AUTOMATIC DETECTION AND SEGMENTATION OF BRAIN TUMORS USING CONVOLUTIONAL NEURAL NETWORK

Abstract

In many clinical applications, Brain MRI is an essential task as the result of the entire diagnosis is
influenced by MRI segmentation. Magnetic Resonance Imaging (MRI) has made enormous progress in
retrieving brain damage and inspecting brain analysis. The advancements in brain MRI have allowed
huge amount of datasets that are extremely good in quality. These kinds of huge and compound MRI
dataset analysis have become a tiresome task for clinicians, one who physically takeout the essential
information from the datasets. Inventions in computerized methods have been made to overcome the brain
MRI data analysis difficulties. High tissue contrast and fewer artifacts are found in MRI images. The
tumor detection demands diverse actions on MRI which includes image pre-processing, image
classification and segmentation. In our proposed work image processing methods gray scale conversion
is implemented. In this, whether a person is diseased or not is concluded by the final classification
process using neural network algorithm like Convolutional Neural Network (CNN) algorithm. The
proposed CNN can be outperforms than the existing machine learning algorithms. The identification
process of two-dimensional image information is carried out by a special design such as Convolution
Neural Network algorithm. CNN is a multilayer perceptron as it contains three different kinds of layers
like Convolution layer, Fully connected layer, Pooling layer through which the diseases can be classified
accurately. Keywords--Classification, Segmentation, Pre-processing, Conversion, Convolutional Neural
Network.