CNN-based Approach for Enhancing Brain Tumor Image Classification Accuracy

Abstract

Brain tumors are one of the deadliest diseases in the world. This disease can attack anyone regardless of gender or certain age groups. The diagnosis of brain tumors is carried out by manually identifying images resulting from Computerized Tomography Scan or Magnetic Resonance Imaging, making it possible for diagnostic errors to occur. In addition, diagnosis can be made using biopsy techniques. This technique is very accurate but takes a long time, around 10 to 15 days and involves a lot of equipment and medical personnel. Based on this, machine learning technology is needed which can classify based on images produced from MRI. This research aims to increase the accuracy of previous research in the classification of brain tumors so that errors do not occur in the diagnosis of brain tumors. The method used in this research is Convolutional Neural Network using the AlexNet and Google Net architectures. The results of this research obtained an accuracy of 98% for the AlexNet architecture and 96% for GoogleNet. This result is higher when compared with previous research. This finding can reduce the computational burden during model training. The results of this research can help physicians diagnose brain tumors quickly and accurately.