A Unified Design of ACO and Skewness based Brain Tumor Segmentation and Classification from MRI Scans

Abstract

Brain tumor is among the major reasons for deaths among cancerous diseases around the world. Medical imaging technologies used to detect brain tumor is very popular these days. However, before time detection is open-ended research and needs to be handled more accurately. Multimodality medical image fusion has emerged with promising results in cancer detection. In this paper, a hybrid technique for extracting tumors using MRI images is presented. This technique consists of five steps, such as de-noising of an image, the extraction of the tumor, feature selection, feature fusion, and classification. Curvelet transformation is implemented in the first step for image de-noising. Then in the second step, Ant Colony Optimization (ACO) is utilized along with the Thresholding method for the extraction of tumors based on MRI scans of the brain. Three distinct kinds of features are extracted depending on texture and shape in the third step. After that, the top 70% features are selected based on the priority approach, and fusion is performed using a concatenation based approach. In the last step, fused features are fed to different classifiers such as SVM. The proposed technique is tested on two datasets named BRATS2013 and private dataset. This new system performed well in comparison to different present systems.