Skip to content

Latest commit

 

History

History
146 lines (119 loc) · 6.1 KB

README.md

File metadata and controls

146 lines (119 loc) · 6.1 KB
Raw

(Nature Communications Engineering 2024) Compressive Confocal Microscopy Imaging at the Single-Photon Level with Ultra-Low Sampling Ratios [PyTorch]

icon visitors

Shuai Liu*, Bin Chen*, Wenzhen Zou, Hao Sha, Xiaochen Feng, Sanyang Han, Xiu Li, Xuri Yao, Jian Zhang†, and Yongbing Zhang†

Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, China.

School of Electronic and Computer Engineering, Peking University, Shenzhen, China.

School of Computer Science and Technology, Harbin Institute of Technology (Shenzhen), Shenzhen, China.

Center for Quantum Technology Research, School of Physics, Beijing Institute of Technology, Beijing, China.

* Equal contribution † Corresponding authors

Accepted for publication in Communications Engineering (Nature Communications) 2024.

Abstract

Laser-scanning confocal microscopy serves as a critical instrument for microscopic research in biology. However, it suffers from low imaging speed and high phototoxicity. Here we build a novel deep compressive confocal microscope, which employs a digital micromirror device as a coding mask for single-pixel imaging and a pinhole for confocal microscopic imaging respectively. Combined with a deep learning reconstruction algorithm, our system is able to achieve high-quality confocal microscopic imaging with low phototoxicity. Our imaging experiments with fluorescent microspheres demonstrate its capability of achieving single-pixel confocal imaging with a sampling ratio of only approximately 0.03% in specific sparse scenarios. Moreover, the deep compressive confocal microscope allows single-pixel imaging at the single-photon level, thus reducing the excitation light power requirement for confocal imaging and suppressing the phototoxicity. We believe that our system has great potential for long-duration and high-speed microscopic imaging of living cells.

Overview

overview

Environment

torch.__version__ == '2.2.1+cu121'
numpy.__version__ == '1.24.4'
skimage.__version__ == '0.21.0'

Data and Pretrained Model Weights

Download the data and pretrained model weights. Unzip the files into ./data and ./code/weight directories, respectively.

The paths of all files should be:

.
├── README.md
├── code
│   ├── model.py
│   ├── test.py
│   ├── test.sh
│   ├── train.py
│   ├── train.sh
│   ├── utils.py
│   └── weight
│       ├── f-actin
│       │   └── layer_9_f_128
│       │       └── net_params_30000.pkl
│       ├── flureoscent_microsphere
│       │   └── layer_9_f_128
│       │       └── net_params_30000.pkl
│       ├── nucleus
│       │   └── layer_9_f_128
│       │       └── net_params_30000.pkl
│       └── potato_tuber
│           └── layer_9_f_128
│               └── net_params_30000.pkl
├── data
│   ├── A_128.npy
│   ├── A_32.npy
│   ├── f-actin
│   │   ├── test_X.npy
│   │   ├── test_X_WF.npy
│   │   ├── test_Y128.npy
│   │   ├── test_Y32.npy
│   │   ├── train_X.npy
│   │   ├── train_X_WF.npy
│   │   ├── train_Y128.npy
│   │   └── train_Y32.npy
│   ├── flureoscent_microsphere
│   │   ├── test_X.npy
│   │   ├── test_X_WF.npy
│   │   ├── test_Y128.npy
│   │   ├── test_Y32.npy
│   │   ├── train_X.npy
│   │   ├── train_X_WF.npy
│   │   ├── train_Y128.npy
│   │   └── train_Y32.npy
│   ├── nucleus
│   │   ├── test_X.npy
│   │   ├── test_X_WF.npy
│   │   ├── test_Y128.npy
│   │   ├── test_Y32.npy
│   │   ├── train_X.npy
│   │   ├── train_X_WF.npy
│   │   ├── train_Y128.npy
│   │   └── train_Y32.npy
│   └── potato_tuber
│       ├── test_X.npy
│       ├── test_X_WF.npy
│       ├── test_Y128.npy
│       ├── test_Y32.npy
│       ├── train_X.npy
│       ├── train_X_WF.npy
│       ├── train_Y128.npy
│       └── train_Y32.npy
└── figs
    └── overview.png

Test

cd code
python test.py --data_type=nucleus
python test.py --data_type=flureoscent_microsphere
python test.py --data_type=f-actin
python test.py --data_type=potato_tuber

The reconstructed images will be in ./code/result.

Train

cd code
python train.py --data_type=nucleus
python train.py --data_type=flureoscent_microsphere
python train.py --data_type=f-actin
python train.py --data_type=potato_tuber

The log and model files will be in ./code/log and ./code/weight, respectively.

Citation

If you find the code helpful in your research or work, please cite the following paper:

@article{liu2024compressive,
  title={Compressive confocal microscopy imaging at the single-photon level with ultra-low sampling ratios},
  author={Liu, Shuai and Chen, Bin and Zou, Wenzhen and Sha, Hao and Feng, Xiaochen and Han, Sanyang and Li, Xiu and Yao, Xuri and Zhang, Jian and Zhang, Yongbing},
  journal={Communications Engineering},
  volume={3},
  number={1},
  pages={88},
  year={2024},
  publisher={Nature Publishing Group UK London}
}