This application addresses Innovations in Biomedical Computational Science and Technology Initiative PAR-09-220. Healthcare is moving towards fully digitized patient records. Surgical pathology represents a challenge in that high resolution imaging of tissue sections on glass slides produces datasets that are large compared to those of clinical pathology (e.g., blood tests) and radiology. Moreover the FDA recently announced its intention to regulate whole slide scanners as Class III medical devices, delaying implementation of digital pathology for primary diagnosis. Nevertheless digital pathology has enormous potential in other applications including education (teaching and training of physician residents and medical students), secondary consultation diagnosis and research. What is more there is a growing sense that application of digital methods can revolutionize diagnostics in anatomic pathology just as genomics technologies are predicted to revolutionize diagnostics in clinical pathology. Although the medical profession in general and anatomic pathologists in particular are currently seen to be slow in adapting the new technology, in the very near future pathology likely will take the lead in improving diagnosis through precision medicine (i.e., analytics of the large multi-part health datasets). Our Phase I efforts were focused on developing a high performance, open-source, client-server system for remote viewing of whole slide images. Our successful prototype system has been used weekly for teaching of Dermatopathology to residents in pathology and dermatology at Harvard University, and will be used at Washington University in St. Louis and at the University of New Mexico. For Phase II we are proposing to extend the client-server system to support peer-to-peer data sharing across a distributed database, and remote collaborative viewing and interaction. We will provide an open data repository for users to upload and share their images, and develop additional content on the 6,000 plus images already gathered in Phase I. Further we will promote innovation and extend our open, agile architecture to support plug-in support for analysis modules (e.g., to support CAD) and demonstrate this capability by using the open-source, medical image analysis toolkit ITK. Finally, through the use of a pilot Challenge Study we will begin to quantify early anecdotal evidence that the digital pathology system is faster and provides better results in certain applications than conventional approaches using physical slides. Our goal is to establish widespread use of the system (server software) and/or service (web portal for uploading and viewing images) to create a large pathology network. During the proposed two-year Phase II we will deploy the software and support the services to partnering institutions (there are six currently). In addition, the co-PI Dr. Faulkner-Jones will develop a curated slide image collection that will build on the current collection of trusted, high-demand content.
The work described here will create and deploy modern digital pathology systems which reduce costs by reducing slide handling. Further, the project will provide new capabilities including data archival, sharing, retrieval and audit;telepathology (remote viewing);annotation and mark-up;3D visualization and informatics;use of automated data processing algorithms;and education (curated and annotated image sets).