A critical and longstanding problem in radiology has been the radiologist's lack of access to a patient's prior imaging studies. While comparisons between """"""""priors"""""""" and new images are the standard of care and have been found to influence diagnoses in 56 percent of cases, change diagnoses in 16-20 percent, and detect new pathology in 6 percent, they are far too infrequent due to the balkanization of medical data among unaffiliated medical institutions. Mainstream acceptance of PACS, DICOM, and the internet, combined with new federal regulations for privacy and security, however, offer an opportunity to rectify this age-old bottleneck for the first time. Our firm has developed a prototype system capable of unifying an arbitrary number of hospitals into a unified, but distributed repository of digital radiological studies that is securely searchable and visualizable over the internet. Here we propose to study the feasibility of our design and implementation by linking two now-separate PACE at the Hospital of the University of Pennsylvania, quantifying our platform's performance and impact upon the hospital information infrastructure, and optimizing our design to minimize the load placed upon the PACS and RIS. Phase II will set the stage for commercialization of the technology by incorporating probabilistic record linkage for cross-institutional indexing of patient records, enhanced security, seamless integration with softcopy diagnostic workstations, and image compression and streaming for efficient web-based viewing. Most importantly, Phase II will involve the addition of Children's Hospital of Philadelphia to the network, creating for the first time a platform supporting cross-institutional sharing of imaging studies between unaffiliated hospitals for routine use in pediatric and adult patient care. Upon successful completion of Phase II, commercialization will involve the introduction of the first service enabling arbitrary hospitals to providing images to their peers while having routine access to much-needed patient-specific data. Future enhancements will include support for multi-institutional collaborative research and retrospective studies.
