3D asymmetric expectation‐maximization attention network for brain tumor segmentation

• Published in: NMR in biomedicine 2022-05, Vol.35 (5), p.e4657-n/a
• Main Authors: Zhang, Jianxin, Jiang, Zongkang, Liu, Dongwei, Sun, Qiule, Hou, Yaqing, Liu, Bin
• Format: Article
• Language: English
• Subjects: 3D U‐Net; Artificial neural networks; asymmetric convolution block; Asymmetry; Attention; Benchmarks; Biological products; Brain; Brain cancer; Brain Neoplasms - diagnostic imaging; Brain Neoplasms - pathology; brain tumor segmentation; Brain tumors; Coders; Computation; Embedding; expectation‐Maximization attention; Humans; Image processing; Image Processing, Computer-Assisted - methods; Image segmentation; light‐weight network; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Maximization; Motivation; MRI; Neural networks; Neural Networks, Computer; Optimization; Tumors; Two dimensional models
• ISSN: 0952-3480; 1099-1492
• DOI: 10.1002/nbm.4657

Automatic brain tumor segmentation on MRI is a prerequisite to provide a quantitative and intuitive assistance for clinical diagnosis and treatment. Meanwhile, 3D deep neural network related brain tumor segmentation models have demonstrated considerable accuracy improvement over corresponding 2D methodologies. However, 3D brain tumor segmentation models generally suffer from high computation cost. Motivated by a recently proposed 3D dilated multi‐fiber network (DMF‐Net) architecture that pays more attention to reduction of computation cost, we present in this work a novel encoder‐decoder neural network, ie a 3D asymmetric expectation‐maximization attention network (AEMA‐Net), to automatically segment brain tumors. We modify DMF‐Net by introducing an asymmetric convolution block into a multi‐fiber unit and a dilated multi‐fiber unit to capture more powerful deep features for the brain tumor segmentation. In addition, AEMA‐Net further incorporates an expectation‐maximization attention (EMA) module into the DMF‐Net by embedding the EMA block in the third stage of skip connection, which focuses on capturing the long‐range dependence of context. We extensively evaluate AEMA‐Net on three MRI brain tumor segmentation benchmarks of BraTS 2018, 2019 and 2020 datasets. Experimental results demonstrate that AEMA‐Net outperforms both 3D U‐Net and DMF‐Net, and it achieves competitive performance compared with the state‐of‐the‐art brain tumor segmentation methods. A novel encoder‐decoder neural network called AEMA‐Net is proposed for MRI brain tumor segmentation. AEMA‐Net first introduces 3D asymmetric multi‐fiber and asymmetric dilated multi‐fiber units to improve the deep feature representation of brain tumors during the encoder step. In addition, it embeds an expectation‐maximization attention (EMA) module on the third skip connection of the baseline network, successfully capturing the long‐range contextual dependence of brain tumor images.