Segmentation of organelles in isotropic electron microscopy data of wild-type, immortalized breast cancer cell - SUM159 (jrc_sum159-1)

The cell interior contains hundreds of different organelle and macromolecular assemblies intricately organized relative to each other to meet any cellular demand. Obtaining a complete understanding of this organization is challenging and requires nanometer-level, three-dimensional reconstruction of whole cells. Even then, the immense size of datasets and large number of structures to be characterized requires generalizable, automatic methods. To overcome this challenge, we developed an analysis pipeline for comprehensively reconstructing and analyzing all known cellular organelles from entire cells imaged by focused ion beam scanning electron microscopy (FIB-SEM) at a near-isotropic size of 4 or 8 nm per voxel. The pipeline involved deep learning architectures trained on diverse samples for automatic reconstruction of 35 different cellular organelle classes - ranging from ER to microtubules to ribosomes - from multiple cell types. Automatic reconstructions were used to directly quantify various previously inaccessible metrics about these structures and their spatial interactions. We show that automatic organelle reconstructions can also be used to automatically register light and electron microscopy images for correlative studies. The data, computer code, and trained models are all shared through an open data and open source web platform OpenOrganelle, enabling scientists everywhere to query and further reconstruct the datasets.Sample: Wild-type SUM-159 cell, treated with 0.5 mM oleic acid for 45 mins prior to high pressure freezing to induce the formation of lipid droplets.Dataset ID: jrc_sum159-1EM Data DOI: 10.25378/janelia.13114352EM voxel size (nm): 4.0 x 4.0 x 4.0 (X, Y, Z)Segmentation voxel size (nm): 4.0 x 4.0 x 4.0 (X, Y, Z)Dataset URL: https://data.janelia.org/LzShCVisualization Website: https://openorganelle.janelia.org/datasets/jrc_sum159-1Publication: Heinrich et el., 2020Included in Dataset: manual annotations细胞内部包含数百种不同的细胞器与大分子复合物，它们彼此间精密排布，以满足各类细胞生理需求。完整解析该组织模式极具挑战，需对完整细胞开展纳米级三维重建。即便如此，数据集的庞大规模与待表征结构的海量数量，仍亟需通用化的自动化分析方法。 为攻克这一难题，我们开发了一套分析流程，可对经聚焦离子束扫描电子显微镜（focused ion beam scanning electron microscopy, FIB-SEM）成像的完整细胞，以各体素近各向同性的4 nm或8 nm分辨率，全面重建并分析所有已知的细胞细胞器。该流程基于在多样本上训练的深度学习架构，可自动重建多种细胞类型中35类不同的细胞细胞器——涵盖内质网（Endoplasmic Reticulum, ER）、微管至核糖体等。 自动重建结果可直接量化此前难以获取的各类结构相关指标及其空间相互作用。我们还证实，自动细胞器重建还可用于自动配准光学显微镜与电子显微镜图像，以开展关联成像研究。 本数据集、计算机代码与训练好的模型均通过开放数据与开源网络平台OpenOrganelle共享，以便全球科研人员查询并进一步重建该数据集。 样本：野生型SUM-159细胞，在高压冷冻前经0.5 mM油酸处理45分钟，以诱导脂滴形成。 数据集ID：jrc_sum159-1 电子显微镜数据DOI：10.25378/janelia.13114352 电子显微镜体素尺寸（nm）：4.0 × 4.0 × 4.0（X、Y、Z轴） 分割体素尺寸（nm）：4.0 × 4.0 × 4.0（X、Y、Z轴） 数据集URL：https://data.janelia.org/LzShC 可视化网站：https://openorganelle.janelia.org/datasets/jrc_sum159-1 发表文献：Heinrich 等人，2020年 数据集包含内容：手动注释