Graphic STI
logo EPFL
text EPFL
english only
Biomedical Imaging Group
3D Deconvolution Microscopy
BIG >  3D Deconvolution Microscopy
DECONVOLUTION

Outline

Open Software

» DeconvolutionLab2

» DeconvolutionLab

» PSF Generator

3D Reference Datasets

Benchmarking

Advanced Methods

Conditions of Use

Acknowledgment

BIG

Home

Research

Publications

Demos

Algorithms

Teaching

DeconvolutionLab2

The remasterized Java deconvolution tool

DeconvolutionLab2 is freely accessible and open-source for 3D deconvolution microscopy; it can be linked to well-known imaging software platforms, ImageJ, Fiji, ICY (not yet implemented), Matlab (not yet implemented) and it runs as a stand-alone application. The backbone of our software architecture is a library that contains the number-crunching elements of the deconvolution task. It includes the tool for a complete validation pipeline. Inquisitive minds inclined to peruse the code will find it fosters the understanding of deconvolution.

splash

Reference

D. Sage, L. Donati, F. Soulez, D. Fortun, G. Schmit, A. Seitz, R. Guiet, C. Vonesch, M. Unser

DeconvolutionLab2 : An Open-Source Software for Deconvolution Microscopy

Methods-Image Processing for Biologists, vol. 115, 2017.

Installation of DeconvolutionLab2

 

Download and installation

Command line interface

Download

DeconvolutionLab_2.jar

Do no unzip this file

  1. To run the user interface of Deconvolutionlab2, enter the following command line in the terminal:

    java -jar DeconvolutionLab_2.jar Lab

  2. To run a deconcolution task, run Deconvolutionlab2 with a command following this example:

    java -jar DeconvolutionLab_2.jar Run -image synthetic Cube 10.0 1.0 size 200 100 100 -psf synthetic Double-Helix 3.0 30.0 10.0 size 200 100 100 intensity 255.0 -algorithm RIF 0.1000 -out mip MI1 -path home


ImageJ

Put the file DeconvolutionLab_2.jar in the plugins folder and restart ImageJ or Fiji. Check the menu Plugins » DeconvolutionLab2.

  1. » DeconvolutionLab2 Lab: Start the complete user interface of DeconvolutionLab2
  2. » DeconvolutionLab2 Run: Run headless a deconvolution command given as a macro, if no macro it is run the Lab
  3. » DeconvolutionLab2 Launch: launch the GUI for a deconvolution command given as a macro, if no macro it is run the Lab

Example of ImageJ macro

	image = " -image synthetic Cube 10.0 1.0 size 200 100 100"
	psf = " -psf synthetic Double-Helix 3.0 30.0 10.0 size 200 100 100"
	algorithm = " -algorithm RIF 0.1000 -out mip MI1 -path home"
	run("DeconvolutionLab2 Launch", image + psf + algorithm + parameters)
		

ImageJ2

Fiji

Matlab

Add DeconvolutionLab_2 in the java path

javaaddpath([matlabroot filesep 'java' filesep 'DeconvolutionLab_2.jar'])

then run a specific algorithm

result = DL2.RIF(image, psf, 0.125 , '');


Stand-alone application
If you have a Java JDK properly installed in your machine, double-click on the DeconvolutionLab_2.jar file to the ser interface of Deconvolutionlab2

Java code

Use the DeconvolutionLab2 as a Java library

Example Java code

	RealSignal r = Lab.getImage("fundus.tif");
	RealSignal h = new DirectionalMotionBlur(2, 20, 20).generate(r.nx, r.ny, r.nz);
	Simulation sim = new Simulation(0, 1, 0);
	RealSignal y = sim.run(r, h);
	TikhonovRegularizedInverseFilter trif = new TikhonovRegularizedInverseFilter(0.001);
	RealSignal x = trif.run(y, h);
	Lab.show(x, "TRIF " + i);

Icy
Not yet implemented
Download and installation
GIT Repository git clone https://c4science.ch/diffusion/2075/deconvolution.git
Status Status of the development
API Java doc
FFT Libraries
JTransforms Visit the page of JTransforms
JTransforms is already included in Fiji and Icy
Get a JTransforms.jar file and put it in the same directory than DeconvolutionLab_2.jar
FFTW FFTW.zip
It is include the FFTW2 dynamic libraries
for Mac OSX, and Windows 32-bits and 64-bits machines, and Linux 32-bits and 64-bits machines.
Download the FFTW.zip folder, unzip it and put it in the same directory than DeconvolutionLab_2.jar

How to use Deconvolution2

▶ Image

  • Drag and drop your input image file (or directory) into the Image bar or in the image table
  • file Select a file containing a z-stack of images (TIF or STK)
  • directory Select a directory containing the list of images using
  • synthetic Create a synthetic image by choosing its shape and its size
  • platform Select one of the open images of the platform (ImageJ)
  • active Select the active image window of the platform (ImageJ)

▶ PSF

  • In the way than the images, select the PSF images.

▶ Algorithm

  • Choose an algorithm
  • Select the parameters

▶ Path

  • Drag and drop a directory used as working directory
  • Select the current directory as main directory

Run

  • Run the deconvolution task in headless mode

Launch

  • Start the deconvolution dialog box, then Run the deconvolution task

Batch

  • The current deconvolution command is added in the Batch table (see Advanced tab)

Algorithms of DeconvolutionLab2

 

Algorithms Shortname Iterative Step Controllable Regularization Wavelets
Deconvolution
Regularized Inverse Filter

Laplacian Regularized Inverse Filter

RIF

LRIF

Direct No Yes
Tikhonov Regularized Inverse Filter TRIF Direct Yes
Naive Inverse Filter

Inverse Filter

NIF

IF

Direct No Yes
Richardson-Lucy RL Iterative No No
Richardson-Lucy Total Variation RLTV Iterative No Yes
Landweber

Linear Least Squares

LW

LLS

Iterative Yes No
Non-negative Least Squares

Landweber+Positivity

NNLS

LW+

Iterative Yes No
Bounded-Variable Least Squares

Spark-Parker

BVLS

SP

Iterative Yes No
Van Cittert VC Iterative Yes No
Tikhonov-Miller TM Iterative Yes Yes
Iterative Constraint Tikhonov-Miller ICTM Iterative Yes Yes
FISTA FISTA Iterative No Yes Haar, Spline
ISTA ISTA Iterative No Yes Haar, Spline
Simulation
Simulation SIM Direct
Convolution CONV Direct
Identity I Direct
Non-stabilized Division DIV Direct

Results of DeconvolutionLab2

 

The results of the deconvolution in terms of image reconstruction are the same than our previous version DeconvolutionLab. DeconvolutionLab2 improves the usability through friendly user-interface and it run on various imaging platform It offers a larger choice a FFT librairies and different ways to cancel the border artefacts. In addition, it allows a scripting for batch processing.

Scripting DeconvolutionLab2

 

The command line of DeconvolutionLab2 consists of a series of arguments that allows a full control of the processing. The command line is written in a single line, space is mostly used as separator. The general format of the argument is:

-keyword [option] parameters

The list of keywords and the options presenting in the following table. The sign | indicate a OR). The default value are written in bold

Keyword Default Options Description
-image file Mandatory Path to a single file (z-stack) usually a TIF or STK file Source of images.
The 3D input data should be a z-stack of images.
-image directory Path to a directory containing 2D images
[pattern
-image synthetic Name and parameters of the shape
[intensity, size, center]>
-image platform Name of the image of the platform (ImageJ or Icy)
-psf file Mandatory Path to a single file (z-stack) used as PSF usually a TIF or STK file Source of PSF.
The 3D PSF data should be a z-stack of images.
-psf directory Path to a directory containing 2D images
[pattern]
-psf synthetic Name and parameters of the shape
[intensity, size, center]
-psf platform Name of the image of the platform (ImageJ or Icy)
-algorithm Mandatory RIF | TRIF | NIF | LW | NNLS | BVLS | RL | RLTV | TM | ICTM | ISTA | FISTA | VC | I | CONV | SIM | DIV

Synonym of the acronym:

RIF = LRIF, NIF = IF, LW = LLS, NNLS = LW+, BVLS = SP, I = ID
Name and parameter of the algorithm
-path current current | path Working directory
Output (several instances of out are possible)
-display yes yes | no The final results is displayed
-out stack intact float

Name of the output:

Note that this name is used as title of the window image and as the filename for the storage

Option for dynamic:

intact | rescaled | normalized | clipped

Option for type:

byte | short | float

Mode:

By default the output is shown and saved.

nosave | noshow

Output as a stack of images (TIF)
-out series intact float Output as series of 2D images (slices, XY)
-out mip intact float Output as a maximum-intensity projection
-out ortho intact float Output as a 3 orthogonal views centered around the keypoint
-out planar intact float Outputs as a 2D side-to-side image of all the z-slices
Controller
-monitor console table console | table | no Selection of the monitoring output
-verbose log log | quiet | prolix | mute Message monitoring
-stats show show | save | no Statistics
-constraint no no | nonnegativity | clipped Spatial constraint on the signal
-residu no no | value Stops when the minimal residu is reached
-time no no | value Limitation of running time
-reference no no | filename Assess the current deconvolved image with the reference image
Preprocessing
-pad NO NO 0 0 NO | X2 | X23 | X235 | E2 Lateral and axial padding and extension scheme
-apo NO NO UNIFORM | NO | HAMMING | HANN | COSINE | TUKEY | WELCH Lateral and axial apodization window function
-norm 1 no | value Normalization factor for the PSF
Resources
-fft fastest academic | jtransforms | fftw2 Indicates the FFT library
-epsilon 1E-6 value Machine Epsilon

Example

-image synthetic Cube 10.0 1.0 size 200 100 100 intensity 255.0 -psf synthetic Double-Helix 3.0 30.0 10.0 size 200 100 100 intensity 255.0 -algorithm RIF 0.1000 -out mip MI1 -norm 10.0 -path home

Course

 

Material Download Size Description
Slide in PDF (without the animation/video) 3D-Deconvolution-Microscopy.pdf 4.6 Mb Course given in Neubias 2020, February 2017
Restoration logo.zip 0.6 Mb 2D simulation, influence of the PSF shape to restore the original image
Naive Inverse Filter naive-deconvolution.zip 3.3 Mb 2D simulation of the inverse filter
Simulation to check the resolution test-resolution.ijm 0 Mb Macro to generate simulated data and simulated PSF
Simulation to check the spectral effect spectral-analysis.zip 0.7 Mb 2D simulation of fine structures
Hollow bars bars.zip 28 Mb 3D reference, 3D corruputed data and 3D PSF
C-elegans embryo c-elegans.zip 183 Mb 3 fluoresence channels, 3D data and 3D theoretical PSF
Synthetic microtubules microtubules-challenge.zip 63 Mb 3D data and 3D theoretical PSF of a realistic specimen
Drosophila (crop) drosophila-crop.zip 19 Mb Small 3D data and 3D theoritical PSF
Synthetic microtubules real-donut.zip 463 Mb 3D data and 3D estimated PSF of real well-defined objects (donut)

Conditions of use

You'll be free to use this software for research purposes, but you must not transmit and distribute it without our consent. In addition, you undertake to include a citation whenever you present or publish results that are based on it. EPFL makes no warranties of any kind on this software and shall in no event be liable for damages of any kind in connection with the use and exploitation of this technology.

© 2017 EPFL • webmaster.big@epfl.ch • 20.04.2017