ctom2 / lucyd-deconvolution Goto Github PK

View Code? Open in Web Editor NEW

14.0 1.0 1.0 1.05 MB

[MICCAI 2023] This is the official code for the paper "A Feature-Driven Richardson-Lucy Deconvolution Network"

Python 100.00%

deblurring deconvolution light-sheet microscopy volumetric-data fluorescence deep-learning denoising

lucyd-deconvolution's Introduction

LUCYD: A Feature-Driven Richardson-Lucy Deconvolution Network

[paper][arXiv]

Run in Colab 🟡▶️

The process of acquiring microscopic images in life sciences often results in image degradation and corruption, characterised by the presence of noise and blur, which poses significant challenges in accurately analysing and interpreting the obtained data. We propse LUCYD, a novel method for the restoration of volumetric microscopy images that combines the Richardson-Lucy deconvolution formula and the fusion of deep features obtained by a fully convolutional network. By integrating the image formation process into a feature-driven restoration model, the proposed approach aims to enhance the quality of the restored images whilst reducing computational costs and maintaining a high degree of interpretability.

Prerequisities:

Python 3.7 or higher
PyTorch 1.12.1 or higher
NumPy 1.22.4
TorchMetrics 1.0.1
Tifffile 2023.7.10

Training:

model = LUCYD(num_res=1)
model = train(model, train_dataloader, test_dataloader)

Testing:

evaluate(model, eval_dataloader)

Cite

Please cite our work if you find it useful to your research.

@InProceedings{10.1007/978-3-031-43993-3_63,
  author="Chobola, Tom{\'a}{\v{s}}
  and M{\"u}ller, Gesine
  and Dausmann, Veit
  and Theileis, Anton
  and Taucher, Jan
  and Huisken, Jan
  and Peng, Tingying",
  title="LUCYD: A Feature-Driven Richardson-Lucy Deconvolution Network",
  booktitle="Medical Image Computing and Computer Assisted Intervention -- MICCAI 2023",
  year="2023",
  publisher="Springer Nature Switzerland",
  pages="656--665",
  isbn="978-3-031-43993-3"
}

lucyd-deconvolution's People

Contributors

Stargazers

Watchers

Forkers

roimautner

lucyd-deconvolution's Issues

No proper requirements file?

Hi. I'm interested in trying out the method but I cannot use your code since the requirements are not properly specified. Installing only pytorch and python does not suffice for the code to run. Could you perhaps update the repo in a way that others could try/cite your work?

Local Execution Fails Upon Loading Weights

Hi Tomas,

I've successfully ran the code on your Colab file, however, when I try to do it locally, I face errors while loading the weights using the following:

from model.lucyd import LUCYD, device
import os
import torch

model = LUCYD().to(device)
weigth_selection = "act"
weight_map = {"mixture": "lucyd-mixture.pth", "act": "lucyd-act.pth", "nuc": "lucyd-nuc.pth"}
WEIGHTS = os.path.join("model", "weights", weight_map[weigth_selection])
model.load_state_dict(torch.load(WEIGHTS))

and end up with this error:


File "/home/username/Desktop/lucyd/test.py", line 11, in <module>
    model.load_state_dict(torch.load(WEIGHTS))
  File "/home/username/anaconda3/envs/lucyd/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1604, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for LUCYD:
    Missing key(s) in state_dict: "Encoder.0.layers.1.main.0.main.0.weight", "Encoder.0.layers.1.main.0.main.1.weight", "Encoder.0.layers.1.main.0.main.1.bias", "Encoder.0.layers.1.main.0.main.1.running_mean", "Encoder.0.layers.1.main.0.main.1.running_var", "Encoder.0.layers.1.main.1.main.0.weight", "Encoder.0.layers.1.main.1.main.1.weight", "Encoder.0.layers.1.main.1.main.1.bias", "Encoder.0.layers.1.main.1.main.1.running_mean", "Encoder.0.layers.1.main.1.main.1.running_var", "Encoder.0.layers.2.main.0.main.0.weight", "Encoder.0.layers.2.main.0.main.1.weight", "Encoder.0.layers.2.main.0.main.1.bias", "Encoder.0.layers.2.main.0.main.1.running_mean", "Encoder.0.layers.2.main.0.main.1.running_var", "Encoder.0.layers.2.main.1.main.0.weight", "Encoder.0.layers.2.main.1.main.1.weight", "Encoder.0.layers.2.main.1.main.1.bias", "Encoder.0.layers.2.main.1.main.1.running_mean", "Encoder.0.layers.2.main.1.main.1.running_var", "Encoder.0.layers.3.main.0.main.0.weight", "Encoder.0.layers.3.main.0.main.1.weight", "Encoder.0.layers.3.main.0.main.1.bias", "Encoder.0.layers.3.main.0.main.1.running_mean", "Encoder.0.layers.3.main.0.main.1.running_var", "Encoder.0.layers.3.main.1.main.0.weight", "Encoder.0.layers.3.main.1.main.1.weight", "Encoder.0.layers.3.main.1.main.1.bias", "Encoder.0.layers.3.main.1.main.1.running_mean", "Encoder.0.layers.3.main.1.main.1.running_var", "Encoder.1.layers.1.main.0.main.0.weight", "Encoder.1.layers.1.main.0.main.1.weight", "Encoder.1.layers.1.main.0.main.1.bias", "Encoder.1.layers.1.main.0.main.1.running_mean", "Encoder.1.layers.1.main.0.main.1.running_var", "Encoder.1.layers.1.main.1.main.0.weight", "Encoder.1.layers.1.main.1.main.1.weight", "Encoder.1.layers.1.main.1.main.1.bias", "Encoder.1.layers.1.main.1.main.1.running_mean", "Encoder.1.layers.1.main.1.main.1.running_var", "Encoder.1.layers.2.main.0.main.0.weight", "Encoder.1.layers.2.main.0.main.1.weight", "Encoder.1.layers.2.main.0.main.1.bias", "Encoder.1.layers.2.main.0.main.1.running_mean", "Encoder.1.layers.2.main.0.main.1.running_var", "Encoder.1.layers.2.main.1.main.0.weight", "Encoder.1.layers.2.main.1.main.1.weight", "Encoder.1.layers.2.main.1.main.1.bias", "Encoder.1.layers.2.main.1.main.1.running_mean", "Encoder.1.layers.2.main.1.main.1.running_var", "Encoder.1.layers.3.main.0.main.0.weight", "Encoder.1.layers.3.main.0.main.1.weight", "Encoder.1.layers.3.main.0.main.1.bias", "Encoder.1.layers.3.main.0.main.1.running_mean", "Encoder.1.layers.3.main.0.main.1.running_var", "Encoder.1.layers.3.main.1.main.0.weight", "Encoder.1.layers.3.main.1.main.1.weight", "Encoder.1.layers.3.main.1.main.1.bias", "Encoder.1.layers.3.main.1.main.1.running_mean", "Encoder.1.layers.3.main.1.main.1.running_var", "Encoder.2.layers.1.main.0.main.0.weight", "Encoder.2.layers.1.main.0.main.1.weight", "Encoder.2.layers.1.main.0.main.1.bias", "Encoder.2.layers.1.main.0.main.1.running_mean", "Encoder.2.layers.1.main.0.main.1.running_var", "Encoder.2.layers.1.main.1.main.0.weight", "Encoder.2.layers.1.main.1.main.1.weight", "Encoder.2.layers.1.main.1.main.1.bias", "Encoder.2.layers.1.main.1.main.1.running_mean", "Encoder.2.layers.1.main.1.main.1.running_var", "Encoder.2.layers.2.main.0.main.0.weight", "Encoder.2.layers.2.main.0.main.1.weight", "Encoder.2.layers.2.main.0.main.1.bias", "Encoder.2.layers.2.main.0.main.1.running_mean", "Encoder.2.layers.2.main.0.main.1.running_var", "Encoder.2.layers.2.main.1.main.0.weight", "Encoder.2.layers.2.main.1.main.1.weight", "Encoder.2.layers.2.main.1.main.1.bias", "Encoder.2.layers.2.main.1.main.1.running_mean", "Encoder.2.layers.2.main.1.main.1.running_var", "Encoder.2.layers.3.main.0.main.0.weight", "Encoder.2.layers.3.main.0.main.1.weight", "Encoder.2.layers.3.main.0.main.1.bias", "Encoder.2.layers.3.main.0.main.1.running_mean", "Encoder.2.layers.3.main.0.main.1.running_var", "Encoder.2.layers.3.main.1.main.0.weight", "Encoder.2.layers.3.main.1.main.1.weight", "Encoder.2.layers.3.main.1.main.1.bias", "Encoder.2.layers.3.main.1.main.1.running_mean", "Encoder.2.layers.3.main.1.main.1.running_var", "Decoder.0.layers.1.main.0.main.0.weight", "Decoder.0.layers.1.main.0.main.1.weight", "Decoder.0.layers.1.main.0.main.1.bias", "Decoder.0.layers.1.main.0.main.1.running_mean", "Decoder.0.layers.1.main.0.main.1.running_var", "Decoder.0.layers.1.main.1.main.0.weight", "Decoder.0.layers.1.main.1.main.1.weight", "Decoder.0.layers.1.main.1.main.1.bias", "Decoder.0.layers.1.main.1.main.1.running_mean", "Decoder.0.layers.1.main.1.main.1.running_var", "Decoder.0.layers.2.main.0.main.0.weight", "Decoder.0.layers.2.main.0.main.1.weight", "Decoder.0.layers.2.main.0.main.1.bias", "Decoder.0.layers.2.main.0.main.1.running_mean", "Decoder.0.layers.2.main.0.main.1.running_var", "Decoder.0.layers.2.main.1.main.0.weight", "Decoder.0.layers.2.main.1.main.1.weight", "Decoder.0.layers.2.main.1.main.1.bias", "Decoder.0.layers.2.main.1.main.1.running_mean", "Decoder.0.layers.2.main.1.main.1.running_var", "Decoder.0.layers.3.main.0.main.0.weight", "Decoder.0.layers.3.main.0.main.1.weight", "Decoder.0.layers.3.main.0.main.1.bias", "Decoder.0.layers.3.main.0.main.1.running_mean", "Decoder.0.layers.3.main.0.main.1.running_var", "Decoder.0.layers.3.main.1.main.0.weight", "Decoder.0.layers.3.main.1.main.1.weight", "Decoder.0.layers.3.main.1.main.1.bias", "Decoder.0.layers.3.main.1.main.1.running_mean", "Decoder.0.layers.3.main.1.main.1.running_var", "Decoder.1.layers.1.main.0.main.0.weight", "Decoder.1.layers.1.main.0.main.1.weight", "Decoder.1.layers.1.main.0.main.1.bias", "Decoder.1.layers.1.main.0.main.1.running_mean", "Decoder.1.layers.1.main.0.main.1.running_var", "Decoder.1.layers.1.main.1.main.0.weight", "Decoder.1.layers.1.main.1.main.1.weight", "Decoder.1.layers.1.main.1.main.1.bias", "Decoder.1.layers.1.main.1.main.1.running_mean", "Decoder.1.layers.1.main.1.main.1.running_var", "Decoder.1.layers.2.main.0.main.0.weight", "Decoder.1.layers.2.main.0.main.1.weight", "Decoder.1.layers.2.main.0.main.1.bias", "Decoder.1.layers.2.main.0.main.1.running_mean", "Decoder.1.layers.2.main.0.main.1.running_var", "Decoder.1.layers.2.main.1.main.0.weight", "Decoder.1.layers.2.main.1.main.1.weight", "Decoder.1.layers.2.main.1.main.1.bias", "Decoder.1.layers.2.main.1.main.1.running_mean", "Decoder.1.layers.2.main.1.main.1.running_var", "Decoder.1.layers.3.main.0.main.0.weight", "Decoder.1.layers.3.main.0.main.1.weight", "Decoder.1.layers.3.main.0.main.1.bias", "Decoder.1.layers.3.main.0.main.1.running_mean", "Decoder.1.layers.3.main.0.main.1.running_var", "Decoder.1.layers.3.main.1.main.0.weight", "Decoder.1.layers.3.main.1.main.1.weight", "Decoder.1.layers.3.main.1.main.1.bias", "Decoder.1.layers.3.main.1.main.1.running_mean", "Decoder.1.layers.3.main.1.main.1.running_var", "Decoder.2.layers.1.main.0.main.0.weight", "Decoder.2.layers.1.main.0.main.1.weight", "Decoder.2.layers.1.main.0.main.1.bias", "Decoder.2.layers.1.main.0.main.1.running_mean", "Decoder.2.layers.1.main.0.main.1.running_var", "Decoder.2.layers.1.main.1.main.0.weight", "Decoder.2.layers.1.main.1.main.1.weight", "Decoder.2.layers.1.main.1.main.1.bias", "Decoder.2.layers.1.main.1.main.1.running_mean", "Decoder.2.layers.1.main.1.main.1.running_var", "Decoder.2.layers.2.main.0.main.0.weight", "Decoder.2.layers.2.main.0.main.1.weight", "Decoder.2.layers.2.main.0.main.1.bias", "Decoder.2.layers.2.main.0.main.1.running_mean", "Decoder.2.layers.2.main.0.main.1.running_var", "Decoder.2.layers.2.main.1.main.0.weight", "Decoder.2.layers.2.main.1.main.1.weight", "Decoder.2.layers.2.main.1.main.1.bias", "Decoder.2.layers.2.main.1.main.1.running_mean", "Decoder.2.layers.2.main.1.main.1.running_var", "Decoder.2.layers.3.main.0.main.0.weight", "Decoder.2.layers.3.main.0.main.1.weight", "Decoder.2.layers.3.main.0.main.1.bias", "Decoder.2.layers.3.main.0.main.1.running_mean", "Decoder.2.layers.3.main.0.main.1.running_var", "Decoder.2.layers.3.main.1.main.0.weight", "Decoder.2.layers.3.main.1.main.1.weight", "Decoder.2.layers.3.main.1.main.1.bias", "Decoder.2.layers.3.main.1.main.1.running_mean", "Decoder.2.layers.3.main.1.main.1.running_var".

Note that I tried all different keys there, but no luck. I believe there could be some issue with the weights.

Here is my pip freeze output:

python==3.10.4
lightning-utilities==0.9.0
numpy==1.22.4
packaging==23.1
tifffile==2023.7.10
torch==1.12.1
torchmetrics==1.0.1
typing_extensions==4.7.1

I have CUDA 11.7.1 and I am on an Ubuntu OS.

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.

Jobs

Jooble