Hapsari Peni Agustin Tjahyaningtijas, Laras Suciningtyas, Andi Kurniawan Nugroho, Cucun Very Angkoso, Lusia Rakhmawati, Naim Rokhmawati
Deep learning is a part of machine learning that has been shown to address challenges in the artificial intelligence field. To express complicated relationships, traditional machine learning methods such as Support Vector Machine (SVM and k-Nearest Neighbor (kNN) require a high number of nodes. In this study, we employ Deep Neural Network (DNN) as a brain tumor classification approach since it can efficiently handle difficult problems without the usage of a large number of nodes like SVM and kNN. This research used MRI scans from the BraTS data set with a total of 112 low grade glioma (LGG) and high grade glioma (HGG). The study was divided into four stages: segmentation applied the U-Net-based segmentation approach, image feature extraction using gray level co-occurrence matrix (GLCM), gray level run length matrix (GLRLM), and gray level size zone matrix (GLSZM) in the second stage, in the third stage we applied the Max-Min normalization method, in the fourth stage we utilized info gain ratio as a scoring method. Finally, the DNN classification approach is compared to the SVM classification method. The classification approach with DNN has a greater accuracy value of 2.75 percent than SVM. © 2022 IEEE.
Universitas Negeri Surabaya, Department of Electrical Enginieering, Surabaya, Indonesia; Universitas Semarang, Department of Electrical Engineering, Semarang, Indonesia; University of Trunojoyo Madura, Informatics Department, Bangkalan, Indonesia; Universitas Negeri Surabaya, Department of Informatics Engineering, Surabaya, Indonesia
