Brain Tumor Classification Using Medial Residual Encoder Layers
This work addresses the problem of accurate and timely diagnosis of brain tumors for physicians, offering a noninvasive alternative to biopsy, but it appears incremental as it builds on existing deep learning approaches.
The authors tackled brain tumor classification from MRI images by proposing a deep learning system with encoder blocks using residual learning, achieving 95.98% accuracy on a dataset of 3064 images, outperforming prior studies.
According to the World Health Organization (WHO), cancer is the second leading cause of death worldwide, responsible for over 9.5 million deaths in 2018 alone. Brain tumors count for one out of every four cancer deaths. Therefore, accurate and timely diagnosis of brain tumors will lead to more effective treatments. Physicians classify brain tumors only with biopsy operation by brain surgery, and after diagnosing the type of tumor, a treatment plan is considered for the patient. Automatic systems based on machine learning algorithms can allow physicians to diagnose brain tumors with noninvasive measures. To date, several image classification approaches have been proposed to aid diagnosis and treatment. For brain tumor classification in this work, we offer a system based on deep learning, containing encoder blocks. These blocks are fed with post-max-pooling features as residual learning. Our approach shows promising results by improving the tumor classification accuracy in Magnetic resonance imaging (MRI) images using a limited medical image dataset. Experimental evaluations of this model on a dataset consisting of 3064 MR images show 95.98% accuracy, which is better than previous studies on this database.