Research On Antetype System Of Medical Image Database

Medical image database is the data storage center in medical information systems, and it's the important content of the PACS and other medical images is involved in the various medical information systems. Medical Image Database is the specific areas of medical image-oriented database technology. Medical Image Database is based on the image database, image processing, computer network technology and medical knowledge-based fields, effective supporting for medical image data storage, transmission, retrieval and management. Most of the medical image databases are currently based on existing mature relational database or expansion of relational database to manage medical image data. They were not forming perfectly storage theoretical models suited to medical image. The inquiry based on text of patient information and image basic information on retrieval does not satisfy the needs of clinical medicine and medical research. The study of effective inquiries based medical image content in mass image database , including feature extraction, similarity measure, and high dimension structure (feature organize), etc, had definite academic value and broad application prospects. Therefore, the study and exploration of medical image database technology suited to the characteristics of medical images has important practical significance.The study is based on analysing object relational database research thinking, techniques and methods, and explore from the organization for medical image database, storage technology, indexing technologies and achieving the initial prototype system and so on. These works include :1. Discussion on a relational database and object-oriented database in the Data Organization. This section detailly analyse the data organization of medical image database based on DICOM, including DICOM information model, the medical image data model, medical image complex object invoking and the expressing form of internal and external memory.2. research on the storage structure of medical image database and design a system directory structure and methods; According to the related information about DICOM standards, The article put forward a overall storage structure for complex object of medical image, and analyse the storage structure and methods of subtable, Accordingly realizing mixed storage for medical image object and relation. The article also summarized three available clustering strategy, and discussed their advantages and disadvantages.3. analyze and summarize the relevant high dimension indexing technologies and typical indexing structure; indicate indexing structure suited for medical image database. The article put forward a splitting algorithm based on the K-means clustering algorithm on analysing the structural features and algorithms of M_tree. It use virtual cluster center (center of mass) as a reference instead of the actual data objects usually on splitting algorithm. It can effectively decrease the radius of coverage and overlap nodes ,and reduce the frequency of visits disk(DAC) in the implementation of similar inquiries. It had the better query performance.4. Medical image database prototype system initially realize. Realize storage and query functions on the basic information of the medical image, including image storage and display. These basic information can be input by manual or be directly input from the DICOM documents. Especially the function that be input directly from the DICOM documents to the related information tables can greatly facilitate the application of clinical medicine and medical research, and has the higher value.