147 SOFTWARE FOR MORPHOLOGICAL CLASSIFICATION OF CATTLE BLASTOCYSTS
F. D. Matos A , J. C. Rocha A and M. F. G. Nogueira ASão Paulo State University – UNESP, Assis, SP, Brazil
Reproduction, Fertility and Development 27(1) 165-165 https://doi.org/10.1071/RDv27n1Ab147
Published: 4 December 2014
Abstract
Embryonic morphological classification has great importance for numerous laboratory techniques (from basic to applied research in assisted reproduction). However, the method used to perform the classification of embryos in varying degrees of quality has always been based on the subjectivity of the evaluator. Although quality standards and descriptions of morphological characteristics that categorize an embryo in each grade are established, currently there is not an accurate method that can generate consistent and reliable results. Thus, our work resulted in the development of software able to perform the classification of morphological quality of bovine blastocysts. Artificial intelligence techniques (such as artificial neural networks) were used in the development. Results indicate an overall accuracy of 79.2% in the classification of bovine blastocysts in 3 degrees of quality. For blastocysts classified as excellent or good (class 1), the hit rate is 82.6%; for blastocysts classified as regular (class 2), the hit rate is 16.7%; and for blastocysts classified as poor (class 3), the hit rate is 91.7%.
