Jiang Yi, Wang Haoyu, Boergens Kevin M, Rzepka Norman, Wang Fangfang, Hua Yunfeng
Shanghai Institute of Precision Medicine, Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, 200125 Shanghai, China.
Shanghai Institute of Precision Medicine, Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, 200125 Shanghai, China.
Cell Rep Methods. 2025 Feb 24;5(2):100989. doi: 10.1016/j.crmeth.2025.100989.
Recent technical advances in volume electron microscopy (vEM) and artificial-intelligence-assisted image processing have facilitated high-throughput quantifications of cellular structures, such as mitochondria, that are ubiquitous and morphologically diversified. A still often-overlooked computational challenge is to assign a cell identity to numerous mitochondrial instances, for which both mitochondrial and cell membrane contouring used to be required. Here, we present a vEM reconstruction procedure (called mito-SegEM) that utilizes virtual-path-based annotation to assign automatically segmented mitochondrial instances at the cellular scale, therefore bypassing the requirement of membrane contouring. The embedded toolset in webKnossos (an open-source online annotation platform) is optimized for fast annotation, visualization, and proofreading of cellular organelle networks. We demonstrate the broad applications of mito-SegEM on volumetric datasets from various tissues, including the brain, intestine, and testis, to achieve an accurate and efficient reconstruction of mitochondria in a use-dependent fashion.
近期，体积电子显微镜（vEM）和人工智能辅助图像处理技术的进展推动了对细胞结构（如线粒体）的高通量定量分析，线粒体普遍存在且形态多样。一个常常被忽视的计算挑战是为众多线粒体实例赋予细胞身份，过去这需要同时勾勒线粒体和细胞膜轮廓。在此，我们提出一种vEM重建程序（称为mito - SegEM），该程序利用基于虚拟路径的注释在细胞尺度上自动分配分割后的线粒体实例，从而无需进行膜轮廓勾勒。webKnossos（一个开源在线注释平台）中嵌入的工具集针对细胞器网络的快速注释、可视化和校对进行了优化。我们展示了mito - SegEM在来自各种组织（包括脑、肠和睾丸）的体积数据集上的广泛应用，以依赖于使用方式的方法实现线粒体的准确高效重建。
