alterlab-histolab

Whole-slide image tiling automates tissue detection, tile extraction, and normalization to generate ready-to-use datasets for deep learning in pathology. You load a slide, choose a tiler like GridTiler, and run extraction to produce tiles and a report.

Tissue mask extraction uses image and morphological filters to enable robust tile selection during whole-slide image preprocessing. This process isolates relevant tissue areas, discarding background artifacts before feeding tiles into deep learning models.

Yes, RandomTiler, GridTiler, and ScoreTiler are supported tiling methods for pathology slide analysis. RandomTiler selects random patches, GridTiler extracts uniform grids, and ScoreTiler ranks tiles by tissue content for targeted dataset curation.

Yes, preprocessing filters and tissue masks are supported to ensure robust tile selection for tissue characterization across cohorts. These filters standardize whole-slide analysis and maintain reproducibility across different pathology slides.

Manual tiling limits scalability and reproducibility compared to automated whole-slide image preprocessing. Automated preprocessing handles tissue detection, tile extraction, and normalization end-to-end, significantly reducing manual effort for large pathology cohorts.