Automated extraction of mathematical variables representing the morphology of in vitro produced equine embryos

• Grant number: 24/23599-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2025
• End date: December 31, 2025
• Field of knowledge: Agronomical Sciences - Veterinary Medicine - Animal Reproduction
• Principal Investigator: Marcelo Fábio Gouveia Nogueira
• Grantee: Malu Maria Felix Manzieri
• Host Institution: Faculdade de Ciências e Letras (FCL-ASSIS). Universidade Estadual Paulista (UNESP). Campus de Assis. Assis , SP, Brazil
• Associated research grant: 23/16156-1 - Prediction of gestational success, using the variables of morphology and morphokinetics of the human embryo and the patient, using the Artificial Intelligence technique, through deep learning and multilayer perceptron., AP.R

Abstract

Embryo quality plays an important role in selecting the embryo with the greatest potential for implantation and, consequently, for establishing pregnancy. In order to select the best embryos, embryologists have classification systems based on morphological aspects and visual analysis. However, the methodology used to categorize different degrees of quality is prone to variability and subjectivity between and within evaluators - there is currently no objective, reproducible method that generates consistent evaluations. Thus, the objective of this study is to develop a digital image processing method in order to obtain mathematical variables that represent the embryo morphology of equine embryos produced in vitro (originating from intracytoplasmic sperm injection; ICSI). Methodologies and algorithms that have already been described for the processing of digital images of bovine (produced in vitro) and human (fertilized in vitro) embryos will be applied. After analyzing which algorithms performed best, additional algorithms that better explore the peculiarities of the equine blastocyst - which differs clearly from bovine and human blastocysts - will be explored (if necessary). By defining the set of algorithms that best process the images and extract the numerical variables, an automated model will be characterized for all digital processing steps (i.e., standardization and segmentation of blastocyst images). If possible (i.e., if there is a bank of around 400 images of equine blastocysts, provided by the company In Vitro Equinos Ltda.) and after characterizing the automated extraction of the mathematical variables, the effectiveness of structuring a software based on artificial intelligence for the classificatory prediction of embryo morphology will be explored.