Brain Tumor Detection Using Shape features and Machine Learning Algorithms [ READ ]
Dena Nadir George, Hashem B. Jehlol, Anwer Subhi Abdulhussein Oleiwi
One of the common methods used to detect tumor in the brain is Magnetic Resonance Imaging (MRI). It gives important information used in the process of scanning the internal structure of the human body in detail. The MR Images classification is not easy task because of the variation and complexity of brain tumors. In the proposed technique, the detecting a brain tumor in the MR Images includes a number of steps are sigma filtering, adaptive threshold and detection region. Numbers of shape features are considered consists Major Axis Length, Euler Number, Minor Axis Length, Solidity, Area and Circularity to extract features for MR Images. The proposed method uses two classifiers depend on supervised techniques; the first classifier was C4.5 decision tree algorithm and the second classifier Multi-Layer Perceptron (MLP) algorithm. The classifiers are used for the purpose of classification the brain case to the normal or abnormal; the abnormal brain is classified into one type of benign tumor and five type of malignant tumor. Maximum precision of about 95% is achieved by considering 174 samples of brain MR Images and using MLP algorithm.
Dena Nadir George, Hashem B. Jehlol, Anwer Subhi Abdulhussein Oleiwi
One of the common methods used to detect tumor in the brain is Magnetic Resonance Imaging (MRI). It gives important information used in the process of scanning the internal structure of the human body in detail. The MR Images classification is not easy task because of the variation and complexity of brain tumors. In the proposed technique, the detecting a brain tumor in the MR Images includes a number of steps are sigma filtering, adaptive threshold and detection region. Numbers of shape features are considered consists Major Axis Length, Euler Number, Minor Axis Length, Solidity, Area and Circularity to extract features for MR Images. The proposed method uses two classifiers depend on supervised techniques; the first classifier was C4.5 decision tree algorithm and the second classifier Multi-Layer Perceptron (MLP) algorithm. The classifiers are used for the purpose of classification the brain case to the normal or abnormal; the abnormal brain is classified into one type of benign tumor and five type of malignant tumor. Maximum precision of about 95% is achieved by considering 174 samples of brain MR Images and using MLP algorithm.
