Brain Tumor Identification by Hybrid CNN-SWT Model.

Abstract

Detecting brain tumor using the segmentation technique is a big challenge for researchers and takes a long time in medical image processing. Magnetic resonance image analysis techniques facilitate the accurate detection of tissues and abnormal tumors in the brain. The size of a brain tumor can vary with the individual and the specifics of the tumor. Radiologists face great difficulty in diagnosing and classifying brain tumors. This paper proposed a hybrid model-based convolutional neural network with a stationary wavelet trans-form named "CNN-SWT" to segment brain tumors using MR brain big data. We utilized 7 layers for classification in the proposed model that include 3 convolutional and 3 ReLU. Firstly, the input MR image is divided into multiple patches, and then the central pixel value of each patch is provided to the CNN-SWT. Secondly, the pre-processing stage is per-formed using the mean filter to remove the noise. Then the convolution neural network-layer approach is utilized to segment brain tumors. After segmentation, robust feature extraction such as information-extraction methods is used for the feature extraction process. Finally, a CNN-based hybrid scheme based on the stationary wavelet transform technique is used to detect tumors using MR brain images. These experiments were obtained using 11500 MR brain images data from the hospital national of oncology. It was proved that the proposed hybrid achieved a high classification accuracy of (98.7 %) as compared with existing methods. The advantage of the hybrid novelty of the model and the ability to detect the tumor area achieved excellent overall performance using different values.