Abstract—The aim of this paper is to propose a matlab toolbox of a comparative study of four brain tumor segmentation methods with specific sequences. Four methods are K-means clustering, fuzzy C-means clustering, Region growing, and Otsu segmentation. Skull stripping algorithm has done before segmentation. Different preprocessing algorithms are applied before segmentation to enhance images and give better segmentation. Median filter, Gaussian filter, and contrast enhancement are applied according to segmentation method. The database used is based on DICOM images. GLCM used for feature extraction. Classification and accuracy measurement are done by ANN.

Index Terms—ANN, fuzzy C-means, GLCM, K-means clustering, otsu segmentation, region growing.

The authors are with Minia University /Biomedical Engineering, Minia, Egypt (email: fatmasayed93@gmail.com).

Cite: Ashraf M. Said, Fatma S. Ibrahim, "Comparative Study of Segmentation Techniques for Detection of Tumors Based on MRI Brain Images," International Journal of Bioscience, Biochemistry and Bioinformatics vol. 8, no. 1, pp. 1-10, 2018.