Big Data based system for Biomedical Image Classification

Medical field deals with large volume of data. Among them, the medical images are extremely significant. The existing traditional methods are not efficient enough to manage this huge amount of data. For the efficient management and storage, big data technology is used. Hence, in this work, the processing and analysis of such huge amount of biomedical image data has been done by conglomerating technologies like Big Data and Machine Learning. The designed system takes as input, the features extracted from various Diabetic Retinopathy (DR) images. These features are used for further classification. The classification has been performed using K-Nearest Neighbor machine learning classifier (KNN) implemented in Hadoop MapReduce framework, in order to detect the absence or presence of DR. The performance of the Hadoop MapReduce framework has been analyzed using the execution time. The analysis has been done by comparing the execution time for the classification performed by Hadoop MapReduce framework and the execution time for the classification performed by the Python framework for datasets of different sizes. At the end of the analysis, it has been found that the Hadoop MapReduce framework can handle bigger datasets more efficiently for classification than the Python framework.