Abstract:
In this study a functional models of Artificial Neural Networks (ANNs) is proposed to aid existing diagnosis methods. ANNs are currently a “hot” research area in medicine, particularly in the fields of radiology, cardiology, and oncology. In this paper an attempt was made to make use of ANNs in the medical field. Hence a Computer Aided Diagnosis (CAD) system using ANNs to classify brain tumors was developed in order to detect and classify the presence of brain tumors according to Magnetic Resonance (MR) Image, and then determined which type of ANNs and activation function for ANNs is the best for image recognition. Also the study aimed to introduce a practical application study for brain tumor diagnosis. Neural network must be able to determine the state of the brain according to MR image. In all procedures, image processing and ANNs design, MATLAB was incleded. From each MR Image a Harlick texture features was extracted to prepare training data which was introduced to neural network as input and target vectors. ANNs was designed using MATLAB tool "nntool". Results obtained explain Elman Network, with log sigmoid activation function, surpassing other ANNs with a performance ratio of 88.24%.
