delestro / openfertility

Open Fertility is an open-source project dedicated to advancing fertility analysis and prediction through machine learning.

Open Fertility is an open-source project dedicated to advancing fertility analysis and prediction through machine learning. The goal is to foster a community-driven initiative that empowers fertility professionals, researchers, and enthusiasts.

Inspired by the paper An annotated human blastocyst dataset to benchmark deep learning architectures for in vitro fertilization the aim is to leverage the open data they published to develop initial models.

The primary objectives are to:

• Cultivate an inclusive and collaborative community, bringing together experts and enthusiasts.
• Provide comprehensive tools for training, evaluating, and applying machine learning techniques.
• Effective data visualization, to gain valuable insights and interpret data with clarity.
• Deliver intuitive user interfaces that grant easy access to the models.

Get in touch if you feel interested in participating!

Installation

During this early develoment stage, a PyPi pip installation is still not provided.

To use the current version, follow these steps:

1. Download and unzip (code button above the files) or clone the repository
2. Navigate to the directory containing the repository
3. Run pip install .

Early features

Download the blastocyst dataset

import openfertility as of

blasto2k = of.datasets.blasto2k.Dataset()
blasto2k.download()

Generate training set

import openfertility as of

blasto2k = of.datasets.blasto2k.Dataset()
blasto2k.download()
train_set = of.datasets.splits.TrainSet(blasto2k)

print(f'Image shape: {train_set[0][0].shape}')
>>> Image shape: torch.Size([3, 384, 512])

Show random grid from dataset

import openfertility as of

# Get the dataset
blasto2k = of.datasets.blasto2k.Dataset()
blasto2k.download()

# Generate train set
train_set = of.datasets.splits.TrainSet(blasto2k)

# By default shows labels on the top left corner
of.display.img.grid(train_set)