COMPUTER VISION TECHNIQUES FOR THE CHARACTERIZATION OF FUNCTIONAL CORPUS LUTEUM DIAGNOSIS BY MEANS OF COLOR DOPPLER ULTRASONOGRAPHY IN BEEF CATTLE

• Grant number: 25/19174-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2026
• End date: January 31, 2027
• Field of knowledge: Agronomical Sciences - Veterinary Medicine
• Principal Investigator: Ricardo Vieira Ventura
• Grantee: Talissa da Silva Vieira
• Host Institution: Faculdade de Medicina Veterinária e Zootecnia (FMVZ). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

Color Doppler ultrasonography, a 2D imaging diagnostic tool that converts regional blood flow into a spectrum of pixelated colors, has been the standard method for qualifying blood perfusion in the corpus luteum of bovine females. However, due to the subjective analysis that depends on the operator's experience, there is a need to more accurately quantify the variance in pixel colors and their subsequent association with biological phenomena related to animal reproduction. Possible changes in the colors concerning the red and blue tones present in the analyzed structure may elucidate parameters related to the biological functioning of investigated structures within the uterine environment.Computer Vision (CV) is a field of artificial intelligence (AI) that allows for the capture of patterns from data such as images and videos, mimicking human visual perception. Such techniques can support the development of a preliminary computational tool for the classification of biological events associated with pregnancy diagnosis. The use of these technologies can also be fundamental to understanding the ideal moment in which the corpus luteum has an adequate blood supply, a decisive factor for the success of reproductive procedures such as artificial insemination and embryo transfer.Therefore, this project aims to apply basic computer vision techniques to investigate parameters related to the blood perfusion of the corpus luteum in Nellore bovine females (obtained via Color Doppler Ultrasound images), and, consequently, to diagnose the functionality of this structure through the analysis of colored frames extracted from ultrasonographic videos. This research will contribute not only to the introduction of predictive technologies capable of identifying blood perfusion patterns in corpora lutea suitable for the establishment and maintenance of gestation, but also to providing valuable parametric information, such as the average of colored pixels present in the ovaries and corpora lutea, establishing classificatory criteria. (AU)