ARM-Net: Attention-guided residual multiscale CNN for multiclass brain tumor classification using MR images

Tapas Kumar Dutta, Deepak Ranjan Nayak, Yu-Dong Zhang

Abstract: Highlights•An attention-based residual multiscale CNN ‘ARM-Net’ is proposed.•A lightweight global attention module is proposed to learn class-specific features.•A comparative analysis with contemporary attention mechanisms is performed.•Results indicate the effectiveness of ARM-Net over existing methods.•The proposed ARM-Net model is lightweight and end-to-end learnable.