Tile Serving Without a Server to Traverse Whole Slide Imaging Data on the Web

Abstract Background: Whole slide imaging (WSI) has become the primary modality for digital pathology analytics. However, due to the size and high-resolution nature of whole slide images, it becomes necessary to use specialized software platforms to operate on them. These platforms engage WSI data by “tiling” the image, essentially operating only on sub-regions of the slide at a time. To execute this computationally intensive operation, these platforms typically demand that the data be locally available, raising governance concerns. Moreover, they come with potentially heavy licensing fees and require costly computational infrastructure, all the while making it difficult to distribute any analyses outside of their “enclaved” environment. Results: To address these issues, we developed ImageBox3, an in-browser tiling mechanism to enable zero-cost, zero-footprint traversal of remote WSI data. The key achievement of this work was to devise a web component that brings the computation behind tiling WSI data to the user’s personal browser. All computation required to view and operate upon WSI data is performed client-side, without compromising user governance for public and private images alike. As a result of operating on the web, ImageBox3 is always available, requires no setup whatsoever to use, and can be distributed immediately at no cost to either the provider or the user. Conclusion: Our work removes significant logistic and regulatory hurdles to collaborative WSI operation. It is a step towards the democratization of analytical tools so as to foster a participative, FAIR data commons for digital pathology. This is delivered as an open source framework (https://github.com/episphere/imagebox3) and as a ready-to-use web application (https://episphere.github.io/imagebox3).