International Journal of Computer Science Trends and Technology (IJCST) – Volume 9 Issue 3, May-Jun 2021 ISSN: 2347-8578 www.ijcstjournal.org Page 67 Brain Tumour Detection and Classification Using SVM S.S. Dharun Raj [1] , S. Hariharan [2] Student, Department of Software Engineering, SRM Institute of Science and Technology - India ABSTRACT The brain is one of the most complicated organs in the human body, with billions of cells working together. When cells divide uncontrollably, they form an abnormal group of cells around or inside the brain, which is known as a cerebral tumour. This cell group has the ability to disrupt brain activity and kill healthy cells. Brain tumours are graded into benign or low-grade (grades 1 and 2) and malignant or high-grade (grades 3 and 4). The proposed method is intended to distinguish between normal and brain tumour (benign or malign). Brain magnetic resonance imaging (MRI) is used to study certain forms of brain tumours, such as metastatic bronchogenic carcinoma tumours, glioblastoma, and sarcoma. Different wavelet transforms and support vector machines are used in the identification and classification of MRI brain tumours. Manually detecting a brain tumour by doctors is a complex and time-consuming operation. To prevent misclassification and save time, brain tumour identification and classification could be performed automatically. Keywords: Brain tumor, SVM classification, otsu method. I. INTRODUCTION Cancer is becoming a more serious health issue as the world's population grows. According to statistics, the population of cancerous people in India is about 12.7 million every year, with 7.6 million people dying as a result of cancer [1]. Most normal cells die as they age or become damaged, and new cells replace them. This procedure may sometimes go wrong. When the body doesn't need new cells, they form, and old or damaged cells don't die as they should. A growth or tumour is a mass of tissue formed by the accumulation of extra cells. There are two types of primary brain tumours: benign and malignant. Benign brain tumours do not contain cancer cells. Benign tumours can usually be removed, and they seldom grow back. Benign brain tumours typically have a distinct edge or border. Benign tumour cells rarely enter the tissues around them. They are not infectious and do not spread to other areas of the body. Cancer cells are present in malignant brain tumours (also known as brain cancer). Malignant brain tumours are more dangerous and also pose a life-threatening threat. They are likely to spread quickly and crowd or invade healthy brain tissue nearby. Malignant brain tumours may cause cancer cells to break free and spread to other parts of the brain or the spinal cord. They only spread to a few other areas of the body in exceptional cases. Basic block diagram of brain tumour classification is as shown in figure 1. The basic block diagram consists of four modules 1. Image pre-processing 2. Image Segmentation 3. Feature extraction 4. Classification The obtained images of medical imaging are very noisy because of the physical process of imaging. The presence of noise will cause the images to be misclassified, lowering the classifier's output. Image pre-processing is a method of enhancing an image using a filtering technique.The quantitative measurement of the images is called feature extraction. Image data is transformed into a statistical numeric value during feature extraction. Contrast, homogeneity, correlation, energy, and entropy are some of the features that can be extracted from an image. The classifier analyses the characteristics of the input data and categorises the images accordingly. Support vector machine (SVM), k-nearest neighbour KNN, artificial neural network (ANN), Hidden Markov Model (HMM), and Probabilistic Neural Network (PNN) are some of the examples of learning classifiers. Every classifier has its own set of advantages and disadvantages. ANN is fast and reliable, but it has a high computational cost, so it uses a lot of the CPU's primary physical memory. SVM outperforms other algorithms in terms of accuracy [1]- [[20]. II. RELATED WORKS In [21]-[25] presented a method for automatically classifying medical images. The KNN classifier is used to separate medical images into two categories: normal and abnormal. KNN is a straightforward approach with a low computational cost. In [26]-[30] suggested a thesis on the brain tumour prediction algorithm and its position in the brain. ANN is a statistical problem inspired by the biological nervous system. The GLCM technique was used to isolate a function, and the extracted features were then identified using an artificial neural network [31]- [40]. A SVM classifier-based MRI image classification technique was proposed. Support Vector-based advanced classification techniques. A supervised learning algorithm RESEARCH ARTICLE OPEN ACCESS