Quantitative Measurements Extraction and Annotation Optimization in Whole Slide Imaging

The adoption of digital pathology is growing rapidly, demanding new tools for decision support and better workflow. The improvements to the software and hardware are promoting a remote-ready environment where the glass slides are acquired in multiple distributed places, sent to a central repository, and reviewed in a collaborative modus operandi. Many of these solutions are already operating in the Cloud making use of standard Web interfaces and applications. However, the review process of a slide is still an arduous process due to the high resolution of images, requesting the integration of artificial intelligence tools for decision support. However, the huge number of pathologies evaluated and their specificities make this process hard and the regulatory authorities only recently started approving the first solutions for clinical use. This article presents a research platform, fully integrated within a PACS ecosystem compliant with the DICOM standard and built using pure web technologies. The platform was used in a scientific project to help pathologists annotate cases for the development of artificial intelligence tools. Preliminary results in the annotation of Helicobacter Pylori samples scanned at 40x magnification show that manual annotation is impracticable for pathologists and requests more automated approaches that are capable of improving the current annotation workflows.