Peningkatan Hiperparameter Framework Deep Learning VGG-16 untuk Pendeteksian Tumor Otak pada Teknologi MRI

Abstract

Detection of human brain tumors through medical images still has limitations, so an accurate method is needed. This study aims to improve the ability of the modified VGG-16 model through hyperparameter adjustment in detecting human brain tumor MRI images. The dataset used comes from the Brain MRI Tumor Dataset on Kaggle, with four categories of brain tumors. The VGG-16 model was adjusted to improve accuracy, adjust brightness and contrast in data augmentation, and add a classification layer. Hyperparameters set include learning rate, batch size, epoch, and optimizer. The results showed an accuracy of 95.63%, precision 95.69%, recall 95.58%, and F1 score 95.57%. The applied model shows potential in improving the accuracy and efficiency of brain tumor diagnosis using MRI technology. Thus, the modification of the VGG-16 model in this study provides improved performance in brain tumor MRI image detection compared to previous studies.