A Novel Approach to Improving the Efficiency of Storing and Accessing Medical Image Files on Hadoop

In this paper, HDFS is used to solve the storage problem of mass medical image data. However, HDFS does not consider the correlation between medical small files. As an open source distributed file system, HDFS has reliability, scalability, low cost of storage capacity and many other advantages. But it has poor performance on the directly processing of small medical image files. In this paper, multi-strategy merge model-HMERGE is presented to solve the problems of size differences and number difference between the common medical image types, like CT, CR and US. On this basis, we design prefetching and caching mechanism based on the correlation between small files to improve storage and access efficiency. Experiments show that the presented scheme can effectively reduce the load of NameNode in HDFS and improve the efficiency of storing and accessing small medical image files.