A Brain MRI Segmentation Method Using Feature Weighting and a Combination of Efficient Visual Features

Determining the area of brain tumors is an essential and fundamental step in automatic diagnosis and treatment systems. The authors present a method based on a combination of efficient visual features and fuzzy c-means clustering to detect brain tumors. For this purpose, first, the background area of the images is removed by the new thresholding method, then the useful and efficient features are extracted. The authors use this new feature space for clustering-based segmentation. The proposed clustering algorithm gives a different importance to the extracted features in the segmentation process, which leads to better detection of the tumor region. Finally, to remove some curved edges of the brain and the border between the background and the skull which are wrongly clustered as tumors, the mode filter is used. The proposed approach is tested using a dataset of magnetic resonance images. On all testing images, the suggested approach’s accuracy and F-score metrics rates are an average of 96.25 and 97.96%, respectively. © 2024 selection and editorial matter, Swati V. Shinde, Darshan V. Medhane and Oscar Castillo; individual chapters, the contributors.