Risk prediction in magnetic resonance imaging brain images using machine learning techniques

Brain diseases is one of the major cause of cancer related death among children and adults in the world. Brain diseases like brain tumor is characterized as a gathering of abnormal cells that becomes inside the brain and around the brain. There are various imaging techniques which are used for brain tumor detection. Among all imaging technique, MRI (Magnetic Resonance maging) is widely used for the brain tumor detection. MRI is safe, fast and non-invasive imaging technique. The early detection of brain diseases is very important, for that CAD (Computer-aided-diagnosis) systems are used. The proposed scheme develops a new CAD system in which pulse-coupled neural network is used for the brain tumor segmentation from MRI images. After segmentation, for feature extraction the Discrete Wavelet Transform and Curvelet Transform are employed separately. Subsequently, both PCA (Principal Componenet Analysis) and LDA (Linear Discriminant Analysis) have been applied individually for feature reduction. A standard dataset of 101 brain MRI images (14 normal and 87 abnormal) is utilized to validate the proposed scheme. The experimental results show that the suggested scheme achieves better result than the state-of-the-art techniques with a very less number of features.