Ultrasound-Based Kinematic Analysis of Human Ambulation

Kremenic, IJ, Medvecky, MJ, Levy, RC*, Hager, AA*, Cheikin, MI*, Ben-Avi, S*, Gleim, GW

NISMAT, Lenox Hill Hospital, New York, NY; *The Cooper Union School of Engineering, New York, NY

Motion analysis systems typically monitor visual or infra-red light. In order to make three dimensional measures at least two, precisely positioned cameras are necessary. Cost and space requirements for such systems are significant enough to preclude their routine clinical use. Motion detection devices which monitor ultrasound can measure three dimensional positions at a fraction of the cost and in limited space. We describe the application of such a system to the measurement of slow speed ambulation in humans.

Materials and Methods

The motion analysis system is shown as a block diagram in Figure 1, and is based around a tracking system called V-Scope. The system consists of buttons having a mass of 14 grams which we attached to the subject, three towers which track the buttons, and a controller which gathers data from the towers and relays it to a PC via a serial port. Analysis of the data is performed off-line, after data collection. We mounted the entire system on a cart, so it was easily moved and positioned within 2 meters of the exercise device.

Figure 1: Motion analysis system.

Each tower sent out a pulse of infra-red, including an address indicating which button should receive this signal. Having received this signal, the particular button responded with an ultrasonic "chirp". Each tower determined the distance to the button by time elapsed from the infra-red pulse, since the speed of sound in air is known. The actual position of each button was calculated by triangulation, obtaining a 3-dimensional (x, y, z) coordinate for each button. The system corrected for temperature variations of the speed of sound. The buttons were polled serially every 25 ms, meaning that the relative positions of the buttons were known at slightly different times, thus making the system insensitive to high speed oscillations. Because the buttons were signaled using infra-red, line of sight was maintained between them and the towers.

Analysis was done off-line using a Fourier-based custom program to determine a fundamental period for one session of walking, smoothing the data and allowing for all gait cycles from that session of walking to be averaged, giving one (mean) cycle to represent the entire session. This technique limited noise and facilitated comparisons between any two sessions of gait. The program calculated the knee angles throughout the cycle in the sagittal and coronal planes, as well as 66% and 95% confidence intervals for these angles. The program allowed comparison between two sets of gait data as shown in Figure 2 (the first gene-ration of this software assumed the gait cycle to start at maximum knee angle).

In order to assess reliability of the measure, 10 healthy adult volunteers walked on a treadmill, with their gait monitored for two minutes. Buttons were placed on the greater trochanter, fibular head, and lateral maleolus. The subjects were instructed to walk with their hands resting lightly on the front of the treadmill. Each volunteer walked two times (on different days), at two different speeds each day (1.5, 2.0 mph), in order to evaluate the reproducibility of knee angles calculated using the V-Scope based motion analysis system.

Results

Figure 2 demonstrates the range of knee motion in the sagittal plane for two tests in one subject. The mean value of the second test is compared to the mean and 95% confidence interval of the first test.

FIGURE 2

The maximum (FLEX) and minimum )(EXT) knee flexion angles are averaged for the 10 subjects in the following tables for test 1 (T1) and test 2 (T2) and shown for speeds of 1.5 and 2.0 miles/hour.

     MAXIMUM KNEE FLEXION ANGLE (DEGREES)
       1.5 FLEX T1     1.5 FLEX T2     2.0 FLEX T1     2.0 FLEX T2
R      53.6 +/- 2.2    52.2 +/- 2.1    56.7 +/- 2.2    54.4 +/- 1.7
L      55.3 +/- 1.9    55.3 +/- 1.9    57.7 +/- 1.6    58.1 +/- 1.6
     MINIMUM KNEE FLEXION ANGLE (DEGREES)
        1.5 EXT T1     1.5 EXT T2     2.0 EXT T1    2.0 EXT T2
R       4.4 +/- 1.0    3.6 +/- 0.9    4.2 +/- 0.8   3.3 +/- 0.8
L       3.1 +/- 0.6    3.3 +/- 0.6    3.4 +/- 0.7   3.3 +/- 0.6

There were no statistically significant differences between T1 and T2 results (P> 0.2 in all cases). Significant increases in max knee flexion ROM were noted with increasing speed (P<.01 for both tests). While showing good reliability between tests, the device is sensitive enough to pick up small increases in knee flexion angle with increasing speed.

Discussion

As with all measurement systems, sources of error must be identified. The ultrasound based system used is potentially subject to intra-subject variability, intra-test variability, temporal random variability, marker dropout and insufficient sampling frequency. At the speeds tested, the system demonstrated good reliability since no statistically significant differences existed between tests.

The statistically significant increase in maximum knee flexion angle with increasing speed has also been noted by others (1). With increasing stride length, greater knee flexion is necessary to clear the increasingly plantar flexed position occurring during pre-swing (1).

Consequently, this study was an initial demonstration of the reliability of an ultrasound based kinematic analysis of human motion. The ability of the system to detect pathological gait is yet to be determined as are the limitations of sampling rate at greater stride frequencies.

References

1. Brinkman JR, Perry, J. Rate and range of knee motion during ambulation in healthy and arthritic subjects. Physical Therapy 65(7): 1055-60, 1985.

Combined Orthopedic Research Societies Meeting, November, 1995.