eipm / stork-v

This project aims to enhance the assessment of fertilized human embryos, a critical step in the process of in vitro fertilization (VF). The research paper behind this project introduced BELA, the Blastocyst Evaluation Learning Algorithm, a novel model for embryo ploidy status prediction.

BELA: Accurate and Noninvasive Ploidy Prediction for Human Preimplantation Embryos

Project Description

This project aims to enhance the assessment of fertilized human embryos, a critical step in the process of in vitro fertilization (VF). The research paper behind this project introduced BELA, the Blastocyst Evaluation Learning Algorithm, a novel model for embryo ploidy status prediction. It surpasses both image- and video-based ploidy models without necessitating any subjective input from embryologists. This project uses deep learning techniques to accurately and noninvasively predict ploidy in preimplantation human embryos.

Set up local environment and install dependencies

Install requirements from requirements.txt pip install -r requirements.txt

Add the models in the folder src/stork_v/models

Execute a model as script

Add the zip file in the expeted location src/data/78645636.zip and run the command python src/test.py

Docker run command

PATH_TO_MODELS=<set the location of your models directory>
PATH_TO_DATA=<set the location of you data directory>
PATH_TO_TEMP=<set the location of the temp directory used for creating the videos>

docker run --name stork-v \
-v $PATH_TO_MODELS:/stork-v/stork_v/models \
-v $PATH_TO_DATA:/stork-v/data \
-v $PATH_TO_TEMP:/stork-v/temp \
-e USERS_DICT="{'user1': 'stork'}" \
-p 8080:80 \
stork-v

Instructions for Use of Training Files

Files are available for training a BELA model in the 'main' folder.

Files available provide code for training and evaluating a BELA model through:

1. Video Creation
2. Annotation File Creation
3. Training and Prediction

Contributors

• Suraj Rajendran, Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA, Tri-Institutional Computational Biology & Medicine Program, Cornell University, NY, USA
• Matthew Brendel, Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA
• Josue Barnes, Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA
• Qiansheng Zhan, The Ronald O. Perelman and Claudia Cohen Center for Reproductive Medicine, Weill Cornell Medicine, New York, NY, USA
• Jonas E. Malmsten, The Ronald O. Perelman and Claudia Cohen Center for Reproductive Medicine, Weill Cornell Medicine, New York, NY, USA
• Pantelis Zisimopoulos, Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA
• Alexandros Sigaras, Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA
• Marcos Meseguer, IVI Valencia, Health Research Institute la Fe, Valencia, Spain
• Kathleen A Miller, IVF Florida Reproductive Associates, Fort Lauderdale, Florida, USA
• David Hoffman, IVF Florida Reproductive Associates, Fort Lauderdale, Florida, USA
• Zev Rosenwaks, The Ronald O. Perelman and Claudia Cohen Center for Reproductive Medicine, Weill Cornell Medicine, New York, NY, USA
• Olivier Elemento, Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA
• Nikica Zaninovic, The Ronald O. Perelman and Claudia Cohen Center for Reproductive Medicine, Weill Cornell Medicine, New York, NY, USA
• Iman Hajirasouliha (Corresponding author), Institute for Computational Biomedicine, Department of Physiology and Biophysics, Weill Cornell Medicine of Cornell University, New York, NY, USA, [e-mail]

Contact

For any questions or comments, please contact Iman Hajirasouliha at [e-mail].