Author : Dhanashri Joshi and Prof.H.P.Channe

Magnetic Resonance Image (MRI) gives an internal structure body. MRI is widely used for brain tumor detection. A brain tumor can have different shapes and can occur at different locations inside the brain. Tumors are of mainly two types of benign and malignant. Cancerous tumors are called malignant tumors, which can spread in surrounding tissues. Non-cancerous tumors are called benign tumors, which can be removed surgically. Traditionally brain tumor detection is done by radiologists, which is time-critical and depends upon the availability of skilled radiologists. This paper presents a brain tumor detection method using image processing and machine learning techniques. The proposed method follows preprocessing, segmentation, feature extraction, and classification stages. GLCM texture features and Local Binary Pattern (LBP) features are taken into consideration for classification. This approach is evaluated on publicly available MRI datasets using SVM and Random Forest with accuracy 92% and 93%, respectively.

Keywords: Brain tumor detection; Structural MRI; Image processing; Machine learning; k-means