An In Silico Approach for Brain Tumor Detection and Classification of Magnetic Resonance Images

Abstract

Background: Early detection of cancer can be done using machine learning approaches with high precision. A brain tumor is a very dangerous disease that may cause the death of cancerous patients. Every year, thousands of people die from that disease all over the world. Proper detection of cancerous cells in the body can save their lives. Methods: To segment the brain tumor region through brain MR images and to classify tumorous and normal brain images into different classes is very crucial to cure death-causing diseases like cancer. There are various techniques or methods for segmenting the tumorous part or area from the medical images. Magnetic resonance imaging is the most important technique to capture the images of the body parts because it has more information than any other imaging method, such as a CT scan, etc. K-means clustering is used for the segmentation of the tumor region, and the SVM classifier is used for classification purposes. Results: The classification was done through the support vector machines in MATLAB 2019a. 350 images were classified with an accuracy of 89.7 %. Conclusion: In this paper, MRI images have been used for tumor detection and classification of those images into different classes with the help of MATLAB software. We calculated the accuracy of the classification using machine learning techniques. Early detection of cancerous regions is effective in curing death-causing diseases.