Parallel processing researches of Medical image based on Spark

Abstract:

With the growing number of medical images, in order to effectively and quickly process medical images and make all kinds of medical images processing method get quickly transplant and apply. A medical image parallel processing method based on Spark is proposed, which integrate traditional medical image processing method with the Spark. Firstly, preprocess images base on binary conversion method, store images in the HDFS distributed file system. Secondly, implements a transfer function way, which realized image parallel processing quickly, avoid convert the image processing algorithms to the MapReduce algorithms. Thirdly, according to the experiment, medical image parallel processing method based on Spark has preferable adaptability and efficiency, which can adapt to the large-scale image processing.