Evaluating the Impact of Scanning Factors on Ultrasound Imaging for Predicting Semen Quality in Boars

Early prediction of semen quality in young boars is crucial to reduce operational costs associated with low-productivity boars at boar studs, where high genetic merit is key. B-ultrasound imaging, a non-invasive method using echo signals to visualize tissue density, has been evaluated as a potential tool for this purpose. Stronger echoes indicate denser tissues, such as the seminiferous tubules responsible for sperm production. This study focuses on investigating the predictive power of using B-ultrasound imaging from boars around 6 to 9 months old to predict semen quality over the next six months in industrial settings. Several scanning factors were considered, including image brightness, imaging area of the testicle, and the imaging angle of the probe. The studied dataset included 1417 images and 3254 semen records from 107 boars. Results showed that the model's performance was significantly influenced by the imaging area, the angle of the testicle, and the pixel brightness of the image without being standardized. While the accuracy of B-ultrasound imaging is not yet sufficient to replace traditional assessments, this study highlights key features in testicular images that may significantly impact model predictions, providing practical guidance for leveraging B-ultrasound imaging in predicting semen quality in young boars.