BioHIPI: Biomedical Hadoop Image Processing Interface

Nowadays, the importance of collecting large amounts of data is becoming increasingly crucial, along with the application of efficient and effective analysis techniques, in many areas. One of the most important field in which Big Data is becoming of fundamental importance is the biomedical domain, also due to the decreasing cost of acquiring and analyzing biomedical data. Furthermore, the emergence of more accessible technologies and the increasing speed-up of algorithms, also thanks to parallelization techniques, is helping at making the application of Big Data in healthcare a fast-growing field.This paper presents a novel framework, Biomedical Hadoop Image Processing Interface (BioHIPI), capable of storing biomedical image collections in a Distributed File System (DFS) for exploiting the parallel processing of Big Data on a cluster of machines. The work is based on the Apache Hadoop technology and makes use of the Hadoop Distributed File System (HDFS) for storing images, the MapReduce libraries for parallel programming for processing, and Yet Another Resource Negotiator (YARN) to run processes on the cluster.