feilab-hust / f-vcd

F-VCD Package for 3D Reconstruction in Fourier light field microscopy

F-VCD package for FLFM 3D reconstruction proposed in our paper:

Video-rate 3D imaging of living cells using Fourier view-channel-depth (F-VCD) light field microscopy

System Requirements

• Windows 10. Linux should be able to run the code (based on python).
• Graphics: Nvidia GPU (RTX 3090 recommended)
• Memory: >64 GB RAM (128 recommended)
• Hard Drive: ~50GB free space (SSD recommended)

Installation instructions

• Matlab: download from official website
• Dependencies installation (using conda for environment installation): We provide package list for required environment installation. To use this, run the following commend inside a conda console

  conda env create -f ./Code/DL_net/environment/f_vcd_environment.yml
  

Note: More details are reffered to the user's manual ("Environment installation",Seciton: F-VCD Training && Validation)

How to use

Training

1. Training data preparation:

We provide example data for network training. Please download from Google Drive

Note: If users want to build their own training dataset, please refer to the user's manual (Seciton:Training paris generation)

2. Network training:

Before training F-VCD, users need to check whether the installed environment was activated or not. Then, users need to modify the parameters in network configuration file: config.py. The detailed descriptions of these parameters are listed below:

label: the name of training model

config.img_setting.img_size: patch size of input noisy LF.

config.img_setting.sr_factor: upsampling factor from noisy LF to clean LF

config.img_setting.ReScale_factor: upsampling factor from clean LF to 3D stacks

config.img_setting.Nnum: the view number of input noisy LF

config.channels_interp: filter number of F-VCD

config.sub_pixel: upsampling factor

Afer entering the baisc parameters, users can train the F-VCD model yourself, run:

python ./Code/DL_net/train.py

Validation

To using the trained model to inference 3D stack from LF, users need to enter the validation data path in config.py:

  Line 76: config.VALID.lf2d_path='{user-defined path}'

and run eval_test.py:

  python ./Code/DL_net/eval_test.py

Note: More details about network traing and validation are refered to user's manual

Citation

If you use this code and relevant data, please cite the corresponding paper where original methods appeared: Yi, C., Zhu, L., Sun, J. et al. Video-rate 3D imaging of living cells using Fourier view-channel-depth light field microscopy. Commun Biol 6, 1259 (2023).DOI: https://doi.org/10.1038/s42003-023-05636-x
Correspondence Should you have any questions regarding this code and the corresponding results, please contact Chengqiang Yi ([email protected])

To Do

Update the manual and format codes for readability