Rashmi Raghu (university Of Auckland)

02-18-2003, 09:56 AM

Dear Members

Subject: Computing Bite Forces from Surface Electromyogram Measurements

I am a Master's student working on a project where we are trying to compute

static bite forces from Surface Electromyogram (SEMG) measurements of some of

the jaw muscles (namely the Temporalis and Masseter muscles). At the moment we

are trying to implement a model based on that described by May et al, 2001 ('A

Three-dimensional Mathematical Model of the Temporomandibular Joint Loading',

Clinical Biomech., 16:489-495). We have preliminary SEMG data and

corresponding bite force values from a handful of subjects. We are encountering

problems with our approach and would appreciate any offer of help.

Since the jaw muscle system is indeterminate we are using optimisation

techniques to find bite forces given a few of the muscle forces (which we

estimate from the SEMG measurements). The main constraints of the system are

the equations of static equilibrium (sum of forces = 0 and sum of moments =

0). The forces involved in this system are right & left Temporalis, Masseter,

Medial Pterygoid, Lateral Pterygoid muscles, vertical bite forces plus 3

orthogonal forces at each TMJ (16 forces in total). In order to estimate

muscle forces from SEMGs we are first integrating the signals and normalising

them with respect to their MVC (Maximal Volunatary Contraction) values. This

normalised semg is then multiplied by the maximum possible muscle force (Fmax_i

for the ith muscle) to give an estimate of the force produced by the muscle for

the measured semg.

The two main problems we are having are as follows:

1. The objective functions found for solving this kind of system have involved

only the muscle forces e.g. minimising sum of muscle force estimates squared or

minimising the sum of the squares of the ratio of normalised muscle forces

(normalised force = muscle force/maximum possible muscle force). But these

functions are minimised by simply keeping the unknown (medial and lateral

pterygoid) muscle forces at zero. If this is the case then the bite forces

predicted are not close to the bite forces measured.

So, what we are wondering is whether there is a relationship between activity

level of the temporalis, masseter (from which semgs are measured) and medial

and lateral pterygoid muscles? Alternatively, would it reasonable to use a

calibration step where measured bite forces, temporalis and masseter forces at

a handful of different bite levels are used to compute the other unknowns in

the system to establish some sort of pattern of pterygoid muscle activity at

different bite force levels? If neither of these approaches is reasonable then

is there anything else we may try?

2. The second problem is as follows: the estimates for Fmax_i for each of the

Temporalis and Masseter muscles have a big bearing on the solution. Fmax_i

values are estimated as follows: Fmax_i = K * CSA_i where K = 37 N/cm^2. We

have used various different values for muscle cross-sectional area (CSA_i)

found in literature to estimate the maximum possible muscle forces but they

don't work very well as they are not subject-specific. Unfortunately we don't

have the ability to non-invasively measure muscle CSAs for each subject. So, we

were wondering if we could estimate the maximum forces possible by calibrating

as follows: fix the two bite forces, leave the rest as unknowns and optimise -

this gives a distribution of forces that could be used as Fmax_i estimates

(however, the force assigned to the lateral pterygoid muscles in this case is

likely far lower than what it can maximally produce along its line of action -

probably because we are only considering vertical bite forces). Alternatively,

is there anything other approach we may try?

I apologise for the length of this email and really appreciate anyone's

comments.

Regards,

Rashmi Raghu

Department of Engineering Science

University of Auckland

New Zealand

---------------------------------------------------------------

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl

For information and archives: http://isb.ri.ccf.org/biomch-l

---------------------------------------------------------------

Subject: Computing Bite Forces from Surface Electromyogram Measurements

I am a Master's student working on a project where we are trying to compute

static bite forces from Surface Electromyogram (SEMG) measurements of some of

the jaw muscles (namely the Temporalis and Masseter muscles). At the moment we

are trying to implement a model based on that described by May et al, 2001 ('A

Three-dimensional Mathematical Model of the Temporomandibular Joint Loading',

Clinical Biomech., 16:489-495). We have preliminary SEMG data and

corresponding bite force values from a handful of subjects. We are encountering

problems with our approach and would appreciate any offer of help.

Since the jaw muscle system is indeterminate we are using optimisation

techniques to find bite forces given a few of the muscle forces (which we

estimate from the SEMG measurements). The main constraints of the system are

the equations of static equilibrium (sum of forces = 0 and sum of moments =

0). The forces involved in this system are right & left Temporalis, Masseter,

Medial Pterygoid, Lateral Pterygoid muscles, vertical bite forces plus 3

orthogonal forces at each TMJ (16 forces in total). In order to estimate

muscle forces from SEMGs we are first integrating the signals and normalising

them with respect to their MVC (Maximal Volunatary Contraction) values. This

normalised semg is then multiplied by the maximum possible muscle force (Fmax_i

for the ith muscle) to give an estimate of the force produced by the muscle for

the measured semg.

The two main problems we are having are as follows:

1. The objective functions found for solving this kind of system have involved

only the muscle forces e.g. minimising sum of muscle force estimates squared or

minimising the sum of the squares of the ratio of normalised muscle forces

(normalised force = muscle force/maximum possible muscle force). But these

functions are minimised by simply keeping the unknown (medial and lateral

pterygoid) muscle forces at zero. If this is the case then the bite forces

predicted are not close to the bite forces measured.

So, what we are wondering is whether there is a relationship between activity

level of the temporalis, masseter (from which semgs are measured) and medial

and lateral pterygoid muscles? Alternatively, would it reasonable to use a

calibration step where measured bite forces, temporalis and masseter forces at

a handful of different bite levels are used to compute the other unknowns in

the system to establish some sort of pattern of pterygoid muscle activity at

different bite force levels? If neither of these approaches is reasonable then

is there anything else we may try?

2. The second problem is as follows: the estimates for Fmax_i for each of the

Temporalis and Masseter muscles have a big bearing on the solution. Fmax_i

values are estimated as follows: Fmax_i = K * CSA_i where K = 37 N/cm^2. We

have used various different values for muscle cross-sectional area (CSA_i)

found in literature to estimate the maximum possible muscle forces but they

don't work very well as they are not subject-specific. Unfortunately we don't

have the ability to non-invasively measure muscle CSAs for each subject. So, we

were wondering if we could estimate the maximum forces possible by calibrating

as follows: fix the two bite forces, leave the rest as unknowns and optimise -

this gives a distribution of forces that could be used as Fmax_i estimates

(however, the force assigned to the lateral pterygoid muscles in this case is

likely far lower than what it can maximally produce along its line of action -

probably because we are only considering vertical bite forces). Alternatively,

is there anything other approach we may try?

I apologise for the length of this email and really appreciate anyone's

comments.

Regards,

Rashmi Raghu

Department of Engineering Science

University of Auckland

New Zealand

---------------------------------------------------------------

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl

For information and archives: http://isb.ri.ccf.org/biomch-l

---------------------------------------------------------------