An artificial intelligence tool for predicting embryos quality

One of the most relevant aspects in Assisted Reproductive Technologies is the characterization of the embryos to transfer in a patient. Objective assessment of embryo quality is actually an important matter of investigation both for bioethical and economical reasons. In most cases, embryologists evaluate embryos by visual examination and their evaluation is totally subjective. Recently, due to the rapid growth in our capacity to extract texture descriptors from a given image, a growing interest has been shown on the study of artificial intelligence methods to improve success rates of IVF programs based on the analysis and selection of images of embryos or oocytes. In this work we concentrate our efforts on the automatic classification of the quality of an embryo starting from the analysis of its image. The artificial intelligence system proposed in this work is based on textural descriptors (i.e. features), used to characterize the embryos, by measuring the homogeneity of their texture and the presence of recurrent patterns. A general purpose classifier is trained using visual descriptors to score the embryo images. The proposed system is tested on a datasets of 257 images with valuable classification results.