a ay ca ie s (pa ien s 5-8) comple ed he es . Pa ien 1 did no a end he
CT scan appoin men and pa ien 10 had an inapp op ia e MRI echnique o
an un ela ed heal h issue and his a ay was displaced so he had o unde go a
epai ing su ge y and so could no comple e he es .
Consequen ly, a o al o 1968 ials we e used o he analysis. O all hese,
1169 we e co ec answe s and 799 we e no . T ials we e g ouped in o uni s
o obse a ions. A uni is he esul o i s a e aging he success o a gi en
e e ence elec ode wi h each o i s nea es and second nea es neighbou s
(4 o elec odes 3 o 20; 3 o elec odes 2 and 21 and 2 o elec odes 1 and
Expe imen al Wo k
!
!!!
93!
22) and hen u he a e aging hese in o a single alue. The success a e is
hus he quo ien be ween he numbe o successes by he numbe o ials,
o a gi en e e ence elec ode (6 ials o elec odes 1 and 22, 9 ials o
elec odes 2 and 21 and 12 ials o he es o elec odes). I s desc ip i e
s a is ics (Table 7.2), he Ja que-Be a no mali y es , oge he wi h he
his og am o equencies a e shown in igu e 7.9. F om he la e i is
concluded ha no mali y canno be ejec ed a 4.60% signi icance (p=0.046)
and hence he a iable beha es has a nea Gaussian dis ibu ion, wi h a
success a e o 59.70%.
Figu e 7.9. Success a e his og am ep esen a ion
Desc ip i e S a is ics
Mean
0.597
Median
0.583
Maximum
1
Minimum
0
S anda d De ia ion
0.247
Asymme y
-0.373
Ku osis
2.470
Ja que-Be a Tes
6.141
Ja que-Be a associa ed p obabili y
0.046
Table 7.2. Desc ip i e s a is ics.
Expe imen al Wo k
!
!!!
94!
To iden i y which ac o s ha e he capaci y o in luence and modi y he
success a e, we s a ed by con as ing whe he he elec ode numbe could
be one o hem. A Le ene’s es o equali y o a iances o he mean success
a e indica es ha he a iance is he same o all elec odes (p=0.35, he null
hypo hesis which s a es ha he a iances a e di e en canno be ejec ed).
Consequen ly, an ANOVA con as was also done, concluding ha he
equali y o he means o he successes o all elec odes canno be ejec ed
(The F s a is ic alue had an associa ed p= 0.82).
A simila analysis o he a iance was conduc ed o he pa ien ac o . The
esul sugges s he exis ence o s a is ically di e en esul s depending on he
pa ien , bo h o mean and a iance alues (p obabili y associa ed wi h
he Le ene s a is ic o equali y o a iances p=0.0001, and p obabili y
associa ed wi h he F s a is ic o Welch o equali y o a iances p=0.000).
Acco dingly, he empi ical e idence suppo s he hypo hesis ha he e is a
pa ien e ec ha signi ican ly a ec s success a e. Figu e 7.10 shows he
s anda d de ia ion plo o he mean success a e o each pa ien .
!
Expe imen al Wo k
!
!!!
95!
Figu e 7.10. Mean success a e alues and s anda d de ia ions pe pa ien (pa ).
As can be seen om he cha , he e a e ou pa ien s ha s and ou om he
es . Pa ien s 6 and, mos p ominen ly, 8 ha e a la ge dispe sion. In addi ion,
pa ien 6 also has a low mean success a e, al hough no as low as pa ien 7,
which is he hi d pa ien ha s ands ou om he sample. On he opposi e end
is pa ien 4, wi h he highes success a e and he smalles dispe sion
be ween elec odes.
O hese pa ien s ha ha e a di e en ial beha iou , h ee o hem, pa ien s 6,
7 and 8, ha e s aigh elec ode a ays (al hough so does pa ien 5). In ac ,
he da a shows a s a is ically signi ican di e ence in success a es be ween
pa ien s implan ed wi h s aigh and pe imodiola a ays. The p obabili y
associa ed wi h he Welch’s F s a is ic (because a iances be ween g oups
we e di e en ) o he equali y o he p obabili y o success in he p esence o
di e en a iances also ga e a alue o 0.000, being he p obabili y o
success 53% o s aigh elec ode a ay ca ie s and 70% o pe imodiola
elec ode a ay ca ie s. This esul p o ides empi ical e idence in a ou o
he hypo hesis ha pe imodiola elec ode a ays p o ide be e elec ode
disc imina ion.
The mean e ec o dis ance on success a e in a e age e ms is e iden when
hese wo a iables a e plo agains each o he ( igu e 7.11). Pa ien s whose
elec odes a e u he away om he inne wall a e hose wi h a lowe success
a e, on a e age.

Expe imen al Wo k
!
!!!
96!
Figu e 7.11. Rela ionship be ween he mean elec ode-inne wall dis ance pe pa ien and
mean success a e.
In any case, he abo e esul s mus be aken as indica i e,-no conclusi e-, o
se e al easons. On he one hand, one canno di e en ia e he pa ien e ec
om he dis ance e ec on success a e. On he o he hand, he esul s a e
based on his a iable ha ing no mal beha iou . Howe e , success is a
binomial a iable. The aw ou pu da a consis ed on a ma ix o ones (success)
and ce os ( ailu e). In ac , e en hough he Ja que-Be a con as does no
ejec he null hypo hesis o no mali y o he success a e, o he con as s
ejec such beha iou o signi icance le els o less han 1% (Kolmogo o -
Smi no , Shapi o-Wilk).
A co ela ion and eg ession analysis was done o o e come hese limi a ions
and o sea ch o g ea e p ecision in he measu emen o he ela ionship
be ween he dis ance o he elec ode o he inne wall o he cochlea and he
Expe imen al Wo k
!
!!!
97!
success a e. To his end, he ollowing addi ional a iables we e also
inco po a ed:
1. NRT: quali y o he ne e measu ed in CL. A highe alue indica es a
wo se s a e o he ne e, as mo e cu en needs o be deli e ed o
he ne e end o elici a measu able esponse. Thus, i is expec ed
ha he inc ease in he alue o his a iable esul s in a dec ease in
he p obabili y o co ec ly disc imina ing be ween elec odes.
2. T le el: minimum cu en needed o elici hea ing sensa ion,
measu ed in CL. Acco ding o he li e a u e, his a iable is expec ed
o ha e a posi i e ela ionship wi h success a e.
3. Impedance: measu ed in kΩ. Because la ge impedances esul in
highe cu en s equi ed o each he neu al ends, his a iable is
expec ed o ha e a nega i e e ec on he success a e.
To do his analysis, as indica ed ea lie , he sampling uni s a e all he di e en
elec ode-pa ien combina ions, i.e. elec ode 1 o pa ien 1, elec ode 1 o
pa ien 2, e c. The desc ip i e s a is ics o he a iables a e shown in able
7.3. On a e age pa ien s ha e been success ul in 60% o he es s, ha e an
elec ode dis ance o 0.87 mm, a mean NRT o 183.61 CL, a T le el o 126.29
CL and an impedance o 9.66 kΩ. The ela i e dispe sion is mode a e in all
a iables, being he a iable dis ance he one wi h g ea e ela i e dispe sion.
Because he a iable de e io a ion o he ne e (NRT) was no measu ed in 12
elec odes, he sample size had o be educed o 158 elemen s.
Table 7.4 con ains he co ela ion ma ix o he a iables and hei le el o
signi icance. I can be concluded ha he a iable o in e es , he success a e,
is signi ican ly co ela ed wi h all explica i e a iables, and wi h he expec ed
signs as well. The e a e also signi ican co ela ions be ween he di e en
explana o y a iables, which make i di icul o quan i y hei indi idual e ec
on he mean success a e. Howe e , hei alues a e no high enough o
p oduce se ious p oblems o mul icollinea i y.
Expe imen al Wo k
!
!!!
98!
Success
Ra e
Elec ode-inne
wall dis ance
NRT
T le el
Impedance
Mean
0.60
0.87
183.61
126,29
9.66
Maximum
1.00
2.09
237.00
176.00
20.33
Minimum
0.00
0.11
109.00
95,00
4.23
S anda d
De ia ion
0.25
0.50
22.06
18.56
3.24
Rela i e
dispe sion
0.41
0.58
0.12
0.15
0.34
No. o
Obse a ions
176
176
158
176
176
Table 7.3. Desc ip i e s a is ics o he a iables: elec ode-inne wall dis ance, NRT, T le el
and impedance.
Success
Ra e
Elec ode-inne
wall dis ance
NRT
T le el
Va iance
In la ion
Fac o
Elec ode-inne
wall dis ance
-0.207***
1.23
NRT
-0.294***
0.421***
1.87
T le el
0.169**
0.260***
0.521***
1.64
Impedance
-0.318***
-0.015
0.217***
-0.222***
1.27
* Signi ican a 10%, ** signi ican a 5%, *** signi ican a 1% o bila e al con as s
Table 7.4. Pea son co ela ion o he a iables: success a e, elec ode-inne wall dis ance,
NRT, T le el and impedance.
The de e mining ac o when modelling he endogenous a iable is he e y
na u e o he a iable i sel . Success a e is bounded be ween 0 and 1 and i s
heo e ical beha iou co esponds o a Be nouilli dis ibu ion. Le us emembe
ha he success a e has been ob ained as he quo ien be ween he numbe
o igh answe s di ided by he numbe o ials pe elec ode pai s o a gi en
e e ence elec ode (6, 9 o 12). When calcula ed in his way he a iable has
a binomial beha iou . The e o e, he assump ion o linea ela ionship wi h he
explana o y a iables, which implies a cons an linea e ec , is no s ic ly alid
because linea models gi e alues below ze o and abo e one. Mo eo e , he
a iance o he success a e is no cons an and so he e is he e ocedas ici y.
These p ope ies limi he use o he s anda d me hod o mul iple linea
Expe imen al Wo k
!
!!!
99!
eg ession and i s es ima ion by o dina y leas squa es and hus we mus
eso o al e na i e me hods be o e gi ing inal conclusions.
Th ee p edic i e me hods we e used, all o hem using he explana o y
a iables desc ibed abo e. Fi s , success a e was modelled using he me hod
o minimum chi-squa es. To adap he model o he na u e o he a iable o
in e es , wo mo e models a e p oposed, bo h aking he logi ans o ma ion o
he success a e as he endogenous a iable.
Linea p obabili y model es ima ed by he me hod o leas chi-
squa es (MCSM)
The i s model desc ibes he p obabili y o success as a linea unc ion o
he p edic i e a iables and he weigh ed leas squa e es ima ion o he
ollowing o m (140):
!!=!!
!!
=!!+!!!!!+!!!!!+⋯+!!!!" +!! [7.1]
This is he simples me hod and he easies o in e p e , bu is no wi hou
se ious limi a ions. pi is he success a e (i is nei he he numbe o
successes, no he numbe o es s); {x1,…,xk} a e he se o p edic ing
a iables; ui is he andom pe u ba ion; a e he pa ame e s o he model
ha quan i y he e ec o each one o he p edic i e a iables on he success
a e and i ela es o each o he sampling uni s (pa ien -elec ode). Gi en he
na u e o he endogenous a iable, he model has a p oblem o
he e oscedas ici y. The e o e, he da a was weigh ed using a new a iable,
Pond1, de ined as:
!"#$1=!!
!!(!!!!) [7.2]
β
i
Expe imen al Wo k
!
!!!
106!
8.- ELECTROMAGNETISM
8.1.- The e ec o e e ence elec ode posi ion on powe consump ion in
Cochlea Implan s
(Published as: Ramos-Miguel, A, Ramos-Macías, A, A iles, JV, Zaballos,
MTP (2015). The E ec o Re e ence Elec ode Posi ion in Cochlea
Implan s. J In Ad O ol, 11(3), 222-8.)
Despi e he use o di e en elec ode designs and coding s a egies ac oss
Cochlea Implan (CI) companies, he a iabili y in audi o y abili ies o CI
ecipien s seems o be mo e simila ac oss de ices han o he same de ice
ac oss indi iduals. This sugges s ha signi ican ecipien -dependen ac o s
limi o e all audi o y abili y a an indi idual le el. Aspec s such as neu al
su i al a e known o ha e a signi ican e ec . Un o una ely, li le can be
done o imp o e pa ien pe o mance in such cases. Howe e , he e migh be
o he easons ela ed o he physical cha ac e is ics o he de ice, a he han
he pa ien ’s e iology. (144)
In his s udy, he e ec o Re e ence Elec ode (RE) posi ion on powe
consump ion and s imula ion in ensi y was in es iga ed using wo di e en
app oaches: 1) 2D nume ical simula ions and 2) eal empo al bone
measu emen s using a es implan de eloped by he Labo a o y o
Psychoacous ics o he O oneu ology Di ision, ENT Depa men o he
Complejo Hospi ala io Uni e si a io Insula Ma e no In an il (CHUIMI).
8.1.1- In oduc ion
The RE posi ion in CI has been su p isingly unde s udied. The e a e no
a icles in es iga ing a co ela ion be ween ba e y li e and posi ion o he RE.
Mos cen e s do no e en use his elec ode, wi h jus he body o he implan

Expe imen al Wo k
!
!!!
107!
being used. F om a physical poin o iew, RE posi ion has a bigge in luence
ega ding powe consump ion han in acochlea elec ode posi ion.
Despi e using di e en elec ode designs and coding s a egies ac oss CI
companies, he a iabili y in audi o y abili ies o ecipien s appea s o be mo e
simila ac oss de ices han o he same de ice ac oss indi iduals. This
sugges s ha a an indi idual le el, signi ican ecipien -dependen ac o s
limi he o e all audi o y abili y. Aspec s, such as ossi ica ions, a e known o
ha e a signi ican e ec . Un o una ely, o imp o e pa ien pe o mance in
such cases spli elec odes ha e o be used, and hey equi e exquisi e
op imiza ion o pa ame e s o ob ain good esul s!(144). Goeh ing e al.
disco e ed ha 12.4% o implan a ions in aope a i ely demons a e high
impedances. In 8.2% o cases, impedances pos ope a i ely emain high.
(145)
The e is ex ensi e li e a u e whe ein many a iables conce ning in acochlea
elec odes a e e alua ed. In con as , e y ew s udies ha e explo ed he
e ec s o posi ion, numbe , o shape o RE, wi hou which CI would no
p ope ly unc ion (146). A ield ha has been explo ed in mo e dep h is he
compa ison be ween mono- and bipola s imula ion on he s ill con o e sial
hypo hesis ha mo e es ic ed cu en ields can p o ide be e speech
pe cep ion. Howe e , he e a e p oblems wi h bipola s imula ion: 1) Bipola
elec odes equi e highe cu en s o sup a- h eshold s imula ion because o
a cu en shun om elec ode o elec ode and 2) high po en ials and high
cu en densi ies o m a ound each o he wo elec odes and sup a- h eshold
exci a ion is possible nea bo h he sou ce and sink elec odes (147) (148)
(149) (150) (151) (152). The e o e, powe consump ion is s ill a majo
d awback in his s imula ion mode, he eby making monopola he p e e ed
choice o mos audiologis s. The e o e, i is impo an o e alua e whe he RE
posi ion in monopola s imula ion can dec ease powe consump ion and
imp o e elec ical ield sha pness.
Expe imen al Wo k
!
!!!
108!
In monopola s imula ion, he cu en is injec ed h ough a con ac in he scala
ympani and e u ned o a a - ield elec ode con ac . B iai e and F ijns s udied
he cu en dis ibu ion in mul iple elec ode con ac s along he scala ympani
using his mode (106). The esul demons a ed ha elec ical po en ial ield
pa e ns caused by monopola s imula ion a e b oad in na u e and ha
po en ial ela i ely and slowly d op o as one mo es away om he
s imula ing con ac . High impedances in he issues can cause losses in he
elec ical impulse quali y, he eby equi ing high powe o he pa ien o
e icien ly de ec his signal. I can e en lead o he deac i a ion o some
elec odes because hei powe consump ion is so high ha hey hinde he
global s imula ion s a egy. When impedance is oo high, pa icula ly a high
cu en le els, hen i is possible ha he implan ol age will no be su icien
o gene a e he p og ammed elec ic cu en le el. In his case, an elec ode is
said o be “ou o compliance.” This leads o he si ua ion whe e pe cep ual
loudness does no inc ease wi h inc eases in he cu en le el. To inc ease
loudness, he wid h o s imula ion pulses can be inc eased. In some cases,
impedance is oo high ha his app oach is no easible because o oo la ge
pulse du a ion equi ed. Hence, high impedance esul s in a dec ease in
ba e y li e bu also in a wide sp ead o exci a ion because o he la ge
cu en in ensi y equi ed.
Howe e , dec easing impedance is easible and can sol e he issue o ba e y
consump ion. How his migh be done depends on he sou ce o impedance. I
could be achie ed by educing he space be ween he RE and signal
elec ode. The RE can be he ex acochlea elec ode o some ypes o
implan s, which can be placed in di e en pa s o he mas oid, bu also he
body o he implan , used as g ound elec ode as well. By educing he
amoun o issue be ween he g ound and in acochlea elec odes,
impedance could be lowe ed, and he pulse in ensi y equi ed could be
ob ained wi h less powe . (153) (154)
In CI su ge y, no much a en ion is gi en o he speci ic posi ion o he RE.
The su gical echnique his o ically sugges s posi ioning he RE as a as
Expe imen al Wo k
!
!!!
109!
possible om he elec onic pa o he implan o a oid in e e ences du ing
s imula ion. This pa is cu en ly isola ed om his ype o in e e ence by
Fa aday cages. The e o e, RE can be cu en ly posi ioned close o he
s imula ion egions. (155) (156)
This s udy aims o examine he a iabili y in ex acochlea elec ode posi ion
as a con ibu ing ac o o powe consump ion in CI. This simple change can
be o g ea ad an age ega ding he ba e y li e o pa ien s wi h cochlea
ossi ica ions. Because hey equi e high cu en s o pe cei e elec ical
s imula ion, a close dis ance be ween he in a- and ex acochlea elec odes
may enable he gene a ion o la ge cu en in ensi ies wi hou he need o
up u n he pulse du a ion o inc ease audi o y sensa ion.
8.1.2- Objec i e, Hypo heses and a ionale o he s udy
The aim o his s udy is o examine a iabili y in ex acochlea elec ode
posi ion as a con ibu ing ac o o a iabili y in audi o y abili y in CI pa ien s, in
he con ex o wo ques ions, 1) whe he RE placemen has a measu able
e ec on consump ion, and 2) i RE posi ion has a unc ional e ec on
s imula ion in ensi y.
The hypo hesis is ha he close he RE is placed om he s imula ion
egion, he lesse powe consump ion will be equi ed o each a poin a
a se in ensi y.
The eason behind his s udy has been o y and inc ease he ba e y li e o
he cochlea implan and o imp o e implan unc ioning in pa ien s wi h
ossi ica ions in he cochlea. Because hey equi e high cu en s o pe cei e
elec ical s imula ion, a close dis ance be ween in a and ex acochlea
elec odes may allow he gene a ion o mo e in ense cu en s wi hou he
need o inc ease he pulse bandwid h o augmen he audi o y sensa ion.
Expe imen al Wo k
!
!!!
110!
8.1.3- Ma e ial and me hods
E hics commi ee app o al was no equi ed o his s udy because i did no
in ol e human subjec s.
Classical Ci cui Analysis
To cha ac e ize he heo e ical physical sys em a much-simpli ied elec ical
model was used.
Th ee elec ical o mulas we e used. The i s is Ohm’s Law, which was
pos ula ed by Geo Simon Ohm in 1826. I s a es ha a cu en I h ough a
conduc o be ween wo poin s is di ec ly p opo ional o he po en ial
di e ence V ac oss he wo poin s, wi h esis ance R being he cons an o
p opo ionali y be ween hem (157):
!=!∙!![8.1]!
In his case, V is he po en ial di e ence p oduced by CI be ween he signal
and he RE, R is he impedance caused by he issue be ween he elec odes,
and I is he elec ical cu en passing h ough i . Sol ing o I, i becomes clea
ha o aise his alue, V needs o be inc eased, o R has o be lowe ed.
Thus, he only way o do his wi hou supplying ex a ene gy is by educing he
esis ance be ween he RE and he in acochlea elec ode.
The second equa ion is he elec ical powe o mula:
!=!∙!=!∙!! [8.2]
I ep esen s he a e o ene gy ha is con e ed om he elec ical ene gy o
he mo ing cha ges o some o he o m, e.g., hea , mechanical ene gy, o
ene gy s o ed in elec ic o magne ic ields. Powe is measu ed in Wa s (W)
and is used as a measu e o he e iciency o a de ice. I is di ec ly
Expe imen al Wo k
!
!!!
111!
p opo ional o he squa e o he elec ical cu en and esis ance o a ci cui .
(157)
Thus, o a ixed I, he highe he R, he mo e he powe equi ed o be
supplied by he ci cui o keep I cons an . Elec ical hea ing h eshold occu s
when he cu en I deli e ed o he ne e ibe s eaches a minimum alue.
This alue can be aken as he ixed alue o I. Then, i becomes clea ha he
u he away he RE is om he ac i e elec ode, i.e., he highe he esis ance
be ween hem, he mo e powe will be equi ed o each ha h eshold.
The las o mula used is he ma hema ical de ini ion o he elec ical esis i i y
o a ma e ial, which is he esis ance opposed by a ma e ial o an elec ical
cu en low ac oss i . This o mula gi es he cha ac e is ic R o each
biological issue be ween he RE and an in a-cochlea elec ode.
!=!!
! [8.3]
Whe e ! is he leng h o he conduc o , S is i s c oss-sec ion a ea, and ρ is he
elec ical esis i i y coe icien o he ma e ial, which is measu ed in Ohms pe
me e (Ω·m). In his s udy, he alues o ρ o endolymph (ρ=0.75 Ω·m), bone
(ρ=20 Ω·m), and muscle (ρ=3.5 Ω·m) we e ob ained om he da abase ha
was c ea ed by IT’IS Founda ion. (IT’IS Da abase Low-F equency
Conduc i i y)
2-D Nume ical Simula ions
The i s echnique used was nume ical simula ion. The ad an age o using
his app oach is ha he physical phenomenon can be ec ea ed in a
comple ely con olled en i onmen whe e all he a iables and geome ies can
be de ined, measu ed, and modi ied o ma ch he eal-li e sys em.
The simula ion so wa e used was Ma hema ica (Wol am Resea ch, Inc.,
Ma hema ica, Ve sion 10.0, Champaign, IL (2014)). A 2D ini e elemen

Expe imen al Wo k
!
!!!
112!
simula ion o elec omagne ic phenomena o low equency cu en s was
build.
A use ul app oach o he calcula ion o elec ic po en ials (Vol age) is o ela e
ha po en ial o he cha ge densi y. Since elec ic cha ge is he sou ce o
elec ic ield, he elec ic ield a any poin in space can be ma hema ically
ela ed o he cha ges p esen . Fo mul iple poin cha ges, a ec o sum o
each o hei elec ical ields is equi ed. Fo a con inuous dis ibu ion o
cha ge, ec o sums a e be e handled h ough calculus. This is done h ough
he di e gence ela ionship. This s a es he di e gence ∇∙ ( he scala
de i a i e) o he elec ic ield ec o E is p opo ional o he cha ge densi y !
(158):
∇∙!=−!!
!!
[8.4]
Whe e !!is he pe mi i i y and !! is he cha ge densi y.
Because he elec ic ield is also he de i a i e o he po en ial, we ob ain
Poisson’s equa ion:
!=∇∙∇V=∇!!=−!
!!
[8.5]
In a cha ge- ee egion o space o when he equency is low, he go e ning
equa ion o he elec omagne ic ield is he s a ic equa ion, which is Laplace’s
equa ion. This is a di e en ial equa ion wi h a smoo h solu ion, wi hou
singula i ies. I is used o de ine he po en ial ene gy ield caused by a gi en
cha ge dis ibu ion:
∇!!=0 [8.6]
The model’s ask will hen be o de e mine he solu ion o his equa ion o a
biological model, which in ol es a ious ma e ials and geome ies. This
Expe imen al Wo k
!
!!!
113!
equa ion gi es he second de i a i e o V; howe e , because i is V i sel ha
is needed, in eg a ion is equi ed. Howe e , while in eg a ing, cons an s
appea ha canno be nume ically de e mined unless he bounda y o ini ial
condi ions o he sys em a e known. These a e he alues a he bo de s o
he sys em’s space o he ini ial alue o V be o e i s a s mo ing h ough
space and ime. (158)
Two ypes o condi ions we e used: Di ichle and null Neumann. The i s
speci ies he condi ions ha he elec ical po en ial equi es o ake along he
bounda y. The second se s he alue o i s de i a i e along his bounda y.
Di ichle condi ions o 0 and 5 V we e imposed on he RE and in acochlea
elec ode, espec i ely. The null Neumann condi ion was applied o he edges
ha limi he geome y o he sys em. (159)
Ano he impo an s ep in designing he compu e model is he geome y,
which is illus a ed in Figu e 8.1. The model ep esen s h ee elec ode
posi ions. They all sha e he same elemen s; howe e , he dis ance be ween
he elec odes dec eases by a ac o o 2 be ween one case and he nex .
Figu e 8.1 Geome ies used in he nume ical model. The in e -elec ode bone laye hickness
was modi ied o s udy i s e ec on impedance, and hus, elec ical consump ion.
Each elemen was assigned he elec ical esis i i y alues desc ibed in he
ci cui analysis sec ion. Ai was included in his model, wi h !=1.3×1012 Ω·m.
(160)
Expe imen al Wo k
!
!!!
114!
Tempo al Bone Measu emen s
A de ice was designed o s imula e and moni o elec ical impulses ha we e
deli e ed o a empo al bone while measu ing impedance, cu en , and
ol age.
The pu pose-buil implan comp ises one in acochlea and six ex acochlea
elec odes. Because only he e ec s o ela i e posi ion be ween he RE and
in acochlea elec ode a e o in e es in his s udy, CI was simpli ied and
comp ised a single in acochlea elec ode. In con as , o assess
ex acochlea elec ode cu en s along di e en egions a he same ime, six
o hem whe e loca ed along he empo al bone. They all ha e sphe ical ends
wi h a mean diame e o 0.9 mm. All ex acochlea elec odes can be se o be
RE, one a a ime, hus p o iding six di e en measu emen posi ions. The
emaining elec odes ac as p obes h oughou a gi en s imula ion ial.
The sys em comp ises a mic ocon olle (A duino UNO, Some ille,
Massachuse s, USA) ha is connec ed o a compu e ia USB. I was
p og ammed o ansmi elec ical signals, sample each o he p obes, and
s o e da a. I also calcula es impedances, cu en s, and powe ha a e
gene a ed du ing s imula ion. The boa d is connec ed o a b eadboa d wi h
he elec odes and p obes o be inse ed in o he empo al bone.
Expe imen al Wo k
!
!!!
115!
Figu e 8.2 Schema ics o he implan ci cui . The e e ence/p obe elec odes we e a ached o
he analog po s o he mic ocon olle boa d. Swi che s S1–S6 de ine which p obe o selec
as RE. S12 disconnec s o econnec s he in acochlea elec ode. Resis o s R1–R7 all ha e
a alue o 1.5 kΩ.
The ci cui is based on a numbe o ol age di ide s. The schema ics a e
depic ed in Figu e 8.2. The wo esis o s in se ies o m a ol age di ide
ci cui . One end o he esis o pai is ed 5 V, while he o he end is
connec ed o he g ound. Fi e ol s ha he A duino p o ides a e di ided
be ween he wo esis o s depending on hei esis ance. The esis o , which
holds he g ea e esis ance, ge s mo e o he ol age, acco ding o Ohm’s law
o mula. The ol age ha alls ac oss a componen is di ec ly p opo ional o
he amoun o esis ance i con ains (157):
!!"#$ =!!"#"!"$%"!!!"
!
!"#
−!!"#"!"$%" [8.7]
Using his p inciple, a model can be se up o de e mine he esis ance on he
basis o he ol age di ision. Wi h his da a, all elec ical cha ac e is ics o he
ci cui can be de ined, i.e., impedances, cu en s, and ol ages.
Discussion
!
!
!
122!

Discussion
!
!
!
123!
III. DISCUSSION
Discussion
!
!
!
124!
Discussion
!
!
!
125!
9.- PSYCHOPHYSICS
9.1.- How he heal hy ea unde s ands speech in noise and he e ec s o
aining on i
The abili y o ocus and unde s and a speake in a noisy en i onmen is a
c i ical social-cogni i e capaci y whose unde lying mechanisms a e s ill
unclea . Many explana ions ha e been p oposed. Howe e , none had looked
a high equencies as he possible cues o speech iden i ica ion in noise.
Thus, in he i s psychophysical s udy o his hesis, he ela ion be ween
speech ecogni ion in noise and equency bandwid h o he audi o y s imulus
was in es iga ed.
Many s udies ha e al eady demons a ed he impo ance o equency
bandwid h in he pe cep ion o speech in quie . Fo example, S elmachowicz
e al. ha e s udied hese e ec s on he pe cep ion o he phoneme /s/ in
No mal Hea ing (NH) child en and adul s. The esul s showed a co ela ion
be ween hese wo ac o s o all g oups. A clea example whe e a limi ed
hea ing bandwid h can be ound is in p esbycusis, which in ol es he ise o
audi o y h esholds o high equencies (161). In hese pa ien s, he mos
common complain is he inabili y o unde s and Speech In Noise (SIN), while
hey can s ill ollow a con e sa ion in a quie en i onmen wi h li le di icul y.
These wo indings ( ha high equencies a e impo an o speech pe cep ion
in quie , and ha he ise in high equency hea ing h esholds hinde SIN
abili y) seem o poin in he di ec ion o high equencies as impo an cues o
speech pe cep ion, speci ically in he case o backg ound noise.
A s udy by S ua e al. analysed wo d ecogni ion pe o mance in 12 NH
adul s in con inuous and in e up ed b oadband noise as a unc ion o SNR
wi h and wi hou il e ing a 2 kHz. Howe e , hei il e ing h eshold was lowe
han he ac ual 8 kHz ha a e age implan s ha e. Mo eo e , hey ocused
p ima ily on he di e ence be ween con inuous and in e up ed noise, a he
han he ole o high equencies. In a pe iod whe e Cochlea Implan s (CI)
only eached a ound 4 kHz, Ho nsby also demons a ed ha be e
Discussion
!
!
!
126!
pe o mance was accomplished when he equency bandwid h was inc eased
o 7 kHz. (162) (163)
The s a is ical ou come in his hesis is in acco dance wi h he indings om
S ua and Ho nsby, o NH subjec s, when noise is ei he highe o lowe han
he speech signal. The esul s sugges ha high equency componen s abo e
8kHz may play a signi ican ole in wo d ecogni ion in noisy en i onmen s.
Howe e , o he si ua ion whe e noise and speech ha e equal le els o
in ensi y his was no he case. The cha ac e is ics o he expe imen do no
allow o a clea explana ion as o why his happens. Acco ding o he esul s
ob ained, he e seems o be a di e ence in how humans ecognize speech in
di e en noise si ua ions. This is in acco dance wi h a neu ophysiological
s udy by Wong e al. whe e hey measu ed co ical ac i i y o NH subjec s
while hey iden i ied wo ds in wo noise condi ions: below and abo e he
speech le el. They ound co ical noise-dependen ac i a ion o he bila e al
middle and le pos e io po ions o he supe io empo al gy us. These
esul s likely e lec demands in acous ic analysis, audi o y-mo o in eg a ion
and phonological memo y, as well as audi o y a en ion. Un o una ely, hey
did no s udy wha happened when noise and speech we e a he same le el,
so he ac i a ion si es in his case a e s ill a unknown. Thus u u e esea ch
poin s in he di ec ion o combined psychoacous ics and neu ophysiological
expe imen s, whe e neu al ac i i y is eco ded while speech ma e ial is
p esen ed using he h ee SNR o he p esen expe imen . (69)
Co ela ion s udies be ween how much a wo d was al e ed by il e ing and i s
sco e did no yield conclusi e esul s. Signi ican co ela ion was only ound
o he case whe e noise was lowe han speech. This was somewha
expec ed, since wo d ecogni ion is a complex phenomenon. Vowels and
consonan s a e pe cei ed h ough di e en mechanisms. Nie e al. sus ain
ha consonan ecogni ion happens h ough empo al cues while owel
ecogni ion elies on spec al cues. The esul s o he p esen expe imen , in
combina ion wi h he s a is ical ou come obse ed, seem o indica e ha e en
Discussion
!
!
!
127!
small addi ions o high equency con en o sounds can lead o an o e all
imp o emen in speech ecogni ion in noise. In he ligh o hese a gumen s, i
is hypo hesized ha owel ecogni ion assumes an impo an ole in speech
ecogni ion in noise. Mo eo e , esul s om F iesen e al. indica e ha p ope
owel ecogni ion equi es app op ia e equency alloca ion. The implica ions
o hei indings and hose ob ained in his hesis lead o he hough ha o
app op ia e speech ecogni ion in noise, i will be necessa y o ul il wo
equi emen s: 1) a co ec equency alloca ion and 2) he p ese a ion o high
equencies. This is because i he audible equency is inc eased, bu he
alloca ion o such equencies is no accu a e, hen he esul will no be
econs uc ed p ope ly by he b ain. (164) (165)
The i s s udy (sec ion 6.1) hus showed how deg aded high equency
h esholds can lead o dec eased abili y o unde s and SIN. He ea e , he
na u al ques ion o ask is: is he e a way o p e en , o e en e e se his
p oblem? I he answe o his ques ion we e posi i e, his would mean ha
pa ien s wi h hea ing loss could signi ican ly imp o e hei SIN pe cep ion
h ough an app op ia e aining p o ocol.
The sea ch o he answe leaded o he second s udy o his hesis (sec ion
6.2). The abili y o ai a ic con olle s (ATC) o unde s and SIN unde limi ed
equency bandwid h condi ions is ex ao dina y. Howe e , his does no seem
o be an inna e abili y. These indi iduals all epo ha ing p og essi ely
acqui ed his acul y h oughou he yea s, bu specially du ing hei wo-yea
aining pe iod. Thus, he second expe imen o he se ies o psychophysical
expe imen s in his hesis ied o objec i y he e ec o aining on SIN
unde s anding.
Many s udies ha e in es iga ed he e ec o SIN exe cises. Song e al.
s udied aining- ela ed malleabili y using a p og am ha inco po a ed
cogni i ely based lis ening exe cises o y o imp o e SIN pe cep ion in
pa ien s wi h hea ing loss. T ained subjec s exhibi ed signi ican
imp o emen s in SIN pe cep ion ha we e e ained 6 mon hs la e .

Discussion
!
!
!
128!
Subco ical esponses in noise demons a ed aining- ela ed enhancemen s
in he encoding o pi ch- ela ed cues. This was he i s ime ha sho - e m
aining was demons a ed o ha e he po en ial o imp o e he neu al
mechanisms o SIN pe cep ion. These esul s in ol e and de ine biological
mechanisms ha con ibu e o lea ning success, and hey p o ide a
concep ual ad ance o he unde s anding o he kind o aining ha can
in luence senso y p ocessing in adul hood. (99)
Mo e ecen ly, Swee how and Sabes p ospec i ely assessed he
gene aliza ion o SIN aining in a coho o indi iduals wi h hea ing loss
(166).. Whi on e al. p ospec i ely assessed he e ec o signal-in-noise,
audio-game aining on SIN unde s anding wi h un ained ma e ials (167). On
he o he hand, Fu and Gal in and Moo e and Shannon ha e shown ha
a ge ed audi o y aining can u he enhance he bene i s o new implan
de ices and/o speech p ocessing s a egies (168) (169). These indings and
hose om his hesis sugges ha SIN aining could be bene icial bo h o CI
use s and o hose wi h mode a e hea ing loss, since i equips pa ien s wi h
he cogni i e ools o con on complex audi o y si ua ions.
Howe e , he only g oups ha ha e been s udied and ha ha e some kind o
long- e m aining a e musicians, who a e audi o y expe s. Fo example
Pa be y-Cla k e al. in es iga ed he e ec o musical aining on SIN
pe o mance. They ound ha musical expe ience imp o es he abili y o
unde s and speech in challenging lis ening en i onmen s. The esul s also
sugges ha his enhancemen is de i ed in pa om musicians’ ema kable
wo king memo y and equency disc imina ion.. (170)
The esul s desc ibed in sec ion 6.2 indica e ha ATC, jus like musicians, a e
mo e capable o iden i ying SIN. This abili y is mos appa en in he mos
ad e se case, when he signal is lowe han he noise. In all cases, hey
ob ained be e sco es han he Con ol g oup. The p oposed explana ion is
ha his is due o he e ec o aining ha occu s na u ally du ing hei daily
ac i i ies. E en hough he se o wo ds used we e comple ely unknown o
Discussion
!
!
!
129!
hem, i seems ha he skills lea n in one pa icula audi o y si ua ion a e
ans e able o a new and di e en one, in his case, he labo a o y es . This
is an in e es ing esul , since he i s hing ha subjec s commen ed a e
being in o med o he pu pose o he s udy was ha hey ‘would p obably no
pe o m be e because hei “ ick” is ha hey know which wo ds o expec
du ing adio communica ions’. Howe e , i is clea om he esul s ha his is
no he case and hey could iden i y mo e wo ds han he s anda d NH
popula ion. Reg e ully, he p e alence o his opinion was no quan i ied in
his s udy. I would ha e been e y in e es ing o see how much hei belie s
co ela e wi h he es esul s. The s udies e iewed he e a ou he
hypo hesis ha hei subjec s, and likely ATC as well, a e be e a ocusing on
he less deg aded acous ic speech cues and hen illing he gaps using
cogni i e skills in he p esence o backg ound noise. This way hey equi e
less ime o adap a ion and he e o e miss less in o ma ion.
Gi en he p omising esul s, i was decided o in es iga e possible co ela ions
be ween yea s wo king and age. Howe e , his analysis did no e u n clea
esul s. The wo cases whe e signi ican co ela ions we e ound (in e se
co ela ions be ween 8 kHz 5 SNR es esul s and 22 kHz 5 SNR) do no
seem meaning ul in he con ex o he o he es s. I a co ela ion o any kind is
o be es ablished be ween hese wo a iables, hen consis ency ac oss es s
is equi ed. Thus, he esul s obse ed a e likely o be caused by he limi ed
numbe o subjec s, which is an un o una e bu common aspec in clinical
s udies, as opposed o pu e physics s udies, due o he di icul y o ec ui ing
olun ee s. In such cases, s a is ics needs o be ea ed wi h ca e. I is
possible and e y likely, gi en he ime aken by subjec s in he s udies
men ioned be o e (app oxima ely 6 mon hs- 1 yea ) (97), ha he lea ning
cu e is s eepe du ing hei i s yea o aining and app oaches a pla eau
he ea e . Thus, ano he s udy eme ges as a e y in e es ing con inua ion o
he p esen one: o es ATC in aining a di e en s ages o hei lea ning
pe iod, o o ollow hem longi udinally. Maybe looking mo e closely in o his
pe iod o hei wo king li e will e eal a signi ican co ela ion be ween SIN
iden i ica ion and aining du a ion.
Discussion
!
!
!
130!
The s udy also looked a how high equencies a ec ed SIN unde s anding in
his popula ion g oup. In e es ingly, bo h con ol and a ge g oups had
s a is ically signi ican ly be e esul s when he un il e ed lis s we e used. The
esul s indica e ha pe o mance dec eases when high equency
componen s o speech a e emo ed. S ill, he ATC g oup pe o med be e in
bo h condi ions, hus indica ing an audi o y lea ning e ec , bu he p esen
s udy could no accep ably co ela e i wi h ei he age o wo king yea s.
Ne e heless, he ole o high equencies is ema kable, since i is p esen in
bo h ained and un ained subjec s. This u he suppo s ha equencies
abo e 8 kHz can help humans o be e unde s and SIN. To d aw u he
conclusions, a mo e speci ic expe imen needs o be designed, whe e g oups
o wo ds wi h simila high equency con en s a e selec ed o y and es ablish
a ela ionship be ween wo d iden i ica ion and i s equency con en , because
he speech ma e ial used has balanced phoneme con en .
The implica ions o hese indings a e p o ound because hey can esul in
be e ehabili a ion he apies o people wi h hea ing loss. Ou aging socie y
is bound o su e SIN unde s anding impai men s. As indi iduals g ow olde ,
his abili y deg ades e en be o e audi o y losses become clinically ele an .
The e o e, i is capi al o come up wi h s a egies o y and slow his p ocess
down as much as possible. The esul s om sec ion 6.2 ha e demons a ed
ha daily exposu e o speech in noise can esul in adap a ion and enhanced
pe o mance in NH indi iduals. This is he i s ime ha ATC a e s udied o
his singula abili y. In addi ion o he ecen p oo s ha lea ning du ing
adul hood is a g ea e han p e iously hough , we p o ide e idence ha
audi o y aining unde deg aded sound quali y can lead o be e SIN
unde s anding and can ha e big implica ions in he way hea ing loss is ea ed
oday. Hence, esea ch in he line o his s udy will ce ainly aid in he
de elopmen o e icien and e ec i e aining p o ocols and ma e ials.
Discussion
!
!
!
131!
10.- ELECTRICALLY EVOKED PSYCHOACOUSTICS IN
COCHLEAR IMPLANT RECIPIENTS
10.1.- How physical and psychophysical a iables a ec elec ode
disc imina ion in cochlea implan ecipien s
The aim o his s udy has been o measu e he e ec ha he dis ance
be ween he in achoclea elec odes and he modiolus has on he abili y o a
pa ien o disc imina e he o igin o a ce ain s imulus, sepa a ing his e ec
om hose o he NRT, T le el and impedance, since hese ac o s may also
in luence he abili y o disc imina e elec ical pi ch. The mo i a ion o his
s udy has been ha a be e abili y o disc imina e elec ical pi ch could help o
design be e aining o CI pa ien s, since i has been seen in sec ion 9 how
equency esolu ion plays a majo ole.
The esul s p o ide e idence ha he abili y o disc imina e be ween
elec odes depends on mul iple ac o s. In his s udy, NRT, T le el, impedance
and dis ance we e iden i ied. They we e all s a is ically signi ican in explaining
elec ode disc imina ion.
Especi ically, he s a is ical analysis p o ides empi ical e idence in a ou o
he hypo hesis ha he elec ode dis ance o he inne wall o he cochlea is
a signi ican p edic ing a iable on he elec ode disc imina ion abili y o he
pa ien , measu ed as success a e. The esul s indica e ha he g ea e he
dis ance, he lowe he success a e will be. Thus, dis ance o he inne wall
will esul in mo e di icul ies o pe cei e elec ode di e ences.
The explana ion can be ound in he ex ensi e esea ch ha has been
conduc ed o desc ibe cu en ield pa e ns o a ious elec ode
con igu a ions wi hin he cochlea (171) (172) (173) (105). These expe imen s
show ha he spa ial g adien o he elec ic ield inc eases as he dis ance
be ween he elec odes and he neu al ends dec eases. Thus, an elec ode
ha is close o a su i ing neu al elemen would equi e less cu en o each