Stylohyoid and Posterior Digastric Measurement with Intramuscular

EMG, Submental EMG and Swallowing Sound

Adrien Mialland

1 a

, Ihab Atallah

and Agn

es Bonvilain

Univ. Grenoble Alpes, CNRS, Grenoble INP, Gipsa-Lab, 3800 Grenoble, France

Otorhinolaryngology, CHU Grenoble Alpes, 38700 La tronche, France

Keywords:

Stylohyoid, Digastric, Electromyography, Swallowing Sound, Implantable Active Artiﬁcial Larynx.

Abstract:

The stylohyoid and the posterior digastric muscles have essentially been measured through indirect imaging

method because of the difﬁculty to measure them. They are small neck muscles, close to each other, that

cannot easily be accessed independently. Yet, they showed promising results for a robust and safe indwelling

detection of swallowing, both in terms of timing and recruitment. The possibility to thoroughly establish

their relevance through their direct functional analysis would enable the development of an implantable active

artiﬁcial larynx, that would protect the airway during swallowing detection. Therefore, we set up the ﬁrst

standardized procedure that allows their direct measurement through intramuscular electromyography (EMG)

and that we report in this paper. We also used submental surface EMG and swallowing sound modalities

to access the major time points of the swallowing process. Finally, various exercises, along with swallowing,

were performed by the volunteers. 16 peoples were measured with our new procedure, and both the stylohyoid

and the posterior digastric could be measured independently with no difﬁculty. Timings and tasks comparison

are therefore ongoing.

1 INTRODUCTION

The past decades have provided detailed descriptions

of the swallowing mechanism, its anatomical struc-

tures and their complex interplay (Shaw and Martino,

2013). But swallowing studies still tend to mostly

report on muscles activity during swallowing tasks

only, most likely because of the need to better under-

stand the swallowing process for a clinical practice.

However, this does not allow to draw conclusions in

a broader perspective to get an extensive picture of

muscles timing and recruitment, with regard to vari-

ous tasks. Besides, the available measurement meth-

ods limit the possibilities. The neck muscles are usu-

ally small and close to each other, and the traditional

surface electromyography (EMG) approach may lead

to measurements that contain unwanted contamina-

tion from adjacent muscles (cross-talk). Therefore,

several muscles got little attention because of the dif-

ﬁculty to measure them independently (Steele, 2015).

Yet, we aim at the feasibility of an implantable ac-

tive artiﬁcial larynx through the real-time, robust and

safe indwelling detection of swallowing, which needs

https://orcid.org/0000-0001-5359-674X

high quality signals. This would be beneﬁcial in the

context of laryngeal removal, known as total laryn-

gectomy, that requires the trachea to be sewn on the

anterior neck, to create a tracheostomy. Indeed, as

the laryngeal functions are lost, breathing can only be

safely performed through an isolated tract. But it adds

several adverse effects and the air no longer passes

through the nose and the mouth anymore, which al-

lowed ﬁltration, warming, humidiﬁcation, olfaction

and acceleration of the air for better tissue oxygena-

tion (Maclean et al., 2009). Therefore, these func-

tions could be restored if the trachea could be set back

in place and protected with an active closure mecha-

nism, during swallowing. This requires the develop-

ment of a robust and real-time swallowing detection

strategy, based on muscles that activate early, provide

stable and dedicated activation pattern, and are not al-

tered by the surgery.

In that regard, we focused on the stylohyoid and

the posterior digastric muscles that showed indirect

but appropriate results. Indeed, few studies hint to-

ward their importance. First, on animals the stylohy-

oid have been shown to activate at the beginning of

the swallowing, at the same time than the submental

muscles (German et al., 2009). On humans, imag-

Mialland, A., Atallah, I. and Bonvilain, A.

Stylohyoid and Posterior Digastric Measurement with Intramuscular EMG, Submental EMG and Swallowing Sound.

DOI: 10.5220/0011628100003414

In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 4: BIOSIGNALS, pages 48-54

ISBN: 978-989-758-631-6; ISSN: 2184-4305

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

ing methods conﬁrmed these results (Okada et al.,

2013) and the analysis of their anatomical structure

revealed a signiﬁcant potential to move the larynx

(Pearson et al., 2011), but direct functional data are

lacking. Second, both muscles were studied as a

whole STH-PD complex, which was shown to mainly

activate during swallowing and jaw opening (Kurt

et al., 2006). The authors also studied their nerve

conduction in comparison to the mylohyoid, which

is part of the submental muscles. They argued that

”electrophysiological identiﬁcation factors may make

easier to work on STH–PD muscles”. In addition,

contractile patterns from animals showed a repeatable

and stable activity of the stylohyoid (Thexton et al.,

2012). Third, these muscles are both directly acces-

sible during the surgery, with no further impairment

required. Alternatively, other muscle could have been

considered for a robust an safe detection of swallow-

ing, but several uncertainties mitigate their interest.

The mylohyoid is part of the submental muscles and

demonstrates the greatest potential in laryngeal supe-

rior movements (Pearson et al., 2011). But it is also

part of the ﬂoor of the mouth and activate for var-

ious tasks, to support the tongue. Also, infrahyoid

muscles are easily accessible during the surgery but

could be removed with the larynx, and may exhibit

a large range of activation for various tasks (German

et al., 2009). Finally, the pharyngeal muscles are ei-

ther impaired during the surgery or would require to

place the sensors deep inside the neck, and therefore

to further damage the remaining functional anatomy

(Lippert et al., 2016).

Besides, the swallowing process have been char-

acterized through various methods that give access to

the major events. In particular, the submental muscles

are acknowledged to lead the reﬂexive part of swal-

lowing and their activation assists the tongue in the

propulsion of the bolus posteriorly (Park et al., 2017).

These muscles are commonly measured through sur-

face electromyography and are considered mostly free

of contamination from adjacent muscles (cross-talk)

(McKeown et al., 2002). Also, the forces applied to

the bolus produce various sound components that are

linked to bolus locations and anatomical events. Es-

pecially, the passage of the bolus through the upper

esophageal sphincter (UES) generates a speciﬁc burst

and is the only sound component to be found 100% of

the time (Morini

ere et al., 2008). Recently, the begin-

ning of that event has been suggested to be linked to

the closure of the laryngeal vestibule (Kurosu et al.,

2019), which protects the airway from bolus aspi-

ration. Therefore, these tow measurement methods

would allow to correlate the stylohyoid and posterior

digastric muscles activity with key swallowing events.

So, this paper intends to describe the measurement

protocol we set up to enable the evaluation of the sty-

lohyoid and the posterior digastric muscles with intra-

muscular EMG. Submental surface EMG and swal-

lowing sound measurement methods were also used

to respectively access the beginning of the swallow-

ing and the moment the bolus goes through the UES.

2 MATERIALS AND METHODS

2.1 Subjects

Twelve healthy adults (8 males/ 8 females) with no

history of dysphagia, neck surgery, immune deﬁ-

ciency or any neurological impairment participated in

this study. The mean age was 36.1 ± 13, 6 and all

enrolled participant met the following inclusion cri-

teria: age of 18 years or higher and a body mass

index (BMI) of 25 or lower. Participants with BMI

higher than 25 were intentionally excluded from this

study to avoid a potentially excessive amount of fat

in the region of the neck that would make the proper

placement of the sensors more complex. Each par-

ticipant received a detailed explanation of the study

and were made aware of their rights, including the

possibility to withdraw at any time. Each participant

returned a written informed consent prior to the par-

ticipation. Each participant underwent a standardized

examination conducted by a registered otolaryngolo-

gist to conﬁrm the absence of any sign of dysphagia.

Each participant were anonymized and no identifying

information were collected. All participant accom-

plished the full protocol. This study was approved

by the research ethics committee of Sud-M

editerran

III of N

ımes in France (Protocol ID: 38RC22.0096).

This study has been carried out in accordance with

the Declaration of Helsinki of the World Medical As-

sociation revised in 2013 for experiments involving

humans.

2.2 Sensors Placement

Prior to the sensor placement, the beard must be

shaved, and the neck area must be cleared from any

jewelry and clothing items that could interfere. Dur-

ing any sensor placement, the participant was asked

for any discomfort and a swallowing was performed

to conﬁrm the absence of pain. The participants were

asked to comfortably sit on a chair and each sensors

were placed (Figure 2) in the following order:

SWALLOWING SOUND: The cricoid cartilage

was ﬁrst located by the otolaryngologist and the ac-

celerometer was placed at its center, as it was sug-

Stylohyoid and Posterior Digastric Measurement with Intramuscular EMG, Submental EMG and Swallowing Sound

cricoid cartilage

hyoid bone

stylohyoid

posterior

digastric

intermediate

tendon

mylohyoid

digastric

sEMG

iEMG

ACC

Analog input connectors

16 bits ADC, 4 kHz

BP 10-500 Hz

BP 10-1000 Hz

LP 2000 Hz

Intrumentation

amplifier x4

Figure 1: Sensors locations, anatomical structures and acquisition setup. EMG: electromyography, sEMG: surface EMG,

iEMG: intramuscular EMG, ACC: accelerometer, BP: band-pass, LP: low-pass. ADC: analogue digital converter. The targeted

anatomical regions are shown with ellipses for each sensors. 1 needle is inserted in the stylohyoid at the level of the lesser

horn of the hyoid bone, next to the intermediate tendon which do not generate myoelectric signal. 1 needle is inserted in

the posterior digastric, where it separates from the stylohyoid. 1 surface EMG measures the submental muscles activity. 1

accelerometer measures the swallowing sound from the top of cricoid cartilage. Each signals are then passed through an

instrumentation ampliﬁer, band-pass ﬁltered and digitized at 4kHz by a 16 bits ADC, before being sent to the laptop.

gested to act as a resonator for the sound (Takahashi

et al., 1994; Cichero and Murdoch, 2002). The ac-

celerometer was ﬁxed with a hypoallergenic paper

tape, so as to record anterior-posterior vibrations. The

length of the tape was chosen so as to span the neck

horizontally from one lateral side to the opposite side.

The tension was adjusted to get a clear event when a

swallowing is performed.

SURFACE EMG: The 2 differential electrodes

were placed under the left part of the submental area,

with their center approximately 2cm apart, and the

ground electrode was placed over the right clavicle.

Both areas were ﬁrst cleaned with a dedicated abra-

sive and conductive paste to reduce the electrode-skin

impedance.

INTRAMUSCULAR EMG: Both the stylohyoid

and the posterior digastric muscles are directly acces-

sible behind the skin, but are close to each other. To

isolate their activity, they are targeted close to their

origin and insertion points: the stylohyoid muscle was

targeted next to its insertion point, at the level of he

junction between the body and the greater horn of the

hyoid bone. This point could be identiﬁed by pal-

pation of the lesser horn of the hyoid, which is at-

tached by its base to the angle formed by the junc-

tion between the body and greater horn. The EMG

needle punctures the skin perpendicularly at the in-

sertion point of the stylohyoid bone and then directed

upward and backward. The posterior digastric mus-

cle was targeted in its posterior portion, next to its

origin from the mastoid notch. The EMG needle is

inserted perpendicularly behind the vertical ramus of

the mandible and in front of the anterior border of

the sternocleidomastoid muscle. It is then directed

upward, medially and backward. The anatomical re-

gions of both the site of puncture are located through

palpation and were marked on the skin before inser-

tion. Besides, while the otolaryngologist was slowly

inserting the needles, a second trained operator mon-

itored the EMG activity on the computer to look for

muscle activities. It could be requested to the partici-

pant to swallow to elicit an event. Once an event were

visible on the signal, the needles were no further in-

serted. They were then secured with a Steri-Strip so

that they cannot come out of the muscles.

submandibular

accelerometer

posterior

digastric

stylohyoid

Figure 2: Sensor placement. The stylohyoid and the pos-

terior digastric muscles are measured with intramuscular

EMG, the submental muscles with surface EMG, and the

swallowing sound with an accelerometer. The needles are

secured with steri strips so that they cannot come out.

2.3 Signal Acquisition

From each participant 4 signals were recorded: 2 in-

tramuscular EMG with concentric needle electrodes

(27-gauge, 30 mm), 1 surface EMG with 3 surface gel

electrodes (2 differential and one ground), and 1 swal-

BIOSIGNALS 2023 - 16th International Conference on Bio-inspired Systems and Signal Processing

lowing sound signal with an single axis accelerom-

eter (Pulse Transducer TN1012/ST, 1600Hz, ADIn-

strument). The 3 EMG signals were fed to a Bio-Amp

(FE234, 4 channels, ADInstrument) pre-Ampliﬁer

with differential inputs, which is in turn connected

to a PowerLab (35 series, 4 channels, ADInstrument)

data acquisition system. The accelerometer were di-

rectly connected to the PowerLab with a built-in ded-

icated connector. On the computer, the LabChart

ADInstrument data acquisition and analysis software

receives the signals from the PowerLab. This set up

allows to have all 4 signals synchronized in time.

Then, the Bio-Amp ﬁlters the 2 intramuscular EMG

signals with a 10 − 1000Hz analogue band-pass ﬁl-

ter and the surface EMG signal with a 10 − 500Hz

analogue band-pass ﬁlter. The swallowing sound sig-

nal is also ﬁltered, with a 2000Hz low-pass analogue

ﬁlters. Finally, each channel uses a 16 bits analogue-

digital-converter (ADC) and all signals were sampled

at 4000Hz, which simpliﬁed the ﬁle format and al-

lowed down-sampling if required.

2.4 Protocol

Both swallowing and non-swallowing tasks were per-

formed by participants. The task is ﬁrst prepared by

asking the participant to take the bolus into the mouth

and/or to adopt a relaxed position. Then, the signal

acquisition is launched and 2 second with no motion

is recorded. The participant is then asked to perform

the task and 2 more second with no motion are ac-

quired. This ensure to record a clear event. Also,

when a task is performed, the keyboard is pressed to

place a marker on the event, which facilitates its loca-

tion in post-processing.

SWALLOWING TASKS: Participants were asked to

perform 5 swallowing of Saliva, water (10ml), thick

liquid (compote) and solid bolus (madeleine).

NON-SWALLOWING TASKS: Participant were

asked to perform (1) 3 times mouth opening, lips

purseing, teeth clenching, smiling, whistling, cough-

ing, blowing through a straw, counting from 1 to 10,

saying ”iii” in ascending and descending order. (2)

3 times movement tasks being jaw movements, lat-

eral head movements and head extension and ﬂexion.

They are performed in neutral - right - left - neutral

or neutral - extension - ﬂexion - neutral order depend-

ing on the task. (3) 5 times chewing, that are actually

recorded at the moment the solid bolus swallowing

tasks is performed. To separate the two, the partic-

ipant is asked to hold the bolus in the mouth once

chewed, ready to swallow, before being asked to swal-

low it. All tasks are performed to a comfortable full

extent and at a natural pace.

2.5 Signal Processing

Ofﬂine analysis was performed with a custom piece

of software. A 2

order high-pass Butterworth dig-

ital ﬁlters with a cut-off frequencies of 20Hz is ap-

plied on the EMG signals and a 4

order Butterworth

notch ﬁlters is applied to all signals to eliminate the

50Hz line noise. The baselines are then delineated to

include 1 second of recording that contains no event

of any sort. Next, all signals are transformed using

the Teager-Kaiser energy operator (TKEO) Ψ[x(n)] =

x(n)

− x(n− 1)x(n +1), which has been shown to im-

prove the SNR (Li et al., 2007). Then, the following

are extracted:

MUSCLE ONSET/OFFSET: We used a method

based on generalized likelihood ratio (GLR) (Xu

et al., 2013). Using a 200ms sliding window of size n

samples, it continuously computes the likelihood ra-

tio of a contracted muscle state hypothesis H

against

a relaxed muscle state hypothesis H

. This allows to

estimate the time instant r of muscle onset and offset,

based on a series of observations y

= [y(0), ..., y(n)]

from the current window. A probability density func-

tion (PDF) P

(y(t)) and P

(y(t)) is associated with

each hypothesis, and the likelihood that y

is drawn

from one of these PDF is expressed in Equation 1.

) = p

) =

∏

t=0

(y(t)) (1)

Where i is the type of hypothesis. The likelihood

ratio is then expressed in Equation 2.

(r, y

)

r−1

)

∏

t=r

(y(t))

(2)

Where L

(r, y

) is the likelihood that y

is drawn

from both PDF, with H

starting from time instant r.

The likelihood ratio is then maximize over the possi-

ble transition time r of the current window. Therefore,

this requires to know the distribution followed by the

data and we opted for an exponential PDF. Indeed, on

the basis of empirical evidences, it closely represents

the TKEO transformed EMG, as long as its absolute

value is considered, which is in line with recent inves-

tigations (Selvan et al., 2018). An adaptive Threshold

is then applied and is chosen to be 10% of the max-

imum current GLR values. The onset and offset are

then located manually according to the threshold, to

avoid any false positive detection.

BOLUS THROUGH UES: the moment the bolus

enters the UES corresponds to the major sound burst

and is suggested to be linked to the closure of the la-

ryngeal vestibule (Kurosu et al., 2019). This allows

to locate the ultimate time point where the airway

Stylohyoid and Posterior Digastric Measurement with Intramuscular EMG, Submental EMG and Swallowing Sound

must be protect. The transformation of the signal with

TKEO allows to bring this event out and facilitate the

manual location of its beginning.

3 RESULTS AND DISCUSSION

The goal of this paper was to describe the protocol

we set up that made feasible the measurement and the

evaluation of the stylohyoid and the posterior digas-

tric muscles. Very few studies focused on those mus-

cles and none enabled the investigation of both their

timing and contraction patterns.

Figure 3: Raw signals of a saliva swallowing. SH: sty-

lohyoid, PD: posterior digastric, SUB: submental, ACC:

accelerometer. All signals provide distinguishable events.

The burst of higher frequency in the ACC signal corre-

sponds to the moment where the bolus goes through the up-

per esophageal sphincter.

The challenging aspect of our protocol is the

proper placement of the concentric needles, to record

the stylohyoid and the posterior digastric muscles sep-

arately. Even though concentric needles are very se-

lective, both muscles are thin and close to each other,

which may lead to the wrong muscle to be measured.

Therefore, we chose to target the stylohyoid close to

its insertion point, at the level of the lesser horn of

the hyoid bones. In that region, the posterior digastric

is composed of its intermediate tendon, which can-

not produce EMG signal and therefore cannot be mis-

taken for the stylohyoid EMG signal. Besides, the

posterior digastric muscle was targeted close to its ori-

gin at the level of the mastoid notch. In that region, it

separates from the stylohyoid, which originates from

the styloid process upper in the neck. In addition,

these anatomical regions were found through palpa-

tion, and the decision to exclude the participants with

a BMI upper than 25 limited the amount of fat and

allowed to ﬁnd them with no difﬁculty. Finally, these

muscles are the ﬁrst to be accessible behind the skin.

Once an activity were visible on the screen, the nee-

dle were not inserted any further and were secured

with steri strips. The needles were, therefore, ﬁxed in

depth but could follow the anatomical movement. The

present paper therefore provide the ﬁrst standardized

procedure that allows to measure both the stylohy-

oid and the posterior digastric muscles independently.

In the past, only one study provided direct functional

analysis of those muscle through intramuscular EMG

(Kurt et al., 2006), but they were only measured as a

whole STH-PD complex, with no possibility to differ-

entiate their activity, both in terms of timing and re-

cruitment. Also, the authors mainly focused on nerve

conduction and did not directly seek to evaluate their

timing. We therefore expend those points with our

new procedure, which provide independent signals.

Figure 4: Signal processing results. Left: raw signals.

Right: processed signals. Dashed lines: onset and offset.

Solid lines: the moment when the bolus goes through the

upper esophageal sphincter (UES). The dashed lines loca-

tions are found with the generalized likelihood ratio (GLR)

method, applied on the EMG (top 3) signals. The solid line

location is found with the Teager-Kaiser energy operator

(TKEO), applied on the accelerometer (bottom) signal.

With regard to the swallowing sound, we focused

on its major component. Early investigation of that

event showed its rise in frequency content compared

to the rest of the signal (Hamlet et al., 1990). The

authors suggested that this was the result of the in-

crease in pressure generated by the contraction of

the surrounding muscles that forces the bolus through

the UES. More recently, modern data analysis meth-

ods conﬁrmed the particular frequency content of that

event (Lee et al., 2008) and formerly linked its oc-

currence with the bolus that passes through the UES

(Morini

ere et al., 2008). This event was also the only

one to occur 100% of the time. Finally, the begin-

ning of that event is suggested to be temporal linked

to the closure of the laryngeal vestibule (Kurosu et al.,

2019). This later event is of prime interest to deﬁne

the ultimate moment when the airway are physiolog-

BIOSIGNALS 2023 - 16th International Conference on Bio-inspired Systems and Signal Processing

ically required to be closed. This would in turn de-

ﬁne the timing for an active closure mechanism to be

closed as well. Therefore, the use of TKEO is jus-

tiﬁed by its ability to highlight the high frequency

contents, and were particularly effective in this study

(Figure 4). However, one may argue about the lack of

bandwidth analysis of the swallowing sounds. Indeed,

only its upper limits have been investigated and no

recommendation allows to formerly delineate a par-

ticular sound component. Yet, these considerations

usually arise when it comes to the precise analysis

of the swallowing process. But the main UES sound

component clearly stands out and the TKEO essen-

tially magniﬁes its frequency content. Besides, the

swallowing sound modality allows to avoid the use

of videoﬂuoroscopy, which requires X-ray exposure.

Therefore, these elements make the swallowing sound

and the TKEO operator, an effective and safe combi-

nation to deﬁne the ultimate closing time, from the

perspective of an implantable active artiﬁcial larynx.

The related time marker were then set manually with

no difﬁculty and is visible Figure 4.

Besides, we looked for an semi-automated muscle

activity segmentation to reliably determine their onset

and offset points. But the literature provides a wide

range of algorithm and none has shown to be effec-

tive for all the various bio-mechanical application that

uses EMG (Tenan et al., 2017). So, we used the GLR

for its ability to modelize the signals and therefore

its robustness to various background noise (Xu et al.,

2013). As an example, intramuscular EMG requires

the insertion of a needle inside the muscles through-

out the whole acquisition. The presence of this in-

vasive foreign body may cause the reﬂexive contrac-

tion of relatively few muscle ﬁbers, which can be vis-

ible on the signal as spurious background spikes (Fig-

ure 3). Besides, it allows to adapt to the channel being

analyzed so that every muscle activity are compared

on the same basis, which limits possible biases.

Finally, the various tasks performed by the partic-

ipants allowed to access a large range of every day

contraction pattern. However, this obviously does not

represent free-living signals because we seek to for-

merly characterize the stylohyoid and the posterior di-

gastric, with regard to the criteria described section 1.

Therefore, each tasks were performed with no addi-

tional motion to get a clear event on the signals. First

sight on the data conﬁrmed the literature on the ac-

tivation of the stylohyoid and the posterior digastric

during swallowing. Further analysis are currently car-

ried out to provide extensive results.

4 CONCLUSION

Very few studies focused on the functional analy-

sis of the stylohyoid and the posterior digastric mus-

cle because of their location. Also, none sought

to formerly characterize both their timings and con-

traction patterns in various tasks, including swallow-

ing. Yet, those muscles showed promising results

for the development of an implantable active artiﬁ-

cial larynx, through the robust and safe real-time de-

tection of swallowing. Therefore, to formerly evalu-

ate their potential independently, we described in this

paper the precise acquisition protocol that we set up

with the use of intramuscular EMG, submental sur-

face EMG and swallowing sound measurement meth-

ods. Intramuscular EMG were used to obtain the

activity of both the stylohyoid and the posterior di-

gastric independently, with the insertion of the nee-

dles following the localization of precise anatomical

structure. In addition, submental surface EMG and

swallowing sound allowed to locate the beginning

of the swallowing and the moment the bolus passes

through the upper esophageal sphincter, respectively.

All these modalities give access to various key swal-

lowing events for a comprehensive characterization.

16 people underwent the full procedure and the volun-

teers were asked to perform a large variety of activity,

along with swallowing, to evaluation the recruitment

speciﬁcity of the stylohyoid and the posterior digas-

tric muscles. Besides, timing investigation requires

the localization of onset and offset of muscle activ-

ity and we proposed the use of a generalized likeli-

hood ratio approach. It allowed to effectively abstract

from background noise, following the modelisation of

baseline and muscle activities. This paper therefore

allowed us to access speciﬁc swallowing signals, to

investigate the potential utility of both the stylohyoid

and posterior digastric muscles in the real-time de-

tection of swallowing. Further work are on-going to

provide extensive characterization.

ACKNOWLEDGMENTS

This research has been carried out with funding from

the R

egion Auvergne Rh

one Alpes.

REFERENCES

Cichero, J. A. and Murdoch, B. E. (2002). Detection of

Swallowing Sounds: Methodology Revisited. Dys-

phagia, pages 40–49.

German, R. Z., Crompton, A. W., and Thexton, A. J. (2009).

Stylohyoid and Posterior Digastric Measurement with Intramuscular EMG, Submental EMG and Swallowing Sound

Integration of the Reﬂex Pharyngeal Swallow Into

Rhythmic Oral Activity in a Neurologically Intact Pig

Model. Journal of Neurophysiology, pages 1017–

1025.

Hamlet, S. L., Nelson, R. J., and Patterson, R. L. (1990).

Interpreting the Sounds of Swallowing: Fluid Flow

through the Cricopharyngeus. Annals of Otology, Rhi-

nology & Laryngology, 99:749–752.

Kurosu, A., Coyle, J. L., Dudik, J. M., and Sejdic, E. (2019).

Detection of Swallow Kinematic Events From Acous-

tic High-Resolution Cervical Auscultation Signals in

Patients With Stroke. Archives of Physical Medicine

and Rehabilitation, pages 501–508.

Kurt, T., G

urg

or, N., Sec¸il, Y., Yıldız, N., and Ertekin, C.

(2006). Electrophysiologic identiﬁcation and evalu-

ation of stylohyoid and posterior digastricus muscle

complex. Journal of Electromyography and Kinesiol-

ogy, 16:58–65.

Lee, J., Steele, C. M., and Chau, T. (2008). Time and

time–frequency characterization of dual-axis swal-

lowing accelerometry signals. Physiological Mea-

surement, page 1105.

Li, X., Zhou, P., and Aruin, A. S. (2007). Teager–Kaiser

Energy Operation of Surface EMG Improves Muscle

Activity Onset Detection. Annals of Biomedical Engi-

neering, pages 1532–1538.

Lippert, D., Hoffman, M. R., Britt, C. J., Jones, C. A.,

Hernandez, J., Ciucci, M. R., and McCulloch, T. M.

(2016). Preliminary Evaluation of Functional Swal-

low After Total Laryngectomy Using High-Resolution

Manometry. Annals of Otology, Rhinology & Laryn-

gology, 125:541–549.

Maclean, J., Cotton, S., and Perry, A. (2009). Post-

Laryngectomy: It’s Hard to Swallow. Dysphagia,

pages 172–179.

McKeown, M. J., Torpey, D. C., and Gehm, W. C. (2002).

Non-invasive monitoring of functionally distinct mus-

cle activations during swallowing. Clinical Neuro-

physiology, pages 354–366.

Morini

ere, S., Boiron, M., Alison, D., Makris, P., and Beut-

ter, P. (2008). Origin of the Sound Components Dur-

ing Pharyngeal Swallowing in Normal Subjects. Dys-

phagia, pages 267–273.

Okada, T., Aoyagi, Y., Inamoto, Y., Saitoh, E., Kagaya, H.,

Shibata, S., Ota, K., and Ueda, K. (2013). Dynamic

change in hyoid muscle length associated with trajec-

tory of hyoid bone during swallowing: analysis using

320-row area detector computed tomography. Journal

of Applied Physiology, pages 1138–1145.

Park, D., Lee, H. H., Lee, S. T., Oh, Y., Lee, J. C.,

Nam, K. W., and Ryu, J. S. (2017). Normal contrac-

tile algorithm of swallowing related muscles revealed

by needle EMG and its comparison to videoﬂuoro-

scopic swallowing study and high resolution manom-

etry studies: A preliminary study. Journal of Elec-

tromyography and Kinesiology, pages 81–89.

Pearson, W. G., Langmore, S. E., and Zumwalt, A. C.

(2011). Evaluating the Structural Properties of

Suprahyoid Muscles and their Potential for Moving

the Hyoid. Dysphagia, pages 345–351.

Selvan, S. E., Allexandre, D., Amato, U., and Yue, G. H.

(2018). Unsupervised Stochastic Strategies for Ro-

bust Detection of Muscle Activation Onsets in Surface

Electromyogram. IEEE Transactions on Neural Sys-

tems and Rehabilitation Engineering, 26:1279–1291.

Shaw, S. M. and Martino, R. (2013). The Normal Swallow:

Muscular and Neurophysiological Control. Otolaryn-

gologic Clinics of North America, pages 937–956.

Steele, C. M. (2015). The Blind Scientists and the Elephant

of Swallowing: A Review of Instrumental Perspec-

tives on Swallowing Physiology. Journal of Texture

Studies, pages 122–137.

Takahashi, K., Groher, M. E., and Michi, K.-i. (1994).

Methodology for detecting swallowing sounds. Dys-

phagia, pages 54–62.

Tenan, M. S., Tweedell, A. J., and Haynes, C. A. (2017).

Analysis of statistical and standard algorithms for de-

tecting muscle onset with surface electromyography.

PLOS ONE, page e0177312.

Thexton, A. J., Crompton, A. W., and German, R. Z. (2012).

EMG activity in hyoid muscles during pig suckling.

Journal of Applied Physiology, pages 1512–1519.

Xu, Q., Quan, Y., Yang, L., and He, J. (2013). An Adaptive

Algorithm for the Determination of the Onset and Off-

set of Muscle Contraction by EMG Signal Processing.

IEEE Transactions on Neural Systems and Rehabilita-

tion Engineering, pages 65–73.

BIOSIGNALS 2023 - 16th International Conference on Bio-inspired Systems and Signal Processing