Research Centre for Biomedical Engineering School of Mathematics, Computer Science and Engineering, Department of Electrical & Electronic Engineering, City, University of London, London, United Kingdom.
Electron Microscopy Science Technology Platform, The Francis Crick Institute, London, United Kingdom.
PLoS One. 2020 Oct 2;15(10):e0230605. doi: 10.1371/journal.pone.0230605. eCollection 2020.
The quantitative study of cell morphology is of great importance as the structure and condition of cells and their structures can be related to conditions of health or disease. The first step towards that, is the accurate segmentation of cell structures. In this work, we compare five approaches, one traditional and four deep-learning, for the semantic segmentation of the nuclear envelope of cervical cancer cells commonly known as HeLa cells. Images of a HeLa cancer cell were semantically segmented with one traditional image-processing algorithm and four three deep learning architectures: VGG16, ResNet18, Inception-ResNet-v2, and U-Net. Three hundred slices, each 2000 × 2000 pixels, of a HeLa Cell were acquired with Serial Block Face Scanning Electron Microscopy. The first three deep learning architectures were pre-trained with ImageNet and then fine-tuned with transfer learning. The U-Net architecture was trained from scratch with 36, 000 training images and labels of size 128 × 128. The image-processing algorithm followed a pipeline of several traditional steps like edge detection, dilation and morphological operators. The algorithms were compared by measuring pixel-based segmentation accuracy and Jaccard index against a labelled ground truth. The results indicated a superior performance of the traditional algorithm (Accuracy = 99%, Jaccard = 93%) over the deep learning architectures: VGG16 (93%, 90%), ResNet18 (94%, 88%), Inception-ResNet-v2 (94%, 89%), and U-Net (92%, 56%).
细胞形态的定量研究非常重要，因为细胞及其结构的结构和状态可以与健康或疾病的状况相关。实现这一目标的第一步是对细胞结构进行准确分割。在这项工作中，我们比较了五种方法，一种传统方法和四种深度学习方法，用于分割宫颈癌细胞（通常称为 HeLa 细胞）的核膜。使用一种传统图像处理算法和四种深度学习架构（VGG16、ResNet18、Inception-ResNet-v2 和 U-Net）对 HeLa 癌细胞图像进行语义分割。使用串行块面扫描电子显微镜获得了 300 张 HeLa 细胞的切片，每张切片的大小为 2000×2000 像素。前三个深度学习架构使用 ImageNet 进行预训练，然后使用迁移学习进行微调。U-Net 架构使用 36000 张大小为 128×128 的训练图像和标签进行从头开始训练。图像处理算法遵循传统步骤的流水线，例如边缘检测、膨胀和形态学操作。通过测量基于像素的分割准确性和 Jaccard 指数与标记的地面真实值进行比较，对算法进行了比较。结果表明，传统算法（准确率=99%，Jaccard=93%）的性能优于深度学习架构：VGG16（93%，90%）、ResNet18（94%，88%）、Inception-ResNet-v2（94%，89%）和 U-Net（92%，56%）。
