Document [original]

A MOBILE GAIT ANALYSIS SYSTEM FOR OPTIMIZATION

OF PROSTHETIC ALIGNMENTS

Westebbe B1, Thiele J1, Kraft M1

1Technical University Berlin, Germany

b.westebbe@tu-berlin.de

Abstract: The prosthesis alignment is of central importance

for a harmonic gait, especially for upper limb amputees.

Today alignment optimization is based on static measuring

and the experience of the orthopaedic technician. To objec-

tify the alignment process a mobile gait analysis system

based on 10 inertial sensors and a 6 DOF force and moment

sensor was developed at the TU Berlin. The dedicated soft-

ware adds dynamic gait parameters into the optimization

process and guides the necessary changes in the prosthesis

alignment. Therefore 19 common alignment changes were

analysed based on measurements with 2 subjects fitted with

the C-Leg knee. The anterior and posterior displacement

of the knee will be the focus of an additional study with 6

subjects.

Keywords: amputee, upper limb, prostheses alignment,

optimization, mobile gait analysis

Introduction

The correct alignment of prostheses is of central importance

for amputees to avoid asymmetrical stress of the muscu-

loskeletal system [1]. Increased biomechanical stress of the

residual leg can force degenerative processes in its joints [2].

Today prosthetic alignment is supported by means of static

measuring systems (e.g. LASAR Posture, LASAR Assem-

bly) which cannot enable the evaluation of all aspects of the

gait. Thus a system for optimization of prosthetic alignments

by dynamic parameters was developed at the TU Berlin.

Methods

Measuring system: A mobile gait analysis system for am-

putees has been designed by combining inertial sensor based

motion tracking with the Oktapod system [3] for measuring

forces and moments in 6 degrees of freedom in a lower limb

prosthesis. The 10 motion trackers (MTw Wireless by

Xsens) combine accelerometers, gyroscopes and magneto-

scopes. Joint positions and angles are calculated based on the

exploitation of kinematic constraints. The advantage of these

newly developed algorithms is that they are neither depended

on an exact mounting orientation of the sensors in relation to

the body segments nor on exact calibration movements [4].

Data collection: With the mobile gait analysis system meas-

urements of gait on even ground with self-selected velocity

and different prosthetic alignments were performed. Overall

19 alignment variations were analyzed based on data from

two subjects. Subsequently the variations knee anterior and

knee posterior displacement will be further examined by

additional six subjects at the Hannover Medical School. All

subjects used their own socket and were fitted with the

C-Leg and either the 1D35 or the 1C40 (all Otto Bock

HealthCare). For validation purposes all measurements were

conducted in a gait lab (Vicon, System 460, M-Cam; Kistler,

Typ 9287A).

Data processing: The collected data of the different sources

is synchronized and physical properties like joint angles,

joint moments, energy expenditure and the load line are

calculated. Inertial sensors on both feet are used to detect the

gait phases. Based on the forces and moments measured in

the prosthesis a step detection and filter has been imple-

mented. The filter is necessary to avoid that braking and

accelerating steps are included in the further data processing.

Results

The collected gait data varies significantly between subjects.

This is shown in Figure 1 where the measured sagittal mo-

ments of the bench alignment based on manufacturers’ in-

structions, the knee anterior and posterior displacement are

compared. Subject A performs stance phase flexion with the

bench alignment and the alignment with knee displaced

anteriorly shown by the negative sagittal moments that cause

knee flexion. Subject B suppresses this shock absorbing

movement actively, which can be explained by a high need

for safety while walking.

Additionally in Figure 1 the expected moments for the two

alignment variations are shown, which are calculated from

the bench alignment data by adding the torque that results

from the additional lever when the knee centre is shifted.

Thereby compensation strategies of the amputee can be

observed. The expected knee flexion moments while displac-

ing the prosthetic knee anteriorly were not reached by sub-

ject A in the first part of the gait cycle (0-30% gc, ant meas-

ured). The expected moments would cause an exaggerated

stance phase flexion. That would lead to a high energy ex-

penditure why it is avoided by active compensation with the

amputees stump. For subject B this compensation strategy is

not necessary as the sagittal moments are positive with every

alignment in the first part of the stance phase (0-30% gc, ant

measured). Both subjects show, that the potentially acceler-

ated knee flexion in the pre-swing phase (50-60% gc, ant

expected) shown by the expected high knee flexion moment

is avoided to proper initiation of the swing phase flexion.

The collected data showed not only inter-subject variability

but also significant differences depending on the physical

and mental state on the day of testing. Measurements of the

same alignment on different days often showed stronger

distinctions than those of the alignment variations.

Bereitgestellt von | Technische Universität Berlin

Angemeldet

Heruntergeladen am | 27.10.17 11:24

Figure 1: Comparison of the measured and the calculated

sagittal moment for bench alignment, knee anterior and

posterior displacement of subject A and B

Discussion

When developing the expert system it is necessary to build a

knowledge base that defines which change on the alignment

has positive effects and which disturbs harmonic move-

ments. The results show that it is not possible to define an

absolute optimum for every parameter (Table 1), although

some variables are useable for optimization without further

processing. For those parameters that have no absolute opti-

mum it is necessary to define a zero level for each patient as

a starting point for the optimization process. This can either

be a measurement with the bench alignment or one with the

actual alignment that the patient is used to. After the baseline

measurement, alignment variations are performed, following

recommendations of this expert system and the experience of

the orthopaedic technician.

Table 1: Examplary optimization parameters

parameter

adjustment

optimum

Energy consumption

Minimize

Relative

Compensation mechanisms

Minimize

Absolute

Gait symmetry

Maximize

Absolute

Inter-step-variance

Minimize

Absolute

Torsion

Minimize

Relative

Stance phase flexion

Maximize

Absolute

In the first version of the expert system, recommendations

for knee displacement in anterior and posterior direction are

given. Therefore no sophisticated statistical training is

needed to implement a robust decision-making ability of the

system. Compensation effects that for example occur if an

alignment feels insecure can disturb the optimization process

and have to be detected properly. In order to be able to de-

cide if extensive compensation effects appeared, the expert

system has to know which change in the alignment has been

made to calculate the difference between the expected and

actual values. If the compensation is predominating, not an

alignment change will be recommended but an adjustment in

patient behaviour.

The final decision if the alignment respectively a parameter

is fully optimized has to be made by the orthopaedic techni-

cian since the functional capability of each patient cannot be

described comprehensively to the system (Figure 2).

Bench-/Actual-

Alignment

Variation

measuring system

computer

orthopaedic technician

gait data parameters

evaluation recommendation

data

OPTIMUM

Figure 2: Flowchart of expert system

In the future it is planned to expand the expert system to give

recommendations for other alignment optimization possi-

bilities beside knee displacement in anterior and posterior

direction. Also online data processing is in development

to accelerate user feedback for patients and orthopaedic

technicians.

Acknowledgement

This work is supported by the German Federal Ministry of

Education and Research (BMBF) within the cooperative

research project mebGO (grant 13EZ1112A) and the Otto

Bock HealthCare GmbH.

Bibliography

[1] L. Fang, X. Jia, and R. Wang, “Modeling and

simulation of muscle forces of trans-tibial amputee to

study effect of prosthetic alignment,” Clin Biomech

(Bristol, Avon), vol. 22, no. 10, pp. 1125–1131, 2007.

[2] R. Gailey, K. Allen, J. Castles, J. Kucharik, and M.

Roeder, “Review of secondary physical conditions

associated with lower-limb amputation and long-term

prosthesis use,” Journal of rehabilitation research

and development, vol. 45, no. 1, pp. 15-30, 2008.

[3] S. Oehler, M. Kraft, and M. Pusch, “Mobilitäts- und

Belastungsmessungen an Oberschenkelamputierten

Ergebnisse einer einjährigen Studie,” Biomedizinische

Technik/Biomedical Engineering, vol. 55, no. S1, pp.

63–66, 2010.

[4] T. Seel, T. Schauer, and J. Raisch, “Joint axis and

position estimation from inertial measurement data by

exploiting kinematic constraints,” in IEEE Multi-

Conference on Systems and Control, 2012, pp. 45–49.

Bereitgestellt von | Technische Universität Berlin

Angemeldet

Heruntergeladen am | 27.10.17 11:24