A New Hybrid Image Segmentation Method Based on Fuzzy C-Mean and Modified Bat Algorithm

Abstract

Magnetic resonance imaging (MRI) plays an important role in clinical diagnosis, because of that it has attracted increasing attention in recent years. The symptom of many diseases corresponds to the brain's structural variants. The detection of various diseases has became very useful through the segmentation methods. Fuzzy c-means (FCM) considers among the popular clustering algorithms for medical image segmentation. However, FCM is sensitive to the noise and falls into local optimal solution easily because of the random initialization of the cluster centers. In this research, we propose a hybrid method based on modified fuzzy bat algorithm (MFBA) and the FCM clustering algorithm named MFBAFCM. This developed approach uses the MFBA to get better initial cluster centers for the FCM algorithm by using a new fitness function, which combines intra cluster distance with fuzzy cluster validity indices. Experimental results on several MRI brain images corrupted by different levels of intensity non-uniformity and noise, show that the proposed method produced better results than the standard FCM and some other recent published works.

Brain Image Segmentation in Recent Years: A Narrative Review

“…For healthy brain tissue segmentation, the performance of hybrid metaheuristic MOPSO and fuzzy clustering region-based active contours was found to be better than that of the others for both Dice score and sensitivity, as stated by Pham et al [51]. Boulanouar and Lamiche [53] reported that good results were obtained for both GM and WM segmentation using the Brainweb dataset. For segmentation of the hippocampus structure in the brain, the performance of hybridized deep learning based on 3D U-net with CRF was found to be better than that of the other methods as measured using the Dice score by Jiang and Guo [62].…”

“…Additionally, this type of hybridized approach is generally used to solve or reduce the major drawbacks of machine learning segmentation methods, such as the possibility of being trapped in a local minimum and sensitivity to noise. Several studies [50][51][52][53][54][55] have employed a combination of metaheuristic and machine learning methods.…”

“…Recently, Boulanouar and Lamiche [53] introduced a new hybrid method based on a modified fuzzy bat optimization algorithm (MFBA) and the FCM clustering approach termed MFBAFCM for brain tissue classification. The MFBA algorithm was utilized to obtain the optimal cluster centers, which were subsequently utilized as the adaptive initial seed for the conventional FCM.…”

Recent advances of bat-inspired algorithm, its versions and applications

“…Another domain where using BA shows its superiority is image and signal processing, in which the algorithm solves image segmentation [ 28 , 35 , 45 , 77 , 78 , 196 , 204 , 221 , 229 , 255 , 322 , 323 , 327 , 328 ], face recognition and fingerprint identification [ 23 , 44 , 80 , 230 ], magnetic resonance (MR) brain tumour classification [ 151 – 153 , 271 ], multilevel image thresholding [ 79 , 207 , 254 ], and signal processing application [ 5 , 161 , 302 ]. Figure 19 plots the applications of BA in image and signal processing domain.…”

Enhanced independent component analysis and fuzzy C-mean clustering based on novel bat algorithm for noisy image segmentation

“…When tested on breast cytology images, the hybrid method, FCMGWO, demonstrated superior performance over the standard FCM, particularly in less detailed images. In [30,31], a new approach is developed to calculate the optimal initial values of clusters centroids based on a novel fitness function. The authors introduce a hybrid MRI segmentation method, MFBAFCM, combining the modified fuzzy bat algorithm (MFBA) and Fuzzy c-means (FCM).…”