This project is aimed at developing a web tool that will allow researchers to automatically detect cells with malaria parasite and quantify its density. Importantly, it is able to identify the developmental stage of the parasite (P. falsiparum), which has not been done before and only became possible due to the unique dataset that was provided to us.
Pipeline
- Cell segmentation from microscope images using cellpose
- ROI extarction
- ROI classification using finetuned Resnet18 and/or SqueezeNet
The app was deployed to Heroku. Note: it is much slower due to the limitations of the free-tier server.
Powered by PyTorch and Streamlit