Journal Watch: An Overview of the Latest Research Using Inertial Sensors in Equine Lameness Measurements

By Kevin Keegan Kevin G. Keegan, DVM, MS, DACVS | Laurie Tyrrell-Schroeder Laurie Tyrrell-Schroeder, DVM | Updated on | Feature, Journal Watch, KG Keegan, LT Schroeder, Research

Journal Watch: An Overview of the Latest Research Using Inertial Sensors in Equine Lameness Measurements


Journal Reviewed: Effects of sensor position on kinematic data obtained with an inertial sensor system during gait analysis of trotting horses.

Moorman VJ, Frisbie DD, Kawcak CE, McIlwraith CW.

Journal of American Veterinary Medical Association | 2017 Mar 1;250(5):548-553z

This study assessed the effects of altering the right forelimb gyroscope and pelvic accelerometer from their midline position on forelimb and hind limb lameness results.  The right forelimb sensor was placed 2cm medial and lateral to the dorsal midline surface of the pastern. The pelvic sensor was placed 2cm cranial to, caudal to, and left and right of the highest midline position of the tuber sacrale. The authors found that alteration of the right forelimb gyro did not produce any change in fore or hind limb measures. Moving the pelvic sensor to the left or right of midline altered results of Diff Min pelvis (the impact component of the stance phase) by approximately 3.0 to 3.5 millimeters. No effect on Diff Max Pelvis, the push off component, was noted. These results are consistent with studies conducted at the University of Missouri, whereby a deviation off of midline of more than half the width of the sensor (~1.5cm) produced alterations in Diff Min pelvis, but below +/-3mm.  

Potential Consequences for Q Users

It should be noted that change in Diff Min Pelvis of about 3.5 mm is very near the 95% confidence intervals for significant change between two trials (+/- 3 mm), thus changes less than 3 mm could be within the probability of random change. Nevertheless, the user should be aware of the expected consequences of pelvic sensor malpositioning and take care to assure proper midline placement.

  • Positioning the pelvic sensor 2cm off midline may induce an artifactual impact type lameness measurement on the side of sensor location.
  • A pre-existing impact type lameness on the side away from sensor location may be masked.
  • A pre-existing impact type lameness on the side toward the sensor location may be exacerbated.

Creation of artefactual push off type lameness should not be expected by sensor misplacement, nor should inaccurate measurement of pre-existing push off type lameness be expected by sensor misplacement.

BE AWARE! To avoid sensor malpositioning, make sure the horses is standing squarely when palpating for the tuber sacrale and securing the pelvic sensor. Check midline position from behind the horse prior to data collection. The sensors are 3 cm wide, so malpositioning them 2 cm (75% of their width), although possible, is unlikely when paying attention to proper positioning.  Also, a sensor malpositioned by 2 cm should be readily identified when evaluating the horse. 


In More Detail...

Understanding the physical principles of obtaining uni-axial acceleration measurements and the mechanics of pelvic vertical movement and rotation will provide insight to the accuracy of placement and sensitivity of misalignment.

So, why doesn’t the right forelimb sensor affect results? 

The right forelimb sensor is a gyroscope that measures rotation in the sagittal plane, if placed on the dorsum of the right forelimb pastern. It is used as an event marker in the determination of stance and has no effect on the measurement of amplitude of asymmetry.   A clockwise rotation of the sensor on the dorsum of the distal limb (when viewed from a position to the right of the horse) will cause a clear positive signal.  This happens with the horse is moving forward and the right foot is on the ground.   A counterclockwise rotation (when viewed from a position to the right of the horse) will cause a clear negative signal.  This happens when the horse is swinging the limb forward.  If placed exactly on the dorsum of the right distal forelimb the gyroscopic sensor rotates with the limb and the sagittal rotation signal is strong, with high fidelity.  Move the sensor off midline and the sensor rotation in the sagittal plane decreases and the positivity and negativity of the signal.  The signal has the same shape, it is only smaller in amplitude (see figure below). 

The Equinosis® Q uses a peak detection algorithm to detect right forelimb position.  As long as there are distinct peaks in the gyro signal right forelimb position will be faithfully recorded.  It is recommended to avoid placement of the gyro sensor more than 45 degrees off of dorsal midline to avoid potential issues. However, unless the sensor rotates to the direct lateral or medial aspect of the limb, the peak detection algorithm still works. 

Placing the sensor on the direct medial or the direct lateral aspect of the pastern will cause odd results or no analysis. Rotation of the sensor to the palmar pastern will convert positive rotation signals to negative rotation signals and exactly opposite in side but exactly equivalent in amplitude of lameness will be measured.

Why is Diff Min pelvis affected and not Diff Max?

The pelvic sensor is a uni-axial accelerometer that measures vertical acceleration as the acceleration normal (perpendicular) to the top surface of the accelerometer device.  Moving the sensor up and down while it is held with its top surface parallel to the ground will create the vertical acceleration signal with greatest amplitude.  Holding the sensor slightly tilted but moving it up and down the same amount (as before you tilted the sensor) will give a vertical acceleration signal that is the same shape but of smaller amplitude.

The same would happen if the sensor was put on the horse’s pelvis in the correct location but tilted. This can happen in horses with asymmetric tuber sacrale; however, unless the asymmetry is severe (i.e. approximately 45 degrees or greater) this should not significantly affect accurate measurement of Diff Min or Diff Max pelvis. Because Lameness Locator® measures the difference in upward and downward movement, and standardizes to the expected amount of vertical movement, no significant change in the lameness measures should be seen.

However, when putting the sensor off midline, slight artifacts, but only in Diff Min pelvis, should be expected.  The pelvis rotates when the horse is trotting.  The pelvis is parallel to the ground, and at a maximum height from the ground, when the horse is “airborne”, or when one hind limb has just pushed off and the opposite hind limb is about to land.  The pelvis moves down and rotates during weight bearing, with the rotation away from the weight bearing limb and toward the hind limb swinging forward.  So, there is a counterclockwise rotation of the pelvis (when viewed from behind the horse) when the right hind limb is weight bearing, and a clockwise rotation of the pelvis when the left hind limb is weight bearing.

It is this rotation, which occurs when the entire pelvis is in a minimum position that causes the artifactual Diff Min pelvis results when the pelvic sensor is malpositioned.

To illustrate this, consider the vertical acceleration signals of a pelvic sensor placed on midline correctly, and one placed off to the right. 

When the horse is airborne and the pelvis is parallel to the ground, the correctly positioned and malpositioned sensor both remain parallel to the ground, i.e. not rotated.  So, the upward vertical acceleration is faithfully measured and Diff Max measurements, even from the malpositioned sensor, are correct.

However, when the horse is weight bearing, the pelvis is tilted away from the weight bearing limb. The effect of the tilt on a malpositioned sensor is greater than on the correctly positioned sensor.  When the sensor is placed away from the weight bearing limb, the downward acceleration is amplified in the weight bearing limb.  This is translated into a greater downward movement of the entire pelvis (i.e. the pelvis falls more or to a “lower” position), during weight bearing of the limb opposite to the direction of pelvic sensor malposition.  When the hind limb on the side of malposition is weight bearing, and the pelvis is rotated away from the weight-bearing limb, the rotation of the malpositioned pelvic sensor is not that much different (and maybe even less) than the correctly positioned pelvic sensor.  Thus, the change in downward vertical acceleration is not amplified, so it is less than during weight bearing of the opposite hind limb.  This is translated into a smaller downward movement of the pelvis (i.e. the pelvis falls less or to a “higher” position) during weight bearing on the same side of the incorrectly positioned pelvic sensor.   This will make Diff Min pelvis more positive if sensor misplaced to the right (creating a false RH impact lameness) or more negative if misplaced to the left (creating a false LH impact lameness).  

image credit:

image credit: 

The end result >>> With an existing impact lameness on one side, moving the sensor toward the opposite side will exacerbate the true lameness.  With an existing impact on one side, moving the sensor over to the other side will mask the lameness.  The authors of the paper only found a significant change in Diff Min pelvis when the sensor was positioned away from the lame limb, which would cause a masking of hind limb impact lameness, but this is simply a factor of case use, as the amplitude of change was almost identical when the sensor was misplaced towards or away from the lame hind limb.

The authors also found that placement of the sensor 2 cm cranial to be statistically significant. However, the mean difference between midline in the correct position on the tuber sacrale and 2 cm cranial was approximately 2 mm, which is within the 95% confidence intervals for trial to trial repeatability (or random change). So, although statistically significant, this difference would be difficult to perceive visually and has little practical significance. The small difference could be the result of the sensor tilting forward, creating a less sensitive sensor. However, as explained previously, if the sensor is on midline, the difference in pelvic positions between the two halves of stride should not be affected. Nonetheless, placement of the pelvic sensor cranial or caudal to the tuber sacrale it is not recommended.

Newer Post

Leave a comment