Multiclass Brain Tumor Classification Using Convolutional Neural Network and Support Vector Machine

Abstract

Brain tumor is caused by abnormal growth of cells inside the skull. It is a lethal disease, the diagnosis of which is a difficult task for radiologists. Most of the tumors are misdiagnosed due to the variability and complexity of lesions, which reduces the survival rate in patients. Diagnosis of brain tumors via computer vision algorithms is a challenging task. Traditional brain tumor identification techniques require manual segmentation or handcrafted feature extraction that is error-prone and time-consuming. This research explores multiclass brain tumor classification methods using Deep Learning (DL) and Machine Learning (ML) techniques. First, the brain MRI images are classified via end-to-end Convolutional Neural Network (CNN) models i.e. ResNet-18 and GoogLeNet. The deep features extracted from the CNN models are also classified using Support Vector Machine (SVM). The proposed method is trained and evaluated on 15,320 MRI images and achieved the highest accuracy of 98% via CNN-SVM based method. Our proposed method outperformed the existing brain tumor identification systems and can assist the doctors to detect brain tumors and make key decisions related to the patient’s treatment.