Coupling AI and Citizen Science
In Creation of Enhanced Training Dataset for Medical AI Training
Crowd Sourcing Platform Demo
Implemented based on Hugging Face, a open-source machine learning platform aiming at democratising the collaboration of models, datasets, and applications.
Our demo can be accessed via: https://fuurei-labelstudio-instruction.hf.space/
This project explores the potential in coupling artificial intelligence (AI) and crowd sourcing to enhance the rate of data gathering for medical image segmentation. Utilising advanced segmentation model MedSAM alongside generative AI pix2pixGAN, the research aims to not only augment existing medical datasets but also refine the accuracy and efficiency of data annotation through public participation. By integrating a user-friendly online platform, this approach facilitates the annotation of medical images by a diverse group, aiming to simplify and expedite the data collection process necessary for training robust deep learning (DL) models. The combination of AI with citizen science is hypothesised to significantly improve the size and quality of datasets available, making it a promising approach to overcoming current limitations in medical image analysis and aiding in the progression of medical diagnostics and treatment planning.
