MAD-Net: Multi-attention dense network for functional bone marrow segmentation

• Published in: Computers in biology and medicine 2023-03, Vol.154, p.106428, Article 106428
• Main Authors: Qin, Chuanbo, Zheng, Bin, Li, Wanying, Chen, Hongbo, Zeng, Junying, Wu, Chenwang, Liang, Shufen, Luo, Jun, Zhou, Shuquan, Xiao, Lin
• Format: Article
• Language: English
• Subjects: Ablation; Benchmarking; Bone marrow; Bone Marrow - diagnostic imaging; Cancer therapies; Cervical cancer; Chemotherapy; Contouring; Datasets; Deep learning; Dense convolution; Female; Functional bone marrow segmentation; Gradient flow; Humans; Image Processing, Computer-Assisted; Labor, Obstetric; Magnetic resonance imaging; Malignancy; Medical imaging; Methods; Modules; Pelvis; Pregnancy; Radiation; Radiation damage; Radiation therapy; Residual-dual attention; Segmentation; Semantics; Slide-window attention; Toxicity
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/j.compbiomed.2022.106428

Radiotherapy is the main treatment modality for various pelvic malignancies. However, high intensity radiation can damage the functional bone marrow (FBM), resulting in hematological toxicity (HT). Accurate identification and protection of the FBM during radiotherapy planning can reduce pelvic HT. The traditional manual method for contouring the FBM is time-consuming and laborious. Therefore, development of an efficient and accurate automatic segmentation mode can provide a distinct leverage in clinical settings. In this paper, we propose the first network for performing the FBM segmentation task, which is referred to as the multi-attention dense network (named MAD-Net). Primarily, we introduce the dense convolution block to promote the gradient flow in the network as well as incite feature reuse. Next, a novel slide-window attention module is proposed to emphasize long-range dependencies and exploit interdependencies between features. Finally, we design a residual-dual attention module as the bottleneck layer, which further aggregates useful spatial details and explores intra-class responsiveness of high-level features. In this work, we conduct extensive experiments on our dataset of 3838 two-dimensional pelvic slices. Experimental results demonstrate that the proposed MAD-Net transcends previous state-of-the-art models in various metrics. In addition, the contributions of the proposed components are verified by ablation analysis, and we conduct experiments on three other datasets to manifest the generalizability of MAD-Net. •We propose MAD-Net, the first model to segment the functional bone marrow from pelvic CT images.•The slide-window attention module is designed to explore interdependencies between features.•The residual-dual attention module is employed to enhance high-level feature representations.•MAD-Net surpasses to state-of-the-art models in segmentation accuracy, while retaining comparable training and inference times.