Medical Data Retrieval Technology Based On Mode Network

The hospital information systems collect more and more types of medical data.How to retrieval and mine these data effectively will find out the doctors’ interest and medical knowledge.Such information will help doctors to make a more accurate diagnosis.As special multimodal data,however,there are many semantic relationships between different modals of medical data.Besides,medical data does not fixed document structure in the retrieval process.Therefore,to retrieval of medical data should meet the needs of multi-modal and cross-modal.The traditional medical data management and retrieval techniques for mono-modal data can not fuse modals of medical data.And the existing mode fusion and retrieval techniques for multimedia document can not meet the needs of multi-modal and cross-modal.This paper analyzes and defines the concept of mode,and then proposes the mode network.Using mode network,this paper completes the modeling and storage of medical data based on the semantic relationships and graph database.It establishes the storage structure for mode fusion and data retrieval based on semantic relations.To improve the efficiency of data modeling and loading,the parallel strategy is used to load medical data.The experiments show that the parallel algorithm has good efficiency and stability.Based on the mode network and graph database,this paper proposes multimodal index graph as the index structure of multimodal medical data.Multimodal index graph describes the semantic relationships between the data’s indexes and the correlation between different modes.Based on the correlation of different modes,this paper implemented the modes fusion.Finally,this paper proposes a multimodal medical data retrieval framework.Based on the framework,a retrieval model and the corresponding algorithm for multimodal medical data are proposed.Experiments show that the retrieval algorithm can meet the needs of multi-modal and crossmodal retrieval.In summary,this paper achieved modes fusion based on the semantic relationships in medical data.And finally achieved the multi-modal and cross-modal retrieval based on modes fusion.