The Multi-Scale Task Multiple Instance Learning library was described in the paper Multi-Scale Task Multiple Instance Learning for the Classification of Digital Pathology Images with Global Annotations by Marini et al., and it is a library containing different scripts to extract patches from several Whole Slide Images’ scales. The scripts, given a WSI and the corresponding tissue mask, extract and store the patches. The scripts developed are four, considering the two modalities developed to extract the patches and the two possible types of tissue masks used. The tools include two possible modalities to extract the patches: grid extraction and multi-center extraction.
The library is freely available on its GitHub repository, together with further details on its functioning and applications.