• Corpus ID: 233289586

Advanced Deep Networks for 3D Mitochondria Instance Segmentation

  title={Advanced Deep Networks for 3D Mitochondria Instance Segmentation},
  author={Mingxing Li and Chang Wen Chen and Xiaoyu Liu and Wei Huang and Yueyi Zhang and Zhiwei Xiong},
Mitochondria instance segmentation from electron microscopy (EM) images has seen notable progress since the introduction of deep learning methods. In this paper, we propose two advanced deep networks, named Res-UNet-R and Res-UNetH, for 3D mitochondria instance segmentation from Rat and Human samples. Specifically, we design a simple yet effective anisotropic convolution block and deploy a multi-scale training strategy, which together boost the segmentation performance. Moreover, we enhance the… 
1 Citations

Figures and Tables from this paper

PyTorch Connectomics: A Scalable and Flexible Segmentation Framework for EM Connectomics
PyTC, an opensource deep-learning framework for the semantic and instance segmentation of volumetric microscopy images, built upon PyTorch, is presented, which achieves the best performance in the CREMI challenge for synaptic cleft segmentation and competitive performance on mitochondria and neuronal nuclei segmentation.


MitoEM Dataset: Large-Scale 3D Mitochondria Instance Segmentation from EM Images
The MitoEM dataset is introduced, a 3D mitochondria instance segmentation dataset with two (30μm)3 volumes from human and rat cortices respectively, 3, 600× larger than previous benchmarks.
Automatic Reconstruction of Mitochondria and Endoplasmic Reticulum in Electron Microscopy Volumes by Deep Learning
This work proposes a novel recurrent neural network to detect and segment mitochondria and a fully residual convolutional network to reconstruct the ER, and uses mitochondrial context information to rectify the local misleading results and obtain 3D mitochondrial reconstructions.
Learning for Structured Prediction Using Approximate Subgradient Descent with Working Sets
We propose a working set based approximate sub gradient descent algorithm to minimize the margin-sensitive hinge loss arising from the soft constraints in max-margin learning frameworks, such as the
Automatic instance segmentation of mitochondria in electron microscopy data
We present a new method for rapid, automated, large-scale 3D mitochondria instance segmentation, developed in response to the ISBI 2021 MitoEM Challenge. In brief, we trained separate machine
Contrastive Learning for Mitochondria Segmentation
A novel and effective contrastive learning framework to learn a better feature representation from hard examples to improve segmentation of mitochondria segmentation in electron microscopy images is proposed.
ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks
This paper proposes an Efficient Channel Attention (ECA) module, which only involves a handful of parameters while bringing clear performance gain, and develops a method to adaptively select kernel size of 1D convolution, determining coverage of local cross-channel interaction.
EM-Net: Centerline-Aware Mitochondria Segmentation in EM Images Via Hierarchical View-Ensemble Convolutional Network
  • Zhimin Yuan, Jialin Peng
  • Computer Science
    2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI)
  • 2020
A multi-task network named EM-Net is introduced, which includes an auxiliary centerline detection task to account for shape information of mitochondria represented by centerline, and is able to enhance the accuracy and robustness of segmentation task.
Learning to Restore ssTEM Images from Deformation and Corruption
This paper proposes a novel strategy for modeling the main type of degradation, i.e., Support Film Folds (SFF), by characterizing this degradation process as a combination of content deformation and corruption, which enables the training of a tailored deep restoration network for ssTEM image restoration.
Mask R-CNN
This work presents a conceptually simple, flexible, and general framework for object instance segmentation that outperforms all existing, single-model entries on every task, including the COCO 2016 challenge winners.
Squeeze-and-Excitation Networks
This work proposes a novel architectural unit, which is term the “Squeeze-and-Excitation” (SE) block, that adaptively recalibrates channel-wise feature responses by explicitly modelling interdependencies between channels and shows that these blocks can be stacked together to form SENet architectures that generalise extremely effectively across different datasets.