WSI2ML – An Open-Source Whole Slide Image Annotation Software for Machine Learning Applications

Machine learning (ML) has emerged as a powerful tool for improving the clinical pathology routine; however, developing novel ML research requires a complex multidisciplinary team effort. Therefore, a software for tackling the challenges of effectively annotating, building, and validating ML models is desirable. In this work we present WSI2ML, a web-based platform that provides a friendly interface suited for ML computational pathology research to be easily performed. Compared to similar tools currently available, the proposed software provides a complete toolset for each stage of ML workflow. We demonstrate the usefulness and functionality of WSI2ML by analyzing the performance results obtained in a tissue recognition task using a novel gastric cancer research dataset that is currently being developed with the tool. The software, documentation, installation instructions and related annotation handling library is freely available at https://luanvcmartins.github.io/WSI2ML/.