Automated Detection of labeled particles in microscopic stacks of Drosophila M. tissues

Abstract

Under development

ORIGINAL	TREATED	DISTRIBUTION	STATISTICS ENABLED

License

Copyright CNRS 2013

This software is a computer program whose purpose is to automatically detect labeled Particles in microscopic stacks of Drosophila M. tissues.

This software is governed by the CeCILL license under French law and abiding by the rules of distribution of free software. You can use, modify and/ or redistribute the software under the terms of the CeCILL license as circulated by CEA, CNRS and INRIA at the following URL: http://www.cecill.info/index.en.html

As a counterpart to the access to the source code and rights to copy, modify and redistribute granted by the license, users are provided only with a limited warranty and the software's author,the holder of the economic rights, and the successive licensors have only limited liability.

In this respect, the user's attention is drawn to the risks associated with loading, using, modifying and/or developing or reproducing the software by the user in light of its specific status of free software, that may mean that it is complicated to manipulate, and that also therefore means that it is reserved for developers and experienced professionals having in-depth computer knowledge. Users are therefore encouraged to load and test the software's suitability as regards their requirements in conditions enabling the security of their systems and/or data to be ensured and, more generally, to use and operate it in the same conditions as regards security.

The fact that you are presently reading this means that you have had knowledge of the CeCILL license and that you accept its terms.

Contributors

CLUET David	[email protected]

Acknowledgments

Under development

Publication

Under development

User guide

1) Requirements

The LIPID_DROPLETS macro requires ImageJ v1.49g or higher (Download).

To ensure the opening of the various microscope files, this macro requires the Bio-Formats plugin. If your ImageJ program does not feature it,, make sure to install it.

For ImageJ, the conversion of the analyzed stacks into animated GIFs requires the Gif-Stack-Writer Plugin that will be automatically installed, if missing.

To read the analysis report Markdown files, you can use the Markdown Preview Plus extension for Chrome. In the Extension menu allow access to files URL.

2) INSTALLATION

The LIPID_DROPLETS macro can be automatically installed with all required files in ImageJ and FIJI. Please follow the specific instructions described below.

1. Open ImageJ.
2. Open the src folder of the LIPID_DROPLETS macro.
3. Drag the Installation.ijm file on the ImageJ Menu bar to open it.
4. In the Menu bar of the macro select the Macros/Run Macro option.
5. The window will be closed automatically and all required files will be installed in the ImageJ/macros/Droplets folder. The shortcut Plugins/Macros/LIPID_DROPLETS will be added in the Menu bar.
6. Restart ImageJ to refresh the Menu bar.

1. Open FIJI.
2. Open the src folder of the LIPID_DROPLETS macro.
3. Drag the Installation_Fiji.ijm file on FIJI Menu bar to open it.
4. In the console select the Run option.
5. All required files will be installed in the Fiji.app/macros/Droplets folder. The shortcut Plugins/Macros/LIPID_DROPLETS will be added in the Menu bar.
6. Restart FIJI to refresh the Menu bar.

3) UPDATE

Follow the same instructions as for the initial installation process.

4) PERFORM AN ANALYSIS

Click on the Plugins/Macros/LIPID_DROPLETS shortcut.

The macro is initiated and the welcome window is prompted.

The next window will propose different pre-set analysis modes:

The settings are saved in the settings.csv file located in your ImageJ/macros/Droplets/ folder. They contain the key parameters for the analysis and are organized as following:

Name	Lipid Droplets Brain	Lipid Droplets Retina	Repo Brain
Extension of the files to analyze	.czi	.czi	.czi
Reference resolution (micron/pixel) in X	0.156	0.156	0.156
Reference resolution (micron/pixel) in Y	0.156	0.156	0.156
Distance xy in pixels between 2 particles	5	5	5
Distance in z between 2 particles	5	5	5
Minimum size in pixel	7	7	164
Maximum size in pixel	15000	15000	15000
Maximum size (to exclude big fat bodies)	500	822	8400
Minimum circularity	0.5	0.5	0.3
Maximum circularity	1	1	1
Number of Iterations	3	2	3
Zone for enlargement (in pixel) and erasing	5	5	5
Number of bins for distributions	50	50	50
Zone of analysis	?	Manual ROI	Whole tissue
Minimal number of new Particles	50	50	5
Enhance signal	true	true	false

If you respect this structure you can add your own settings. The various parameters will be described below. Once an analysis mode is selected you can modify all parameters using the main GUI.

This interface displays all the options and parameters that will be used batchwise for the analysis of all your image . In a parameters setting phase it could be usefull to be able to reuse always the same manual (optional) and slices selections. For this purpose, you can recall a previously determined Manual ROI and (starting, ending) slices couple, by selecting YES in the first listbox. If so you will be prompted later to select a previously created (by the macro) *_Parameters.txt file. All the new values of the key parameters modified in the interface will be applied.

Parameters

Extension of the stacks files: This correspond to the format of the files that will be manipulated (for example .czi, .tif, ...).

Initial resolution used for the calibration: The two following parameters correspond to the pixel value in micron of the set of stacks used to set your parameters values. Thus, in case of microscope change, and a subsequent different resolution, a correction coefficient will be applied.

Region to process: In order to provide versatile analyses the macro can manipulate the particles of interest can be detected and analyzed using one of the three following options:

• Whole tissue with Sub-Selection: The manipulator draws a Region Of Interest for each stack. During the automated analysis, the program detects the tissue using the "Huang" thresholding method and will identify the particles of interest using “Max-Entropy” threshold method. The particules of interest and then classified depending if they are located within the Tissue, specific of the Region of Interest, and within the Tissue but not in Region of Interest.
• Whole tissue: During the automated analysis, the program detects the tissue and will identify the particles of interest present.
• Manual ROI: The manipulator draws a Region Of Interest for each stack. During the automated analysis, the program detects the particles of interest present in this Region of Interest.

Thresholds between particles: As this program is designed to perform analyses on stack images, it requires to remove duplicates of a same particle present on several slices. For this purpose the macro requires minimal xy and z distances to distinguish two separated particles. Concerning the duplicates, only the largest one will be conserved.

Parameters of the initial low-resolution scan: In order to precisely detect all particles of interest and their shape, the program iteratively enhance the solidity of the bright particles using the Gaussian Blur... and Maximum... treatments. The “Max-Entropy” threshold method, then permits to detect the particles of interest. Thus at each iteration the program identify the brightest particles. They are stored in the ROI manager and removed from the image to allow the detection of less bright particles during the next iteration. This approach permits to precisely characterize the shape and size of the particles, but requires to remove any false positive bright particle. In this aim a first low resolution analysis detects even extremely big particles (Maximal surface) but with a minimal threshold (Minimal surface).

Note that these two parameters are presented in microns in the GUI, but are expressed in pixels (as seen by the user, when performin visual inspection) in the settings.csv file.

Parameters for the high-resolution scan: Once the potential particles of interest are identified, the program remove all of them that are not within the Tissue or that do not fit the searched properties:

• Maximal surface in microns (expressed in pixels in the settings.csv).
• A specific shape characterized by the Minimal and Maximal circularity. Note that this couple should be defined by values between 0 and 1.

Iteration parameters: Depending on the heterogeneity of the labeling of your particles of interest, the number of iterations has to be optimized. In order to stop the analysis if this maximal number of iterations is too high, a minimal number of new particles is set. If the number of newly identified particles is below this value, the analysis stops. Depending on the noise of the labeling it is not enough to remove only the false positive and the discarded duplicates. It could be required to remove a larger zone that the identified shape. For this purpose an enlargement correction factor can be applied before the removal from the picture. Finally, the user can indicate the number of bins used to draw the distribution curves in the report and statistic files.

For some labeling with small and extremely bright noise, the enhance signal option can be unchecked. This will affect the research engine in two ways. First the Maximum... treatment will not be applied. Secondly, the Minimal surface filter will not be applied during the first iteration. This will allow the program to detect all small noise particles and remove them only at the second iteration. Indeed during the development phase, we encountered stacks that generated only noise particles during the first iteration, leading to the premature abortion of the program. The non enhance signal setting avoids this issue.

Batch analysis

SCREEN SHOTS
	When all parameters are set the program prompts use to identify the root folder containing all your stacks.
	The macro will analyze the folder and its sub-folder to identify all the files with the correct extension. the program then indicates how many files have been found.
CHANNEL SELECTION
	The program will then load in background the first file. If you are using hyperstacks with several channels, the macro will ask you to specify in which channel the particles have to be found.
TAILORED STACK ANALYSIS
	Then the program will ask you for each file to select the starting and ending slices to be analyzed. If the Whole tissue with a Manual ROI option has been selected the program ask you to draw the sub-selection of interest. These file specific parameters will be saved in a _Parameters.txt file within the root folder of your analysis, allowing later key parameters optimization.

Research Engine

ORIGINAL PICTURE
First the tissue is detected on all selected slices.
TISSUE DETECTION	PARTICLES DETECTION iteration 1	PARTICLES DETECTION iteration 2
run("Gaussian Blur...", "sigma=20 slice");	run("Gaussian Blur...", "sigma=1 stack");	run("Gaussian Blur...", "sigma=1 stack");
setAutoThreshold("Huang dark");	run("Maximum...", "radius=5 stack");	run("Maximum...", "radius=5 stack");
run("Analyze Particles...", "size=100000-Infinity pixel show=Nothing add slice");	setAutoThreshold("MaxEntropy dark");	setAutoThreshold("MaxEntropy dark");
This shape is used to characterize the tissue-embed particles.	run("Analyze Particles...", "size="+ 1 +"-"+SizeMax+" add slice");	run("Analyze Particles...", "size="+ 1 +"-"+SizeMax+" add slice");
Continue with particles detection ->
	Removal of all unwanted particles.	Removal of all unwanted particles.
	When several slices, removal of duplicates of the same particle.	When several slices, removal of duplicates of the same particle.
	Removal of all particles on the Original Picture
	Continue to iteration 2 ->
		RESULT ON 5 SLICES