Brain tumor is a hazardous disease. It has to be treated rightly because the patient cannot survive for even a year. This research work is to design a mechanized system that discerns benign and malignant tumor images and improves the classification accuracy. In this system, histogram equalization is used to raise the intensity of tumor so that it can be detected more precisely. In the projected system, segmentation is performed by k-means clustering and Gaussian mixture model (GMM). Along with them, the authors are extracting the features from discrete wavelet transform (DWT) and feature reduction from principal component analysis (PCA). The classifiers support vector machine (SVM) and artificial neural networks (ANN) are used to classify benign and malignant tumor from brain images.
TopI. Introduction
Signal Processing is a vast area of research consisting of various fields, one among them are, the Digital Image Processing (DIP) (Gonzalez, R. C., et.al. 2002,) which allow us to play with components of images as required in desired application. DIP has vast area of research and is used in various fields such as medical imaging, satellite images of planets, and also many industrial applications. Among all these applications, medical field mostly depends on images such as MRI, X-rays, Ultrasound, CT scan and other bio-medical images to identify the exact problem in the patient’s body. These images give the detail study of various diseases such as brain tumor, cancer, swelling, etc. So for the physicians to treat and diagnose the problem in a better way needs the images to be in good quality giving all the necessary information about the infected body part (Devasena, C. L., & Hemalatha, M. 2013). Using image enhancement techniques to improve the images visual quality help us to make better localization of pixels present in the input image which will then result in good contrast images. MRI images are low contrast images. Various methods of image enhancement help to improve the brightness and contrast of the image for practitioners to analyze and treat the infected area, (Amien, M. B., et.al., 2013).
After enhancement of images and identifying the tumor region, it is necessary to define the grade of the tumor. Brain tumor can be classified into two grades i.e. low grade and high grade and also four stages. In this paper we will first segment the tumor using K-means algorithm, this method will help us to extract the infected area from the body part (Ramaswamy Reddy, et.al., 2013).
The two grades of tumor can be expressed in a well defined manner. The low grade tumor is called “Benign” and the high grade tumor as “Malignant”. The benign tumor is a low grade disease which does not spread over the body. Although they can be life threatening. In this the tumor cells do not grow and remain confined to a particular area but starts destroying the normal cells and tissues of that part of the brain. Therefore this grade is also a serious issue for the patient’s health. On the other hand if we talk about the high grade tumors, they spread in the body part with time. These are very dangerous for patient’s health and should be treated as early as possible. They spread in the brain exponentially replacing the normal cells with the infected ones by killing the tissues and veins of the brain, which eventually result in the brain to die slowly.
Cancer is an anomalous cell expansion that has the possibility of invading to remaining organs. It involves a group of diseases.Cancer is prominent factor of death worldwide. Brain tumor is a collection of anomalous cells in brain. Brain is very rigid and encircled with skull any growth inside this restricted place can lead to problems. When these tumors grow inside the brain it raises intra cranial pressure, which can cause brain damage and may also cause throttle of life.
The benign tumors are those which do not invade to other parts of the body. Benign brain tumors are normally characterized as a class of identical cells that grow patterns and do not result normal cell partition and grow into a group of cells that do not have the peculiar appearance of a cancer under the microscope. The brain tumors at the most benign are identified by MRI brain scans and CT scans. These kinds of tumors normally grow slowly and does not spread into nearby cells or invade to remaining parts of body. These tumors hardly evolve as metastatic i.e., cancerous tumors. Mostly the brain tumors which are benign can be removed and normally these benign tumors do not reoccur after removal. Malignant tumor is formed of tumor tissues, and can spread to neighboring tissues. The process in which the tumor tissues can invade into blood flow or lymph cells and spread to the corresponding cells present in the body is called metasis.
Figure 1.
Benign tumorMalignant tumor
Based on level of hardness, different grades of tumor have been present. They are
- •
Grade-1 (Pilocytic astrocytoma): least dangerous tumor
- •
Grade-2 (Low grade astrocytoma): Grade 2 tumor grows normally but casts as anomalous when we view in a microscopic
- •
Grade-3 (Anaplastic astrocytoma): This type of tumor is malignant but there is no much difference between grade 2 and grade 3 types.
- •
Grade-4 (Glioblastoma): Grade 4 is complete malignant and dangerous tumor.