View Full Version : Summary about "Event detection during treadmill walking"

08-11-2006, 02:56 AM
Dear All,

I have attached a summary from your kind answers to my question about "Event detection during treadmill walking". To be more clear I have also added the original question.

With re grads

Behdad Tahayouri

Original question:

I am looking for the techniques which are used to determine "initial contact" and "end contact" events during each cycle of treadmill walking based on the kinematic data. Also I appreciate if you could introduce some of the references which discus validating methods for these techniques or even methods used to validate event detection with “foot switches".

Thanks a lot

B. Tahayori

Summaru of the answers:

From: "Omar Mian"

Here are some papers that you may find useful

Ghoussayni, S., Stevens, C., Durham, S. & Ewins, D. 2004. Assessment and
validation of a simple automated method for the detection of gait events and
intervals. Gait Posture 20, 266-272.

Tirosh, O. & Sparrow, W.A.T. 2003. Identifying heel contact and toe-off
using forceplate thresholds with a range of digital-filter cutoff
frequencies. J Appl Biomech 19, 178-184.

Hreljac, A. & Marshall, R.N. 2000. Algorithms to determine event timing
during normal walking using kinematic data. Journal of Biomechanics 33,

Mickelborough, J., van der Linden, M.L., Richards, J. & Ennos, A.R. 2000.
Validity and reliability of a kinematic protocol for determining foot
contact events. Gait Posture 11, 32-37.



Omar Mian PhD
Institute for Biophysical and Clinical Research into Human
Movement Manchester Metropolitan University
Hassall Road
Alsager, ST7 2HL
United Kingdom

E-mail: O.Mian@mmu.ac.uk
Tel: +44 (0)161 247 5515
Fax: +44 (0)161 247 6375
Web: http://www.irm.mmu.ac.uk/

Date: Thu, 06 Jul 2006 09:40:28 +0100
From: Brian Caulfield
Subject: Re: [BIOMCH-L] Event detection during treadmill walking

I can certainly help you with heel strike detection.
We have recently conducted some work in our lab to identify a method. We
searched the literature last year and found nothing that was acceptable for
treadmill walking. However Hreljac and his team have done a lot of work on
event detection during overground walking and running. His methods have been
validated on a small number of subjects (2?) and are accurate to approx 5ms.
We found that heel strike during treadmill walking corresponded to the point
where the vertical component of acceleration of the lateral heel marker
crossed zero. In tests of this method (aginast footswitches) we found it
to be accurate to within 6ms. We identify the period firstly by means of
looking at the stick figure and then identify the exact point. If your
sampling frequency is not high enough then you should use an algorithm to
correct for timing errors related to low sampling frequencies (I think that
this will be necessary if you have a frequency lower than 200hz). Hreljac
has outlined a very good correction algorithm in his papers.
Other authors have looked at other methods but they are usually only
accurate to within 20-30ms, which is not acceptable in my opinion.
As regards detecting toe off I have used the point of maximum knee flexion
angular velocity in previous work. It looks accurate against stick figure
and video analysis but I have not seen any work that has validated this
Let me know how you get on or if you come across any better methodologies

Dr Brian Caulfield
University College Dublin, Ireland

Date: Thu, 06 Jul 2006 11:39:05 +1000 (EST)
From: Mohammad Reza Fotoohabadi

Subject: Re: [BIOMCH-L] Event detection during treadmill walking

Dear Behdad,
I actually have read the paper of you supervisor, Dr. Fariba
Bahrami, titled" Biomechanical analysis of sit-to-stand in healthy and
paraplegic subjects, Clinical Biomechanics 15, 123-133, (2000)". I am
working on sit-to-stand (STS) in young and pathologic conditions such as
hip OA, or LBP. In our laboratory, we used PEAK Motion Measurement System
to investigate both kinematic and kinetic of different motions. For
instance, I used the x,y coordinates of markers to determine the "Start,
Lift-off, and End" events of STS.

I think that using any switch is not precisely help to determine an event,
since it has some "time lag", however if you do not have access to
complicated instruments such as, PEAK or Vicon, that one has to be the
only choice. Then you have to validate your method of measurement.

At the moment, have a look at our article (attached) to get an idea.

If you are Behdad, then pass my regards to all our friends as well as your
colleagues, particularly Dr. Bahrami whom I used her paper for the
attached article and my PhD thesis.

Take care,

Mohammad Reza Fotoohabadi

Ph.D Student (Scholarship)
School of Physiotherapy,
Faculty of Medicine, Dentistry and Health Science
Victoria, Australia 3010

Telephone: +61 (3) 8344 8802 & +61 (3) 8344 4171
Mobile : 0401913763
Fax : +61 (3) 8344 4188
University: http://www.unimelb.edu.au/
School: http://www.physioth.unimelb.edu.au/

Date: Wed, 05 Jul 2006 17:14:04 -0400
From: "Andrew Kwarciak"

Subject: Re: [BIOMCH-L] Event detection during treadmill walking
I am currently working on a technique to determine pushrim contact and
release during wheelchair propulsion using the angular velocity of the
hand and its proximity to a projected position in space. A similar
approach may work for your application. You may consider the
orientation of the foot and its speed throughout the step to approximate
initial and end contact. I do not have any references, since I have not
found any similar use of kinematic data. If you would like any further
information, please give me a call.

Good luck,

Andrew M. Kwarciak, M.S.
Biomedical Engineer
Human Performance and Movement Analysis Laboratory
Kessler Medical Rehabilitation Research & Education Corporation
1199 Pleasant Valley Way
West Orange, NJ 07052
Tel: 973-243-6903
Fax: 973-243-6984

From: "Wayne Board"
Subject: RE: Event detection during treadmill walking
Date: Wed, 5 Jul 2006 11:15:46 -0500
Dear Fariba Bahrami,

novel pressure/force systems can automatically calculate gait events, such as initial contact, end of contact, stride, step, and swing time. In addition, novel sensors can be used as sophisticated footswitches. On a treadmill, the pedar insole pressure/force measurement system from novel works great. The measurement insoles fit inside the subject’s footwear and the pressure, force, and area loaded can be measured specifically by the 99 sensors in each insole.

If you are interested in discussing this matter further, please do not hesitate to contact me.

The following are references where the pedar was used for gait timing events:
1: Foot Ankle Int. 1996 Apr;17(4):204-9.

Reliability of an in-shoe pressure measurement system during treadmill walking.

Kernozek TW, LaMott EE, Dancisak MJ.

Division of Kinesiology, University of Minnesota, Minneapolis 55455, USA.

We examined the reliability of in-shoe foot pressure measurement using the Pedar
in-shoe pressure measurement system for 25 participants walking at treadmill
speeds of 0.89, 1.12, and 1.34 meters/sec. The measurement system uses EMED
insoles, which consist of 99 capacitive sensors, sampled at 50 Hz. Data were
collected for 20 seconds at two separate times while participants walked at each
gait speed. Differences in some of the loading variables across speed relative
to the total foot and across the different anatomical regions were detected.
Different anatomical regions of the foot were loaded differently with variations
in walking speed. The results indicated the need to control speed when
evaluating loading parameters using in-shoe pressure measurement techniques.
Coefficients of reliability were calculated. Variables such as peak force for
the total foot required two steps to achieve a coefficient of reliability of
0.98. To achieve excellent reliability (> 0.90) in the peak force, force time
integral, peak pressure, and pressure time integral across the total foot and
the seven regions, a maximum of eight steps was needed.

Foot Ankle Int. 2000 Sep;21(9):749-52.
Related Articles, Links

Reliability and running speed effects of in-shoe loading measurements during slow treadmill running.

Kernozek TW, Zimmer KA.

Department of Physical Therapy, University of Wisconsin-La Crosse, Health Science Center, 54601, USA. kernozek.thom@uwlax.edu

In-shoe measurement systems allow the clinician and researcher to examine the loading parameters within the shoe. This study sought to investigate the test retest reliability and speed effects of in-shoe loading parameters using the Pedar System (Novel GMBH Munich) during slow treadmill running. The results indicated good to excellent test retest reliability between the two days test-Intraclass correlation coefficients (ICC's) ranged from 0.84-0.99 depending on the plantar region and variable analyzed. All plantar loading variables increased (peak pressure, peak pressure time impulse, peak force, and force time impulse) with the exception of contact area when treadmill running speed was increased from 2.24 m/s to 3.13 m/s. Results indicate that control of running speed is essential in obtaining reproducible data using this system to measure in-shoe loading data.

Clin Biomech (Bristol, Avon). 2000 Dec;15(10):781-5.
Related Articles, Links

Corrected and republished in:
Clin Biomech (Bristol, Avon). 2001 May;16(4):353-7.

A comparison of vertical force and temporal parameters produced by an in-shoe pressure measuring system and a force platform.

Barnett S, Cunningham JL, West S.

Faculty of Health and Social Care, University of the West of England, Room 1E13, PS5 Glenside Campus, Blackberry Hill, Stapleton,BS16 1DD, Bristol, UK. sue.barnett@uwe.ac.uk

OBJECTIVE: To investigate the ability of Pedar in-shoe system to measure vertical force accurately, by comparing it with the Kistler force platform. DESIGN: In vivo experiment in normal subjects. BACKGROUND: It has been suggested Pedar is highly reliable, but absolute accuracy of the system with regard to force measurement has not been comprehensively tested. METHODS: Sampling at 99 Hz, using five healthy subjects, simultaneous data were collected barefoot, and in three types of shoes (Trainers, Oxfords, Slip-on Deck Type). Six variables obtained from the force/time curve from each footstep were compared. RESULTS: The temporal data recorded by Pedar correlated well with that obtained using Kistler, with significant differences only in overall duration of the step in Deck shoes (P