One of the biggest challenges in developing solutions for medical image diagnosis is the lack of efficient open-source annotation tools which have the ability to manage and label large-scale datasets. Thus, the Medical Imaging Department, Vingroup Big Data Institute (VinBigdata) decided to release VinDr Lab – our DICOM annotation tool into the open-source.
For the development of Artificial Intelligence (AI) applications, data is the key factor. To be clean and valuable for algorithms, the data needs to be labeled and annotated by humans.
In fact, there are many open source software available for labeling data of natural images, speeches, and documents. However, there is no open-source annotation tool that allows label management and assignment for large-scale medical datasets.
VinDr Lab is an open-source software for managing and labeling medical image data. Developed by VinBigdata, the software aims at eliminating difficulties that engineers and organizations face in the process of building medical Artificial Intelligence-based solutions. Users can completely customize the source code to serve their own purposes.
Currently VinDr Lab is supporting to label X-ray images of lungs, breasts and bones. The annotation features for CT and MRI images are under development and expected to release in the near future. Sharing open source code on Github, VinDr Lab developers hope for technology community support and contribution to gradually upgrade the solution.
1, Project management
2, Label management
3, Advanced annotation tools
4, Task management
Main github project as well as member projects have instruction file (README.md) for user reference.