| Medical imaging is moving toward filmless, fully digital media. This is changing the healthcare industry, as it promotes the ability to perform remote diagnoses of patients. To fully benefit from this advancement in technology, a medical storage network must be developed so that medical images may be shared among medical professionals throughout the United States and retrieved quickly in emergencies. In this work, medical imaging data storage requirements are developed through an analysis of imaging data. In addition, models and algorithms are developed to determine the optimal placement of medical storage facilities, as well as number of servers required.; Key network design issues for a large radiological database include the location of the data repositories relative to the generating and retrieval (reading) sites and the number and location of the copies of the files that are generated. In this work, three different classes of location models aimed at assisting in the design of such a data repository network are outlined. The first model is a median-like model that minimizes the average time to retrieve an image. The second model presented is in the class of covering problems. The third model is in the class of center problems.; Each problem results in a large integer-programming model, incapable of being solved by commercial integer programming solvers. To address this issue, a nested Lagrangian relaxation algorithm is devised. The algorithm is outlined, and the results of the algorithm are then compared to provably optimal results obtained using a commercial solver for small problems for even moderate-sized problem instances. The computational results for each problem are discussed on medium-sized problems in which the number of nodes is several times the maximum size capable of being solved using commercial codes. The qualitative conclusions about the key determinants of service in such networks are presented.; From a storage retrieval perspective, since the storage retrieval rate is constant, the goal is to ensure that requests from all sites are equally served, and all requests may be served within the given timeframes. Therefore, the key storage design concern is the number of servers per site. Within the models, the number of servers required to support a specified number of requests within a given time period is determined.; Finally, suggestions for future work are proposed including the prioritization of network traffic, as well as broadening the scope of the medical imaging network to include other medical data. |
