Measurement and extraction of base-of-support gait parameter using a novel accurate human locomotiontracking system via UWB radios

MEASUREMENT AND EXTRACTION OF BASE-OF-SUPPORT GAIT PARAMETER USING A NOVEL ACCURATE HUMAN LOCOMOTION TRACKING SYSTEM VIA UWB RADIOS M. Abou El-Nasr and Heba Shaban

R.M. Buehrer

Arab Academy for Science, Technology & Maritime Transport (AASTMT), P.O.Box 1029, Alexandria, Egypt.

Wireless @ Virginia Tech, Virginia Polytechnic Institute and State University, P.O.Box VA 24061, Blacksburg, USA.

ABSTRACT In this paper, we use our recently proposed highly accurate gait analysis system presented in [1] to measure the base-of-support (BOS) gait parameter, which is defined as the distance from heelto-heel while walking. This particular gait parameter is known to be of clinical importance, and its measurement accuracy reported in the literature using current highly accurate optical tracking systems is not sufficiently accurate to be clinically accepted. We further develop a simulation environment using MATLAB to extract gait parameters from the raw-marker-data measured using sophisticated optical tracking systems, which commonly require dedicated and highly sophisticated software programs. Then, we compare the accuracy of the measured BOS using our proposed system to the extracted BOS from optical tracking systems. We show that our proposed system outperforms the corresponding highly accurate optical tracking systems. In particular, we show that our system provides an accuracy of 1.2% for the BOS measurement compared to 14.6% accuracy for current optical tracking systems.

on limb segments [4], [5], [6]. These systems allow for the assessment of a complete three-dimensional kinematic analysis of human movement [5], [7]. For the base-of-support (BOS), the reported measurement accuracy is not sufficiently good to be clinically accepted. In this paper, we use our recently proposed ambulatory gait analysis system [1] to measure the BOS gait parameter. Furthermore, we develop a simulation environment using MATLAB [8] to extract the gait parameters measured using optical tracking systems, which commonly require dedicated software programs placed on servers associated with these measurement systems. We further compare the accuracy of the measured BOS using our proposed system to optical tracking systems. The organization of this paper is as follows. Section 2 gives an overview of the BOS gait parameter measurement accuracy, provides common values for normal gait, and gives a brief overview of our recently proposed gait analysis system. Then, Section 3 introduces the BOS measurement procedure, and provides results using our proposed system. Section 4 summarizes the proposed extraction procedure of optical system gait parameters using MATLAB. Finally, conclusions are provided in Section 5.

1. INTRODUCTION Gait analysis is the systematic study of human walking. Observational gait analysis, the standard method of evaluating gait, refers to the visual assessment of a patient’s gait. Gait analysis by observer assessment does not use any specialized equipment, and is simply used to observe abnormalities in gait. Clinical gait analysis, also termed as quantitative gait analysis, provides a detailed clinical introduction to understanding and treating walking disorders [2], [3]. The identification of gait disorders is commonly assessed by the measurement of the spatial and temporal parameters of gait1 . However, it is worth noting that the techniques and technologies that work well for measuring normal gait often fail when applied to abnormal gait [4], [5]. Standard quantitative gait analysis is based on either optical, magnetic, or ultrasonic motion tracking systems. Sophisticated measurement systems employ optical tracking techniques to track the displacement of markers2 placed at particular anatomical sites

2. OVERVIEW OF BOS MEASUREMENT AND PROPOSED SYSTEM In this section, we give an overview of common BOS values reported in the literature for normal gait. Then, we give a brief overview of our recently proposed ambulatory gait analysis system. 2.1. Common BOS Values for Normal Gait The BOS is defined as the distance from heel-to-heel while walking; as shown in Figure 1 [9]. It is known to have clinical importance. The BOS is typically equal to 8.5 cm for normal adults with a reported error of 1.17 cm with a relative error ≈ 14.6 % [2], [10]. Thus, the accuracy and reliability of the BOS measurement need to be addressed, as it is one of the important parameters to clinicians. A summary of BOS values in addition to other spatiotemporal gait parameters are provided in Table 1 for normal gait.

978-1-4244-7000-6/11/$26.00 ©2011 IEEE 1 Spatiotemporal parameters of human gait include the step-length, stride-length, velocity, etc. 2 Markers are arranged based on well-defined marker-sets. Markers should be placed at specific anatomical positions, where the objective of data collection is to capture the movement of the underlying skeleton.

188

2.2. Brief Overview of Proposed System Recently, we proposed a wearable human locomotion tracking and gait analysis system using ultra wideband (UWB) radios [1], [11]. In this paper, we use our system to measure and extract the BOS

3. MEASUREMENT PROCEDURE

Figure 1: A diagram representing the BOS for the right foot (A) and the stride-length for the left foot [2].

Table 1: Gait parameters for normal subjects [2]. Gait parameter Young subjects Older subjects Step-length, R(cm) 77 cm 65 cm Step-length, L(cm) 77 cm 63.6 cm Base-of-support, R(cm) 8.5 cm 8.5 cm Base-of-support, L(cm) 8.1 cm 8.5 cm

distance. Our system is based on UWB transceivers attached to the test subject’s body, or sewn into clothing specifically designed for this application. Thus, it does not require a complex setup unlike optical tracking systems. Our proposed system acquires the distances between the different points on the body during movement. A simplified diagrammatic representation of the proposed system’s data acquisition approach is shown in Figure 2. Our system is designed based on a target ranging accuracy equal to 1 mm. This value was particularly chosen for achieving a ranging accuracy that is ten-times better than the inter-marker distance accuracy reported in the literature for current systems. Thus, the measurement error for the BOS distance using our system is ≈ 1.2 %, which makes it an attractive candidate for the application in clinical gait analysis.

In order to evaluate our proposed system’s performance based on actual-data, on-body UWB measurements were taken at the MPRG3 labs. Two UWB transmit and receive antennas were attached to the heels of the test subject in order to estimate the inter-spacing distance based on the time-of-arrival (TOA) of the received pulses. The following equipments were used: HP33120A function generator, Tektronix CSA8000B Digital Sampling Oscilloscope, Geozondas pulser (GZ1106DL1, GZ1117DN25), and two antennas manufactured by the Time Domain corporation [12], as shown in Figure 3(a). The test subject was allowed to walk, and the received pulses were recorded and stored. The length of the pulses was 4000 samples. In this measurement set, we measured the BOS distance for normal gait. The actual measurement setup is depicted in Figure 3(b). An example received pulse normalized to unit energy is depicted in Figure 4.

(a)

(b)

Figure 3: (a) UWB antennas manufactured by the Time Domain corporation. and (b) BOS measurement setup.

0.025 0.02 0.015

Amplitude(V)

0.01 0.005 0 −0.005 −0.01 −0.015 −0.02 0.98

Figure 2: Simplified diagrammatic representation of the on-body intersegmental measurements using UWB radios, for the initial and during movement measurements [1].

1

1.02

1.04 Time (sec)

1.06

1.08

1.1 −7

x 10

Figure 4: Normalized received pulse for the BOS measurement set. 3 Mobile and Portable Radio Research Group, a part of the Wireless @ Virginia Tech research group.

189

In order to estimate the TOA and the corresponding distances, the measured data was further processed to estimate the TOA and the resulting distance every 25 ms. The extracted BOS from the UWB measured data is shown in Figure 5.

shown in Figure 8 for an arbitrary frame of a normal walking MoCap file. This file was then processed using MATLAB to extract the raw-marker data, and estimate the BOS distance. The extracted BOS distance is shown in Figure 9.

12 Measured distance Average value

Base−of−support distance (cm)

11 10 9 8 7 6 5 4 3

0

5

10 Time (s)

15

Figure 6: Basic configuration of a gait analysis laboratory [9].

Figure 5: Person ”A” BOS distance for normal gait measured using the proposed UWB system.

4. OPTICAL TRACKING SYSTEM GAIT PARAMETER EXTRACTION Optical tracking systems are based on the use of charge-coupled device CCD-cameras and a set of markers attached to the subject’s body, where marker positions are estimated via the triangulation of the position and orientation of two or more cameras [4], [7], [13]. A simplified description of a gait laboratory equipped with an optical tracking system is shown in Figure 6. Optical systems are accurate and use high sampling rates, which enable the acquisition of real-time data. The main disadvantage of these systems is that they require dedicated laboratories, complex settings, and highly skilled operators. In addition, they have the line-of-sight (LOS) restriction, where if markers are not detected by at least two-cameras their positions are not recorded [4], [13]. If two or more cameras detect a marker and the positions and orientations of these cameras are known, then it is possible to detect the three-dimensional position of the marker [4], [13]. The three-dimensional (3D) coordinates, also termed raw-marker data, are then stored in special file formats called motion-capture (MoCap) data files. In order to examine and compare the measurement accuracy of our proposed system to commercial optical tracking systems, we developed a set of MATLAB codes for the entry, extraction, and estimation of gait parameters. Figure 7 shows a simplified diagrammatic chart of the data processing procedure. First, a motion capture data file4 representing normal walking movement was obtained from [15]. The raw-marker data was plotted after extraction using MATLAB, as 4 Data obtained from an open-source database for raw motion capture marker data measured with a commercial optical tracking system; Vicon motion capture system consisting of 12 infrared MX-40 cameras [14].

190

Figure 7: A diagrammatic chart of the gait parameter extraction procedure. As can be seen in Figures 5 and 9, the BOS measured with our system and optical tracking system, follow the same pattern, which represents the typical pattern of the BOS distance for normal walking. It is worth noting that the two curves are not expected to be identical since they are for two different persons. According to the plots, the average value of the BOS measured with the optical tracking system is 9.1 cm. The corresponding value for the BOS measured with our proposed system is 8.5 cm, which agrees with the value reported in the literature for normal adults and summarized in Table 1. 5. CONCLUSIONS In this paper, we used our recently proposed highly accurate wireless wearable UWB-based gait analysis system to measure the BOS gait parameter. Then, we developed a simulation environment using MATLAB for the extraction and estimation of gait parameters of raw-marker data measured using commercial optical tracking systems. We showed that the gait parameter measurement using our system provided accurate results for the BOS distance. Measured data for normal gait showed that the average BOS measured

14 Actual distance Mean value 2000

40

39

38

1000

30 16 21

500

0 1500

Base−of−support Distance (cm)

1500 Z [mm]

12

Markers

18 19

37 24

12 6 11 41 10 2 9 13 1 5 8 4 3 7 15 14 20 17 322536 3529 2827 31 34262322 33

10

8

6

1000

1000 500 500 0

Y [mm]

4

0 −500

X [mm]

Figure 8: Three-dimensional coordinates (raw-marker) data extracted from a MoCap file for normal gait.

using our system agrees with the value reported in the literature for the BOS of normal gait. Typically, our proposed approach results in a 1.2% accuracy for the BOS measurement compared to a 14.6% accuracy for current optical tracking systems.

Acknowledgment The authors would like to thank Haris Volos, PhD candidate at the Wireless @ Virginia Tech research group, for taking the UWB measurements. 6. REFERENCES [1] H. Shaban, M. Abou El-Nasr, and R. M. Buehrer, “Toward a highly accurate ambulatory system for clinical gait analysis via UWB radios,” IEEE Transactions on Information Technology in Biomedicine, vol. 14, no. 2, pp. 284–291, Mar. 2010. [2] A. Tiedemann, M. Kwan, S. Lord, H. Menz, and M. Latt, “Reliability of the gaitrite walkway system for the quantification of temporo-spatial parameters of gait in young and older people,” Gait and Posture, vol. 20, no. 1, pp. 20–25, Aug. 2004. [3] E. Rosen, J. Gross, and J. Fetto, Ed., Musculoskeletal Examination, 2nd Ed. USA: Blackwell Science, Inc., 2002. [4] A. Leardini L. Chiari A. Cappozzo, and U. Della Croce, “Human movement analysis using stereophotogrammetry part 1: theoretical background,” Gait and Posture, vol. 21, no. 2, pp. 186–196, Feb. 2005. [5] Z. Knoll R. Kiss, and L. Kocsis, “Joint kinematics next term and spatialtemporal parameters of gait measured by an ultrasound-based system,” Gait and Posture, vol. 26, no. 7, pp. 611–620, Sept. 2004. [6] E. Pavan, C. Frigo, D. Bettinelli, M. Rabuffetti, P. Crenna A. Leardini, A. Ferrari, and M. Grazia Benedetti, “Quanti-

2

0

1

2

3 Time (s)

4

5

Figure 9: Person ”B” BOS distance for normal gait measured using an optical tracking system.

tative comparison of five current protocols in gait analysis,” Gait and Posture, vol. 28, no. 2, pp. 207–216, Aug. 2008. [7] J. A. Corrales, F. A. Candelas, and F. Torres, “Hybrid tracking of human operators using IMU/UWB data fusion by a kalman filter,” in HRI ’08: Proceedings of the 3rd ACM/IEEE international conference on Human robot interaction. New York, NY, USA: ACM, 2008, pp. 193–200. [8] Mathworks, “MATLAB and simulink for technical computing,” 2010. [9] H. A. Shaban, “A novel highly accurate wireless wearable human locomotion tracking and gait analysis system via UWB radios,” Ph.D. Dissertation, Virginia Tech, Blacksburg, VA, Apr. 2010. [10] H. Hillstorm S. Barker, and W. Freedman, “A novel method of producing a repetitive dynamic signal to examine reliability and validity of gait analysis systems,” Gait and Posture, vol. 24, no. 4, pp. 448–452, Dec. 2006. [11] H. Shaban, M. Abou El-Nasr, and R. M. Buehrer, “Performance of ultralow-power IR-UWB correlator receivers for highly accurate wearable human locomotion tracking and gait analysis systems,” IEEE Global Telecommunications Conference, GLOBECOM ’09., pp. 1–6, 30 Nov. - 4 Dec. 2009. [12] T. Corporation, “Markets planar elliptical dipoles under the brand name ”broadspec”,” 2010. [13] A. Leardini L. Chiari A. Cappozzo, and U. Della Croce, “Human movement analysis using stereophotogrammetry part 1: instrumental errors,” Gait and Posture, vol. 21, no. 2, pp. 197–211, Feb. 2005. [14] “Vicon motion systems preparation,” [Available Online] http://www.udel.edu/PT/Research/MAL/, 2009. [15] “CMU graphics lab motion capture database,” [available online]: http://mocap.cs.cmu.edu, 2009.

191