An Accurate Image Analysis Method for Estimating Body Measurements in Horses.

This study proposes a standardized image analysis method for assessing horse conformation. A total of 34 adult American Quarter Horses (mean age = 6.7 ± 2.9 years, body weight = 489 ± 44 kg) were used. Reference angular measurements were performed using a digital goniometer, and reference linear measurements were determined using a standard measuring tape. For image analysis, photographs from the animal's left profile were taken at a distance of 3 m using eight markers to facilitate visualization. Images were processed using ImageJ software. Reference and estimated values showed great accuracy, with v < 1%, Pearson's correlation coefficient > 0.999 (P < .001), and coefficient of variation < 2.70. Image analysis and reference results were similar (P > .05) for most variables. The highest coefficient of variation (3.05%) was observed for coxae-femur angle, where the software obtained an underestimation of 2.8° compared with the reference method. Body length resulted in an overestimation of 3.4 cm in relation to reference values, but coefficient of variation was low (1.54%). Discrepancies between estimated and reference results can be attributed to possible errors when collecting biometric data using portable measuring tools, angles and long measurements being more difficult to obtain. Software analysis of standardized horse photographs is an accurate and precise method for obtaining body measurements and is therefore recommended for future studies on horse conformation.