Overview of convolutional neural networks architectures for brain tumor segmentation

Ahmad Al-Shboul (Jordan University of Science and Technology)
Maha Gharibeh (Jordan University of Science and Technology)
Hassan Najadat (Jordan University of Science and Technology)
Mostafa Ali (Jordan University of Science and Technology)
Mwaffaq El-Heis (Jordan University of Science and Technology)

International Journal of Electrical and Computer Engineering
Vol 13, No 4: August 2023

Due to the paramount importance of the medical field in the lives of people, researchers and experts exploited advancements in computer techniques to solve many diagnostic and analytical medical problems. Brain tumor diagnosis is one of the most important computational problems that has been studied and focused on. The brain tumor is determined by segmentation of brain images using many techniques based on magnetic resonance imaging (MRI). Brain tumor segmentation methods have been developed since a long time and are still evolving, but the current trend is to use deep convolutional neural networks (CNNs) due to its many breakthroughs and unprecedented results that have been achieved in various applications and their capacity to learn a hierarchy of progressively complicated characteristics from input without requiring manual feature extraction. Considering these unprecedented results, we present this paper as a brief review for main CNNs architecture types used in brain tumor segmentation. Specifically, we focus on researcher works that used the well-known brain tumor segmentation (BraTS) dataset.