Semantic Segmentation of HeLa Cells: An Objective Comparison between one Traditional Algorithm and Three Deep-Learning Architectures

In this work, images of a HeLa cancer cell were semantically segmented with one traditional image-processing algorithm and three deep learning architectures: VGG16, ResNet18 and Inception-ResNet-v2. Three hundred slices, each 2000 x 2000 pixels, of a HeLa Cell were acquired with Serial Block Face Scanning Electron Microscopy. The deep learning architectures were pre-trained with ImageNet and then fine-tuned with transfer learning. 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%).