Pancreatlas™: applying an adaptable framework to map the human pancreas in health and disease

Abstract

Deep phenotyping of human tissues generates complex spatial information from a variety of experimental modalities, yet images are typically condensed into small, static figures for publication and contain substantial data and metadata that never become widely available to the scientific community. Because image files are usually large, data is not easily shared or transferable even amongst collaborating researchers. While comprehensive image maps are available for some organs such as the brain, most resources are limited in their ability to accommodate multiplexed imaging with any degree of user interactivity. In recognition of this unmet need, we developed an online resource called Pancreatlas™ that integrates information technology infrastructure with enterprise imaging storage and visualization solutions. Through our custom interface, users can access curated, easy-to-navigate web pages, drill down to individual images, and deeply interact with them – all online, without lengthy downloads or software installation. Images are annotated with structured metadata, enabling users to dynamically build image datasets with biological and clinical relevance. The first version of Pancreatlas (v1.1) contains over 700 unique images acquired as whole-slide scans, confocal images, and imaging mass spectrometry, and is available at http://www.pancreatlas.org. While the overall system – entitled Flexible Framework for Integrating and Navigating Data (FFIND) – was deployed as a human pancreas-specific biological imaging resource, it can be configured to meet a myriad of imaging or other modular data management needs.