Executive Summary : | Infertility is a global health problem and one in every eight couples is suffering from this. In India, around 45-60% of married couples are having some form of subfertility/infertility problem and opting for assisted reproductive techniques (ART) to have a baby. The success of ART depends on the selection of good quality, viable embryo out of the multiple cultivated embryo for transfer. Selection is based on morpho-kinetic (shape and organization of the cells as well as rate and pattern of the cell division) parameters, which at present is done by manual inspection, taking out the embryo at a particular time period from the incubator and observing under a microscope. In this process, the embryos are exposed to the outside environment (temperature, humidity, volatile organic compound, pH etc.) which affects the embryo quality, as well prone to subjective bias. Therefore, minimal human intervention is crucial while continuous distant monitoring of embryos was taken up. Hence a system that allows continuous objective observation of the embryos in a safe environment like an incubator is sought. There has been work leveraging TLS camera system that allows near-continuous images of the embryos reducing environmental exposure as well as some working on guided assistance for selection through basic computer vision algorithms. However, these systems are in addition to the existing workflow as well rudimentary looking at the computer vision/ machine learning advances like AutoML/ TensorFlow advances. On top of it, being imported equipment, these are like white elephants for a cost-conscious Indian market. Taking an example this additional equipment costs more than the entire IVF infrastructure of many labs, making them unavailable and inaccessible. So, the investigators propose an IoT-enabled remote monitoring device that will not only be affordable, accessible to people but performance-wise at par with the recent machine learning architectures. Exploiting the possibilities of 3D-printing, TLS module of android smartphones, we can build an add-on image acquisition device that can be embedded inside available incubator designs. The acquired image will be analysed and annotated based on advanced CV/ML algorithms for morpho-kinetics. Investigators will be building a customized app which with the help of a high-quality phone camera, image sensor, along with TLS will help to take continuous videos of embryo development and by video analytics and machine learning algorithm, we can even select remotely the best quality embryos for transfer. |