Medial Residual Encoder Layers for Classification of Brain Tumors in Magnetic Resonance Images

Correct and timely diagnosis of the brain tumor will make treatment more effective. To date, several image classification approaches have been proposed to aid diagnosis and treatment. In this work, we developed a deep learning approach to increase tumor type classification accuracy in MRI images by considering medical image dataset limitations. Our system is based on deep learning, containing encoder blocks. In addition to more organized and simple architecture, a residual approach is used in the proposed method. Encoder blocks are fed with post-max-pooling features as residual learning. Experimental evaluations of our approach show promising results by improving the tumor classification accuracy in Magnetic resonance imaging (MRI) images using a limited medical image dataset. Experimental assessment of this model on a dataset consisting of 3064 MR images shows 95.98% accuracy, which is better than previous studies on this database.