Detection and classification of glioma, meningioma, pituitary tumor, and normal in brain magnetic resonance imaging using deep learning-based hybrid model

Abstract

Pituitary, meningioma, and glioma tumors are the primary widespread brain tumors. Treatment algorithms for these tumors may differ from each other. For this reason, the detection and typing of brain tumors are necessary to determine the appropriate treatment quickly. Since imaging findings vary, even experts often have difficulties in classifying brain tumors. In this study, a convolutional neural network-based hybrid model is developed to classify glioma, meningioma, pituitary tumors, and normal brain magnetic resonance images. In the proposed hybrid model, features are obtained using pre-trained Efficientnetb0 and Shufflenet architectures. Also, the images in the existing dataset are improved, such as colored and used as the second database. With the proposed two pre-trained models, the features of individual images from two datasets are extracted. Later, these features are concatenated, and the best of these features are selected using the mRMR feature reduction method and classified using the Support Vector Machine (SVM) classifier. The proposed method achieved better results than the previously trained Efficientnetb0 and Shufflenet architectures. The accuracy value of the proposed hybrid model is 95.4%. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023.