Article Preview
Top1. Introduction
The brain tumor is the deadliest disease that affects and ruins several lives in the world. The brain plays a prominent role in every human being in terms of controlling and handling any activity performed by the entire body. The activities of the whole body part are controlled by an organ called the brain. The tumor has been stated as very dangerous as there will be an uncontrolled generation of cancerous cells. The brain tumor is said to be unchecked growth of these cancerous cells in the brain. These dangerous tumors are having the ability to destroy the cells in the brain and damage the cells by increasing stress within the skull and producing inflammation. The MR Images produced top-quality anatomical structures of the human body particularly the brain. Correct and fully automatic human brain classification is significant in clinical studies.
Normally brain tumor has two kinds: MBT - malignant brain tumor and BBT - benign brain tumor. Figure 1 shows the brain MRI images of normal brain (a), brain with benign tumor (b), and the brain with malignant tumor (c). The BBT is having a uniformity of structure, however, does not have any cancer cells, whereas, in the case of MBT, it is having a heterogeneous structure and has made of cancer cells. For this, some of the instances of BBT are meningiomas & gliomas, while some of the instances of MBT are astrocytomas & glioblastoma (Huang et al., 2014). The above-mentioned tumors are life-threatening since they have the potential to damage brain cells and therefore the cells will be destroyed by generating inflammation and stress will be increased in the skull. The techniques like image processing have been utilized widely for extracting information from the images. One of the techniques is Image segmentation which will be explained as a procedure of segmenting the image into regions that are non-intersecting as in contributions (Gholami et al., 2013; Islam et al., 2013; Zhu & Yan, 1997).
Figure 1.
Input MRI images: (a) Brain without tumor, (b) Brain with benign tumor, (c) Brain with malignant tumor
Moreover, it will not generate any damage in healthy areas of the brain with their radiation at the time of the detection. The main contribution of our proposed work is identifying and classifying the benign type and malign types of brain tumors with hybrid optimized classification techniques such as Grey Wolf Optimization (GWO) and Recurrent Neural Network (RNN). For image preprocessing, the High pass filter is used. Here, it will get the MRI image as input, and then it tries to remove the background structures which are not needed, meanwhile it sharpens the essential portions. After preprocessing the optimized method, Modified Gray Level Co-occurrence Matrix (MGLCM-GWO) is applied for feature extraction to take away the imperative features from the tumor image. Hybrid methods, Recurrent Neural Network(RNN) and Grey wolf Optimization(GWO) is applied for the classification of tumors into the harmless and harmful tumor from the original image.
Top2. Literature Survey
Several models have been projected by various researchers in terms of disease prediction. Here, a review of some of the prominent contributions or research of some techniques used in the disease prediction field is depicted in this segment. For identifying the infected area of the tumor from medical-image, the strategy called segmentation had employed or considered in the works of E. Mbuyamba et.al (2017) and M. Sasidhar et.al (2011) developed top-quality anatomical structure images of the human body produced by the MRI imaging technique and it gives crucial data for medical purposes. Depending on characteristics such as brightness, boundaries, contrast, grey levels, color, and texture in an image, this image will be divided into various regions. A. Demirhan et.al (2015) and A. M. Vagu et.al (2015) presented the classification of brain tumors. In this research, tumor cells are categorized from the normal brain by the support of MR (Magnetic resonance) images or any relevant medical sense in their work.