Automatic Reconstruction of Mitochondria and Endoplasmic Reticulum in Electron Microscopy Volumes by Deep Learning
Automatic Reconstruction of Mitochondria and Endoplasmic Reticulum in Electron Microscopy Volumes by Deep Learning
复制标题
通过深度学习自动重建电子显微镜体积中的线粒体和内质网
DOI：
10.3389/fnins.2020.00599
复制
发表时间：
2020-07-21
影响因子：
4.3
通讯作者：
Han, Hua
中科院分区：
文献类型：
--
作者：
Liu, Jing;Li, Linlin;Han, Hua
线粒体和内质网(ER)共同占据了细胞体积的20%以上，形态异常可能导致细胞功能障碍。随着大规模电子显微镜(EM)的迅速发展，在生物学研究中，这些细胞器的人工轮廓和三维重建已经完成。然而，从EM图像中手工分割线粒体和内质网非常耗时，无法满足大数据分析的要求。在这里，我们提出了一种用于线粒体和内质网重建的自动化管道，包括线粒体和内质网接触点(MAM)。我们提出了一种新的递归神经网络来检测和分割线粒体，并提出了一种完全剩余卷积网络来重建内质网。基于突触稀疏分布的特点，利用线粒体上下文信息对局部误导的结果进行校正，得到三维线粒体重建结果。实验结果表明，该方法达到了最好的性能。
Together, mitochondria and the endoplasmic reticulum (ER) occupy more than 20% of a cell's volume, and morphological abnormality may lead to cellular function disorders. With the rapid development of large-scale electron microscopy (EM), manual contouring and three-dimensional (3D) reconstruction of these organelles has previously been accomplished in biological studies. However, manual segmentation of mitochondria and ER from EM images is time consuming and thus unable to meet the demands of large data analysis. Here, we propose an automated pipeline for mitochondrial and ER reconstruction, including the mitochondrial and ER contact sites (MAMs). We propose a novel recurrent neural network to detect and segment mitochondria and a fully residual convolutional network to reconstruct the ER. Based on the sparse distribution of synapses, we use mitochondrial context information to rectify the local misleading results and obtain 3D mitochondrial reconstructions. The experimental results demonstrate that the proposed method achieves state-of-the-art performance.