Brain magnetic resonance image segmentation using novel improvement for expectation maximizing.

Objective: To improve the quality of expectation maximizing (EM) for brain image segmentation, and to evaluate the accuracy of segmentation results.

Methods: This brain segmentation study was conducted in Universiti Putra Malaysia in Serdong, Malaysia between February and November 2010 on simulated and real images using novel improvement for EM. The EM-1 (proposed algorithm) was compared with neighborhood based extensions for fuzzy C-mean (FCM). The EM-1 was also applied to all 20 normal real MRI volumes and compared with reported results from the Internet Brain Segmentation Repository.

Results: In simulated images, the EM-1 outperforms neighborhood based extensions for FCM. The average similarity index value of the proposed algorithm for all 20 normal images is 0.802. The EM-1 produces the average Jaccard indices ρ higher than other algorithms and near to manual results. The average similarity indices ρ for EM-1 and FCM extensions (FCM with spatial information [FCM-S], Fast Generalized FCM [FGFCM]) for all 20 normal images are: EM-1=0.802, FCM-S=0.7517, enhanced FCM=0.7581, and FGFCM=0.7597.

Conclusion: Experimental results show that the proposed algorithm performs better than other studied algorithms on various noise levels in terms of similarity index, ρ.