An Efficient Hybrid Approach for Brain Tumor Detection in MR Images using Hadoop-MapReduce

The sheer expanding of brain tumor MR image in terms of volume demands swift and accurate processing mechanism. To segregate tumor from brain MR effectively an efficient computer aided detection system is required. The research to maximize the classification accuracy rate is improving at fast pace in field of medical image processing. However, somehow setbacks the execution time of the overall system. The objective of the research is now slanting to be proficient enough in both the performance parameters. In this manuscript, the focus is to develop a framework using fusion based approach to balances the accuracy and time simultaneously. The hybrid fuzzy c-means and k-means (FKM) methodology is implemented using Hadoop MapReduce platform to increase the scalability of clusters. Experimentation is verified using DICOM images of variable size datasets processed through multi-node Hadoop platform. The size of the images is taken is slots of 200, 400 and 600 images. The accuracy rate core fusion approach (FKM) to be 96% considering the maximum number of images. The execution time decreases as the number of nodes increases due to the parallel and distributed image data. The system without hadoop framework takes 30% more time as compare to the system with single nodes using 1000 images.