A comprehensive dataset of magnetic resonance enterography images with bowel segment annotations

Inflammatory bowel disease (IBD) is a kind of recurrent bowel disease and usually requires magnetic resonance enterography (MRE) examinations for diagnosis and monitoring. However, radiologists’ recognition of bowel segments from MRE images is challenging and time-consuming. Deep learning-based medical image segmentation has shown the potential to reduce manual efforts and provide automated tools to assist in the management of disease, but it requires a large-scale fine-annotated dataset for training. To address this gap, we collected MRE data from 114 IBD patients. The bowel images per patient were contoured and annotated as ten segments (stomach, duodenum, small intestine, appendix, cecum, ascending colon, transverse colon, descending colon, sigmoid colon, and rectum), with fine pixel-level annotations labeled by experienced radiologists. Further, we validated the efficiency of several state-of-the-art segmentation methods on this dataset. This work established a high quality, publicly available whole bowel segment MR dataset with benchmark results and laid a groundwork for IBD’s AI research.炎症性肠病（Inflammatory Bowel Disease, IBD）是一类复发性肠道疾病，临床诊断与病情监测通常需借助磁共振小肠造影（Magnetic Resonance Enterography, MRE）检查完成。然而，放射科医师从MRE影像中识别肠道节段的工作既具有挑战性，又耗费大量时间。基于深度学习的医学影像分割技术已展现出减少人工工作量、提供自动化工具以辅助疾病管理的潜力，但此类模型的训练需依赖大规模精细标注的数据集。为填补这一研究空白，本研究收集了114名IBD患者的MRE影像数据。每位患者的肠道影像均由经验丰富的放射科医师完成轮廓勾画与像素级精细标注，共划分为10个肠道节段：胃、十二指肠、小肠、阑尾、盲肠、升结肠、横结肠、降结肠、乙状结肠及直肠。此外，本研究在该数据集上验证了多款当前前沿的影像分割方法的效能。本研究构建了一个高质量、可公开获取的全肠道节段MR数据集，并提供了基准测试结果，为IBD相关人工智能研究奠定了基础。