Brain Tumour Detection Using Unsupervised Learning Based Neural Network

The task of MRI (Magnetic resonance Imaging) brain tumor images Classification is difficult due to the variance and complexity of tumors. This paper presents an unsupervised learning based Neural Network technique for the classification of the magnetic resonance human brain images. Brain tumour diagnosis requires a detailed histological analysis, which involves invasive surgery that can be painful and can cause discomfort to patients. In this paper, the brain tumour diagnostic procedure is divided into the following phases. The first phase comprises of image pre-processing which includes histogram equalization, edge detection, noise filtering, thresholding etc. In second phase, the features of the MR brain image are extracted using Independent Component Analysis (ICA). In third phase, brain tumour diagnosis is performed using Self Organized Map (SOM). Finally, a kmeans clustering algorithm is applied to segment the brain into different tissues. Classification results on a variety of MR images for different pathologies indicate this technique to be promising.