IMPROVING BRAIN TUMOR DETECTION WITH DEEP LEARNING MODELS: A COMPARATIVE ANALYSIS USING MRI AND RO-SVM CLASSIFICATION

Abstract

Brain tumors are one of the most important health problems that threaten human life. Therefore, accurate diagnosis at an early stage is vital. Magnetic resonance imaging (MRI) is one of the most effective and important methods for detecting brain tumors. It is thought that instead of disease detection using traditional methods, artificial intelligence-based computer applications can make significant contributions to experts in detecting brain tumors. Deep learning (DL) models, which are very popular in the scientific world today, are used extensively in the processing of images obtained in the field of health and in the detection of diseases. In this study, VGG-19, ResNet-101 and DenseNet-121 DS models trained on the ImageNet dataset were used to detect brain tumors with magnetic resonance (MR) images. The MR images used in the study were pre-processed and the excess images were identified and cropped to obtain high efficiency. After the image ratios were adjusted, random forest (RF) and support vector machines (SVM) classification algorithms were used. In the experimental studies, the highest accuracy values were achieved with DenseNet-121 using RF and SVM classification algorithms in detecting brain tumors with the proposed method. Results of 87.67% were obtained with SVM and 85.83% with RO.