Astrocytoma Type of Brain Tumor Classification using Artificial Neural Network

Brain tumor is a complex disease and its early detection and classification is very challenging. Astrocytoma is one of the commonest type of brain tumors. The detection and classification system of this type of tumor is based on magnetic resonance images (MRI). Brain tumors are classified into four grades (grade I – IV) according to world health organization (WHO) classification. The grade-I tumors are less malignant and grade-IV tumors are highly malignant. The classification based on only imaging findings is useful but occasionally equivocal in some cases which do not have typical characteristics of specific grade of tumor. In this work, image processing algorithm is developed to segment the tumor from the surrounding soft tissues and then classify the tumor into different grades. The work is carried out with MRI of different patients with astrocytoma type of brain tumors. The work involves two phases, namely learning/training phase and recognition/testing phase. In learning/training phase the artificial neural network (ANN) is trained for recognition of different astrocytoma types of brain tumor. The texture features are extracted from MRI whose tumor grades are known using gray level co-occurrence matrix (GLCM) and Gabor filters. These features are saved in knowledge base and are used to train the neural network. The brain MR images whose diagnosis is unknown are used for testing in recognition/testing phase. Tumor segmentation is achieved on these test images using watershed segmentation algorithm. The texture features in the detected tumor region are extracted. These features are compared with the stored features in the knowledge base. Finally a neural network classifier has been developed to recognize different grades of brain tumors.