1. UKM Learning and Research Repository
2. PHD Thesis / Tesis PHD
3. Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Title:	Dynamic fuzzy c-mean based firefly photinus search algorithm for MRI brain tumor image segmentation
Authors:	Waleed Khamees Ali Alomoush (P56199)
Supervisor:	Khairuddin Omar, Prof. Dr.
Keywords:	MRI brain tumor Firefly algorithm Image segmentation
Issue Date:	3-Aug-2014
Description:	Fuzzy C-Mean (FCM) is a part of Fuzzy clustering algorithms which also falls under unsupervised machine learning. FCM has the advantages of considering the overlapping clusters since it is a common problem in MRI brain tumor images segmentation to determine normal and abnormal tissue. However, there are two main issues plague FCM algorithms: initialization sensitivity of cluster centres and an unknown number of actual clusters in a given dataset. Previous studies have considered metaheuristic-based clustering approach such as Firefly Algorithm (FA) because of its feasibility and practical searching solution. However, Firefly algorithm has some disadvantage such as getting trapped into local optima. Firefly algorithm parameters are set fixed and they do not change with the time. In addition Firefly algorithm has weaknesses on optimizing high-dimensional problems. Besides that, it does not memorize or remember history of situation for each firefly and this causes them to move regardless of its previous situation, and they may end up missing their situations. Next, the automatic MRI brain image segmenting is a difficult and complicated process due to pattern diversity in tumour tissue appearance, overlapping tissues and partial volume effects. First is to improve the performance of MRI brain image segmenting, the FA has been enhanced by introducing Firefly Photinus Algorithm (FPA). An FPA is proposed by initializing mate list and new absorption parameters to solve problems of trapped into local optima, remember history of situation, and change the parameters with the time. FPA is done simultaneously in order to produce significant partitioning of the given dataset. Here, FPA is evaluated and compared to FA using ten benchmark problems. The outcome shows that the proposed FPA outperformed FA. Secondly is to propose fuzzy clustering algorithms based on firefly photinus called (FPFCM) algorithm to solve challenges of selecting the initial cluster centers and compute the neighborhood term of per iteration. The proposed FPFCM have been evaluated by MRI brain images such as simulated data. The experimental results shows significant improvements compared to other state of the art approaches in the same domain. Thirdly, is to propose Dynamic FCM based on firefly photinus Algorithms called (DCFPA). Overall, DCFPA algorithm will give better performance in terms of accuracy by finding an optimal number of clusters and suitable location of cluster centroids. The proposed DCFPA have been evaluated by MRI brain images such as simulated, real brain data and synthetic images. The experimental results show significant improvements compared to other state of the art approaches in the same domain.,Ph.D
Pages:	174
Call Number:	TA1638.4 .A466 2015 3
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.