Brain tumor segmentation and classification by improved binomial thresholding and multi-features selection

A malignant tumor in brain is detected using images from Magnetic Resonance scanners. Malignancy detection in brain and separation of its tissues from normal brain cells allows to correctly localizing abnormal tissues in brain’s Magnetic Resonance Imaging (MRI). In this article, a new method is proposed for the segmentation and classification of brain tumor based on improved saliency segmentation and best features selection approach. The presented method works in four pipe line procedures such as tumor preprocessing, tumor segmentation, feature extraction and classification. In the first step, preprocessing is performed to extract the region of interest (ROI) using manual skull stripping and noise effects are removed by Gaussian filter. Then tumor is segmented in the second step by improved thresholding method which is implemented by binomial mean, variance and standard deviation. In the third step, geometric and four texture features are extracted. The extracted features are fused by a serial based method and best features are selected using Genetic Algorithm (GA). Finally, support vector machine (SVM) of linear kernel function is utilized for the classification of selected features. The proposed method is tested on two data sets including Harvard and Private. The Private data set is collected from Nishtar Hospital Multan, Pakistan. The proposed method achieved average classification accuracy of above 90% for both data sets which shows its authenticity. © Springer-Verlag GmbH Germany, part of Springer Nature 2018.