Predicting Chances of Live Birth in IVF procedure based on Day 3 Embryo Analysis and Patient Characteristics

Abstract

The number of couples pursuing In-Vitro Fertilization (IVF) continues to grow worldwide, making infertility a global health problem. An approach to enhance traditional IVF is to find the success rate at the early phase of IVF considering both the patient characteristics and embryo formation as well. Taking this approach into account, we propose a paper that uses a novel method to determine the likelihood of live birth. This is the first research that we are aware of that predicts live birth outcome based on both intrinsic and extrinsic factors. The proposed method is to analyze the embryo image and the patient characteristics, two different algorithms have been developed to achieve this. Initially, by considering patient characteristics, a prediction is obtained using Multiple Linear Regression, and thereafter Hough Transform is applied over day 3 cleavage stage embryo to get prediction based on number of cells. At the end the results are averaged and a final outcome is presented. According to our test results, we found high prediction chances up to 42.89% and 32.27% for the patients with positive live birth and 12.27%, 23.34%, and 17.8% for the patients with negative live birth.