Netlab: A platform for crowdsourcing virtual microscopy knowledge

Abstract

The Netlab project is part of an ongoing effort in image informatics that began in early 1985. This early work resulted in the development of the first digital histology atlas system, which was presented at a NASA conference in 1987. That system used a technology called MetaMAP to make every pixel of the image an independently‐addressable hotspot, linkable to histology knowledge data relating to the cell or structure at that location in the image. The MetaMAP system was then applied to larger anatomy imaging projects, first in the Visible Embryo Project, at UCSF, in 1994, then in the Muritech Atlas of Mouse Development, in 1996, and the Eolas AnatLab project in 1998 and later. In the AnatLab system, MetaMAP allowed the creation of over 800,000 3D anatomical annotations to be made on the image data from the Visible Human Project. The scale of that effort exposed the extreme difficulty involved in validating this kind of data, even when the annotators were trained anatomy students supervised by a professor. The Netlab system is an attempt to lay the groundwork for both building virtual microscopy resources of unlimited scale, and leveraging the power of crowdsourcing to facilitate detailed and validated annotation of the microscopic image data accessible through the system. The Netlab client software allows users to access Aperio‐scanned slide image data streamed in real time over the cloud from the Netlab servers to either a Web browser or a mobile app. Users can pan and zoom in and out across the magnification ranges typical of microscopes in the lab, and can place textual annotations at any locations on any slides in the collection. Each annotation can be classified as private, group or public. Any user with access permission to view another user's annotation can validate that annotation. Each validation is assigned a weight based upon the role of the validator (peer or professor, for example). Group moderators (e.g., professors) can set thresholds for validation. Annotations above the threshold have a “verified valid” icon displayed when viewed. Future work will focus on both expanding the image collection and the development of a more automated validation system, using an approach inspired by the proof‐of‐stake distributed consensus algorithms currently popular in the blockchain technology field.This abstract is from the Experimental Biology 2018 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.