Computerized detection of supporting forelimb lameness in the horse using an artificial neural network.

This research explores the use of technology, specifically artificial neural networks, in detecting lameness in horses, using data derived from automated infrared motion tracking.

Methodology

• The researchers used artificial neural networks, a type of machine learning, to process and analyze the data.
• The experiment monitored the movements of 175 horses as they trotted on a treadmill. The SELSPOT II system, a motion analysis technique, was used to record this motion.
• Infrared (IR) markers were placed on the horses’ heads and left forehooves, which were then tracked by two cameras.

Procedure and Analysis

• The recorded motion of the horses’ heads was transformed into a manageable computational form using a Fourier Transform, a mathematical technique that breaks down a complex signal into simple sinusoidal parts.
• A multilayer feed forward artificial neural network was used to further process the data, and this was trained to distinguish between healthy and abnormal gaits, as well as to quantify the degree of lameness.

Results

• The artificial neural network successfully identified abnormal gaits in 78.6% of cases, suggesting that this method can be effective in detecting supporting forelimb lameness in horses.
• However, the network gave contradictory results in 12% of cases, failed to reach a conclusion in 5.9%, and produced incorrect results in 3.5% of cases.
• While this system was not 100% accurate, the researchers concluded that properly trained artificial neural networks have the potential to provide a non-human diagnosis of equine lameness.

Implications

• This study paves the way for further research into the application of artificial intelligence in veterinary medicine.
• Such technology could dramatically reduce the time taken to diagnose conditions such as lameness in horses, and reduce reliance on human judgement, which can often be subjective.