Analisis Performa Model Deep Learning VGG16 dan ResNet dalam Klasifikasi Jenis Tumor Otak

Abstract

Brain tumor is a disease of abnormal cell growth in the brain. Brain tumors can affect various groups of people from adults to children. The use of deep learning, one of which is the Convolutional Neural Network (CNN), is able to help in the medical field to help classify the presence of brain tumors in humans. The deep learning system will capture data in the form of MRI images which will then be processed. In previous studies, the use of the MobileNetV2 model was able to obtain an accuracy of 88.64% with 30 epochs. The use of the MobileNetV2 model is able to run with light computing but the model cannot generalize well the heavier model for larger and more complex datasets. The use of the VGG16 and ResNet models is expected to provide higher accuracy and can be used to read more complex and larger datasets. The process of classifying brain tumors consists of four classes, namely glioma tumors, meningioma tumors, pituitary tumors, no tumors. The final results of the VGG16 model research were able to provide an accuracy of 96.07% and the Resnet model was able to provide an accuracy of 87.52%.