Brain Cancer Medical Diagnostic System Using Grey Scale Features and Support Vector Machine


Automated segmentation and the classification of brain cancer based on Magnetic Resonance Imaging (MRI) is a significant medical development of the last twenty years. Based on computer systems, there are several techniques developed for diagnosis, but the automated diagnosis of cancer type is still a challenge. In this research, a cancer detection system has been proposed and tested to virtually segment the tumor and classify it based on the MRI images. To implement this, a k-mean clustering method is used in the segmentation step. In the features extraction step, each greyscale, symmetrical, and texture features are used. Then, a Principle Component Analysis (PCA) is used to minimize the number of features and Support Vector Machines (SVM) is applied to classify them. To implement the proposed methodology, a computer system was designed and simulated. A database of images was utilized to evaluate how the system is performing under testing. Finally, the test results of the experiments showed the effectiveness of the techniques used to segment and classify tumors.