Development of a Web-Based Digital Pathology System Used for Dermatopathology Teaching
Beverly Faulkner-Jones is a practicing Dermatopathologist at Beth Israel Deaconess Medical Center in Boston, and an Instructor in Pathology at Harvard Medical School. She received her medical degree from the University of Birmingham Medical School, UK, and then her PhD from the Walter and Eliza Hall Institute of Medical Research in Melbourne, Australia. Her postdoctoral work in Melbourne and St. Louis focused on the developing vertebrate central nervous system and eye. She completed pathology residency and fellowship training at Barnes-Jewish Hospital / Washington University School of Medicine in St. Louis, MO and is board-certified in Anatomic Pathology and Dermatopathology. Her current research interests revolve around the application of digital pathology to education and research.
Background: We have developed a web-based client-server digital pathology system that supports high-performance viewing of whole slide images over standard networks. The system is currently being used to teach dermatopathology to residents and fellows at Harvard teaching hospitals. This software is freely available via a public facing repository and has a liberal open source license.
Methods: The system is built upon a standard HTTP server and distributed Mongo databases. We use a multi-resolution database slide format to make the system scalable with respect to image size. Our web-based system manages: data transfers, the processing of whole slide images, and authenticated access to slide image data (integrated with Facebook authentication). A WebGL client provides high performance viewing on desktop computers. Multitouch-enabled image navigation is supported on mobile devices (e.g. iPad). The system also has scale-invariant annotation capability.
Results: We set up a dermatopathology slide-atlas website for weekly interactive teaching sessions, using the digital pathology system. Images are organized into study sets that can include videos, lectures, papers and notes. Access is controlled at the level of individual study sets. Multiple users can stream image data simultaneously from locations around the country without noticeable loss of performance. We have demonstrated the scalability of the system by serving out a terapixel montage of tiled electron micrographs. Resident feedback, including a subjective increase in diagnostic confidence, has been uniformly positive. The slide-atlas website demonstrates the usefulness of this system for teaching and training.
Conclusions: The slide-atlas website is a useful tool for sharing and annotating whole slide images. In the near future we will extend the system to support search and access to images across different servers. The website will serve as a unified interface for the network of servers. Users will be able to create accounts, upload whole slide images, add comments and annotation, and control which users and groups can have access to their images. We also plan to develop algorithms in ITK, the insight image segmentation and registration toolkit, for automated quantitative analysis.
Understand the design and application of an open source digital pathology system for teaching dermatoapthology.
Click here to see the list of other speakers and additional education information.