Classification and Contour Detection of Brain Tumors on Magnetic Resonance Imaging (MRI) Images Using Machine Learning Algorithms with Convolutional Neural Network and Computer Vision Methods

Abstract

The objective of this study is to improve the accuracy of brain tumor diagnosis and segmentation, with the aim of aiding physicians in identifying specific brain tumors. Brain tumors are solid neoplasms within the skull. These tumors are caused by uncontrolled growth of abnormal cells. Classification of brain tumors is divided based on the location of the tumor, the type of tissue produced, and whether the tumor is malignant (malignant) or benign (benign) and several other considerations. In addition, a surgical biopsy of the suspected tissue (tumor) is required to obtain more information about the type of tumor. Biopsy takes 10 to 15 days for laboratory testing. With the advantages of machine learning algorithms, especially the CNN method in classifying brain tumors and contour detection, this research was conducted on the performance of machine learning on MRI image results for patients with brain tumor using transfer learning EfficientNet-B7 and U-Net.