Medical Image Data Storage And Semantic Retrieval

Medial images are necessary to the clinical diagnosis and treatment. Managing magnanimous medical information with fast developing network technology, sharing medical information in distributed environment, retrieving the interested medical image from the database are the top three issues to be solved presently. The traditional medical image system is PACS which known as the picture archiving and communication system. Such system is based on the central server and the enclosed architecture. It can only serve limited medical image workstations. The enclosed architecture limits the system capacity. Moreover, the medical images can only be retrieved by case number or date but the diagnosis or the content of the image.Three issues were discussed. The first issue was the management of distributed large-scale medical image data. The second issue was the safe sharing of medical image data in grid environment. The third issue was the building of medical image database with MPEG-7 technology and retrieving medical images by semantic information such as abnormal structures in images.Distributed storage technology and the iSCSI technology were applied in the management of medical image data. The developing of large-scale medical image center was introduced from the aspect of distributed storage structure. The schema of using Globus technology to share the medical information was proposed. Based on the above schema, a medical image sharing system using GridFTP protocol was designed. The data safety under Globus environment was also discussed. The medical image meeting the DICOM3.0 (digital imaging and communications in medicine) criterion was studied. The technology of content-based image retrieval (CBIR) can search the interested image from the database by low-level character such as color, texture, and shape. Using MPEG-7 technology to describe the medical image semantic information, such as the diagnosis and the abnormal structures, was proposed.As the result, medical image storage problems were solved from three aspects. First, a scalable medical image storage system was established in the local area network. Second, a secure medical information sharing system was designed in the wide area network. Third, taking mammography as an example, a medical image database with MPEG-7 was built. The above improvements enhanced the retrieval efficiency and can search the image by high-level semantic information such as the diagnosis, the abnormal structure and their positional relationship.