During an IVF cycle, many embryos are usually created from a single patient, but not all of them have equal chances of development and pregnancy. Out of this pool of embryos, the embryologists must select the best one or two embryos for transfer. The process of embryo selection is of paramount importance, and can make the difference between success and failure of IVF treatment. An accurate and rapid way of evaluating IVF embryos is therefore a need in the IVF lab.
Morphological evaluation of embryos can often be misleading and may not reflect the true viability of an embryo. The result of the current uncertainly to identify the single best embryo is that more than one embryos are placed back to the woman’s womb. This may slightly increase the chance of pregnancy, but also increases multiple pregnancies, which are a common complication of IVF treatment, putting babies and their mothers at risk.Get Started
We have developed an Artificial Intelligence (AI) application for evaluating human embryos from images of blastocysts of known implantation outcome. IVFvision.ai predicts the probability of implantation of a single blastocyst using computer vision, offering a fast and intelligent solution to the problem of laborious manual embryo evaluation.Get Started
CNNs have a set of filters working on localised regions that make connections in small two-dimensional areas of the input image, called the local receptive fields. CNNs use the same weights and biases for each of the hidden neurons. By sharing the weights, the network is forced to learn invariant features at different regions of the image. Thus, all the neurons in the layer detect the same feature but at different locations in the image.
Pooling layers are a type of layer typically used after Convolutional layers. They summarise the information from the convolution layer by performing a statistical aggregate function, typically average or max, applied to each feature map and by producing a compressed feature map. Forward propagation evaluates the activations, and backward propagation computes the gradient from the above layer and the local gradient to calculate gradients on the layer parameters.
We fed images to the AI System with a sequence of convolution and pooling layers; convolution to extract spatial invariant features from a subsample using the spatial average of maps and a multilayer Neural Network (MLP) as a final classifier (fully connected layers); and a sparse connection matrix between layers (weight sharing) to avoid a large computational cost and reduce overfitting.
IVFVision.ai is able to identify blastocysts that are most likely to implant with high accuracy, sensitivity and specificity. Our approach has high predictive power and offers a fast, non-invasive and inexpensive solution to IVF laboratories around the world, seeking to maximize their success rates. The system only requires the image of a blastocyst, and can be compatible with time-lapse systems, as well as with standard microscopy observations in laboratories without time-lapse. The application can be adapted to any type of equipment configuration, to suit all laboratory requirements.