Brain tumor diagnosis systems based on artificial neural networks and segmentation using MRI

Automatic defects detection in Magnetic Resonance Images (MRI) is a crucial factor in several diagnostic applications. This paper presents an intelligent Neural Networks (NN) and segmentation-based system to automatically detect and classify various brain tumors types that might be depicted in MRI. The proposed intelligent system is divided into two main parts: the first part is composed of hybrid neural networks composed of the Principal Component Analysis (PCA) for dimensionality reduction to extract the global features of the MRI cases. The second part is based on the segmentation of the MRI cases using the Wavelet Multiresolution Expectation Maximization (WMEM) algorithm to extract the local features of the cases. Then Multi-Layer Perceptron (MLP) is applied to classify the extracted features from either the first part or from the segmentation process. A comparison study between the performances of MLP when one accomplished the two approaches. The purpose of this research is to save the radiologist time, increases accuracy, and so helps non-experts doctors in diagnosing brain tumors.