Algorithms show worth in analysis of a horse’s stride

File image.

Two algorithms developed by researchers for use with inertial measurement units have shown promise in automatically detecting the timing of the hooves hitting the ground and lifting off during locomotion.

Marij Tijssen and her fellow researchers say that hoof-on and hoof-off events are fundamental locomotion characteristics of interest in gait analysis work.

Their work, reported in the open-access journal PLOS ONE, follows on from related work in which one of the same two algorithms proved accurate in identifying the crucial break-over phase in a horse’s stride.

The timing of hoof placement — that is, hoof-on and hoof-off events — can be examined visually, but this is limited by what the human eye is capable of detecting.

Instead, objective measurement tools such as force plates, optical motion capture systems and inertial measurement units are used.

In general, the force plate is considered the gold standard for kinetic gait analysis. However, data collection is time consuming and this method can be used only in a laboratory setting.

A generic illustration of the movement of the hoof (A), modified from a 2004 study by T.H. Witte and others, and the signals of the acceleration (B), angular velocity (C) and vertical force (D). The hoof-on and hoof-off events are depicted with the vertical dashed lines and the dots show the detected hoof-on and hoof-off events from the different signals. These hoof-events occur at the start and end of the stance phase, shown as the period not underlined by a dark beam. The horizontal dashed line in D shows the threshold used to detect the hoof-events from the force signal.

The researchers set out in their study to test the ability of an angular velocity algorithm and acceleration algorithm to detect the hoof-on and hoof-off events automatically and, if successful, to assess their accuracy.

Seven Warmblood horses were used in the study. Each was equipped with an inertial measurement unit on the sidewall of their right front and right rear hooves.

Horses were walked and trotted over a force plate for internal validation while, simultaneously, the three-dimensional position of three reflective markers also attached to the hoof area was measured by six infrared cameras as part of an optical motion capture system.

A total of 147 trials were analyzed.

For hoof-on detection, the results indicate that the agreement with the force plate was, in general, better for the angular velocity algorithm, with an accuracy between 2.39 and 12.22 milliseconds depending on the gait and hoof, and a precision of around 13.83ms.

For hoof-off detection, the results indicate that the agreement with the force plate was better for the acceleration algorithm, with an accuracy of 3.20ms and precision of 6.39ms.

“These algorithms seem promising for gait classification,” they concluded, “although a more extensive validation process should be performed.

“Also, the applicability of these algorithms should be investigated under different circumstances, such as different ground surfaces, gaits, speed and different hoof trimming conditions.”

Members of the study team are variously affiliated with Utrecht University in The Netherlands, the Swedish University of Agricultural Sciences. One is with Inertia Technology B.V.

Tijssen M, Hernlund E, Rhodin M, Bosch S, Voskamp JP, Nielen M, et al. (2020) Automatic hoof-on and -off detection in horses using hoof-mounted inertial measurement unit sensors. PLoS ONE 15(6): e0233266.

The study, published under a Creative Commons License, can be read here

Leave a Reply

Your email address will not be published. Required fields are marked *