Research On Storage And Fast Retrieval Of Medical Images

With the rapid development of society and the continuous popularization of medical modernization,the rapid development of medical digitalization,CT,MRI images are increasingly widely used in clinical work and auxiliary diagnosis,and medical images are also increasing.Massive data has brought huge challenges to data processing,data preservation,data sharing,etc.Medical image data has become an effective way to assist doctors in diagnosing patients’ conditions.Effective medical retrieval can help doctors make more accurate diagnosis of diseases,and obtain images of similar cases in the past to assist doctors in treatment.Main research work of this paper:(1)Research on storage methods of medical images: With the emergence of massive medical data,it is necessary to improve the speed of data processing,sharing and transmission.There are a lot of unstructured data,such as images,videos,audio,etc.Traditional relational database can not effectively store these data,and the data transmission speed drops rapidly.To solve this problem,object storage technology is adopted in this paper.This technology also provides users with file storage and file management functions,to study the system structure;At the same time,to address the single point of failure of the storage system,the erasure code technology is adopted in this paper to ensure that data is evenly stored on each storage node and the load is balanced on each node.This ensures the storage efficiency of the storage system and does not cause the entire storage cluster to break down due to the damage of one node,which effectively improves the storage efficiency.(2)Research on medical image retrieval methods: Different from natural images,medical images focus on extracting lesion area information to retrieve images of similar cases.Most medical images are gray maps with large noise,images of the same category have high similarity,small lesion area and are difficult to distinguish,leading to easy omission of important lesion information during feature extraction,affecting retrieval accuracy and other problems.As a result,existing feature extraction methods pay less attention to the features of focal areas and have low retrieval accuracy.In order to further improve the accuracy of medical image retrieval,we propose a medical image retrieval method based on deep hash,which is used to solve the problems such as small lesion area and difficulty in finding in the process of medical image feature extraction.In order to fully describe the detailed information,attention mechanism is added to the deep learning network,so that the network can better highlight the detailed information of the image.Then input the extracted high-dimensional features into the hash function to generate a compact hash code,which can effectively improve the retrieval accuracy and efficiency.Experimental results show that the proposed deep hash image retrieval method can effectively improve the retrieval accuracy.