Skip to content

Mitochondria Segmentation using 3D Convolutional LSTM U-net

Notifications You must be signed in to change notification settings

nguyen14ck/mito_3d_clstm

Repository files navigation

This repository contains the implementation of our paper of 3D Convolutional Long Short Term Memory (CLSTM) network.

Nguyen, N. P., White, T., & Bunyak, F. Mitochondria Instance Segmentation in Electron Microscopy Image Volumes using 3D Deep Learning Networks (2021). IEEE/Applied Imagery Pattern Recognition Workshop, AIPR.

3D_CLSTM

It relies on the following projects:
CFCM-2D
ConvLSTM
MONAI
TorchIO
Segmenation Models Pytorch
Connectomics

Install

pip install git+https://github.com/zudi-lin/pytorch_connectomics

pip install -r requirements.txt

Train

python train_mito.py --model path_to_model --data path_to_data

with
--model: check point path (default is None)
--data: data path

Test

python test_mito.py --model path_to_model --data path_to_data

with
--model: check point path
--data: data path

Rat volume

References

The dataset used in this work is downloaded from

Wei, D., Lin, Z., Franco-Barranco, D., Wendt, N., Liu, X., Yin, W., Huang, X., Gupta, A., Jang, W. D., Wang, X., Arganda-Carreras, I., Lichtman, J. W., & Pfister, H. (2020). MitoEM Dataset: Large-scale 3D Mitochondria Instance Segmentation from EM Images. Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention, 12265, 66–76. https://doi.org/10.1007/978-3-030-59722-1_7

About

Mitochondria Segmentation using 3D Convolutional LSTM U-net

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages