A Deep Learning-Based Feature Extraction Model for Classification Brain Tumor

Abstract

AbstractThe computer-aided diagnostic-based that supports deep learning (DL) algorithms consists of several processing layers, which symbolize data with several stages of the construct. In current years, deep learning has increased speedily in almost all areas, especially in medical imaging, medical image investigation, or bioinformatics. Therefore, deep learning has effectively untouched or enhanced the methods of recognition, calculation, or diagnosis in many medical and health areas such as pathology, brain tumors, lung cancer, stomach, heart, or retina. As we know there are many applications of deep learning, the purpose of this paper is to appraise the most important deep learning perception related to tumor analysis detection and classification. In recent pretrained models, usually, features are taken from last layers that are different for each dataset from natural to plants to medical images. GLCM feature extraction and ResNet-50 techniques are used for feature extraction and support vector machine (SVM) are used for brain tumor detection and classification to overcome this difficulty in the proposed method. A practical and efficient deep learning model is proposed, in which backpropagation neural network feature is used to predict brain stroke through CT/MRI scan images. The performance and accuracy of proposed model are evaluated and compared with preexisting models, and checked whether it produces high sensitivity, specificity, precision, and accuracy.KeywordsFeature extractionDeep learningMagnetic character imagingGLCMSupport vector machineBrain tumor classificationPretrained model