Hybrid clustering algorithm for an efficient brain tumor segmentation

Abstract

This work describes the data mining methods, techniques and algorithms used for implementation. It is an emerging field of IT industry and research. There are many other fields such as Artificial Intelligence, Machine Learning, Deep Learning, Virtualization, Visualization, Parallel Computing and Image Processing. The human internal Brain can be seen or visualized by the Magnetic Resonance Imaging scan or Computerized Tomography scan. The MRI image is scanned and will be taken as input for processing. The MRI scan is more advantageous and more comfortable than CT scan for diagnosis. MRI scan provides detailed picture of organs. It does not affect the human health and body condition. It doesn't use any radiation. It is purely based on the magnetic field and radio waves. LIPC technique makes the training samples from the patients and arranges them into different group of classes used to construct different dictionaries. Image segmentation is a technique of dividing an image into different multiple portions, which is used to spot out objects and boundaries in images. There are many image segmentation techniques applicable for image processing. No acceptable method is available for solving all kinds of segmentation problem. Every method has merits and demerits. So, choosing good method is the challenging task. The hybrid clustering method is proposed in this work. The k-means algorithm and fuzzy c-means algorithm is proposed for brain tumor segmentation. The algorithm is implemented in synthetic and real time dataset. From the experimental results, this method provides better results in the form of accuracy.