RAAGR2-Net: A brain tumor segmentation network using parallel processing of multiple spatial frames

Mobeen Ur Rehman, Jihyoung Ryu, Imran Fareed Nizami, Kil To Chong

Abstract: Highlights•A novel architecture is proposed for the segmentation of brain tumor regions.•Proposed pre-processing module allows to learn insight features of the tumor region.•An encoder-decoder based segmentation model with optimized modules is introduced.•Modules are designed to minimize the data loss while deep feature extraction.•Depthwise convolution for brain tumor segmentation is explored in this work.•Proposed model helps to prevent location of tumor region during decoding process.