HELP!!!-segmentation and quantification of DIC images

classic Classic list List threaded Threaded
19 messages Options
Reply | Threaded
Open this post in threaded view
|

HELP!!!-segmentation and quantification of DIC images

Hui Zhang
<http://imagej.1557.x6.nabble.com/file/t382540/DIC-droplets.png> Hi

I am a phD student in Germany. recently I am focusing on the phase
separation property of a protein.
As tested the GFP tag can enhance the ability of the protein to phase
sepatate, thus forming bigger droplets.
So I can only purified and utilize the untaged protein for assays and
generate a series of two dimentional DIC images with round shapes.
now the problem is that I need to quantify the number, size and roundess
(aspect ratio), which requires a good program to segmentate.
I attached an image also in case you want to have a look. you can see that
the background is not even, maybe it is not qualified, Do I need to generate
new images?

does someone of you know how to process or have some good program, software,
or plug-in that I can apply.

Best wishes
Hui Zhang



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Robert Smith
Hello Dr. Zhang (you'll have to get used to it sooner or later),
The main problem is the angle of incidence of the light source being too steep. If illumination could be more normal to the surface and more evenly spread, it would be much better
ImageJ/Fiji is as good a software as any you will find to accomplish your goal.
Back to the illumination, because of the angle the droplets are highly illuminated on one side, and dark on the other leaving a terminator boundry between the two sides making a split . If you improve one side you decrease the other side and still maintain the split. The best option is to retake the images, diffuse the light and bring it more to the normal of the plane.
That being said, it can be done much better than I have here. I have included it only as an example as to what can be done.
Let us know if you will attempt again or just want to work with what you have.
Good Luck,
Bob


________________________________
From: Hui Zhang <[hidden email]>
Sent: Thursday, February 27, 2020 6:52 AM
To: [hidden email] <[hidden email]>
Subject: HELP!!!-segmentation and quantification of DIC images

<https://nam12.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382540%2FDIC-droplets.png&amp;data=02%7C01%7C%7Ced10770afbb14ef57ee208d7bb7e0a0f%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184022309147093&amp;sdata=pPYP6aCGdKotGIVWJNRrSmCI34GQ5W6ND570Qd6HQ6M%3D&amp;reserved=0> Hi

I am a phD student in Germany. recently I am focusing on the phase
separation property of a protein.
As tested the GFP tag can enhance the ability of the protein to phase
sepatate, thus forming bigger droplets.
So I can only purified and utilize the untaged protein for assays and
generate a series of two dimentional DIC images with round shapes.
now the problem is that I need to quantify the number, size and roundess
(aspect ratio), which requires a good program to segmentate.
I attached an image also in case you want to have a look. you can see that
the background is not even, maybe it is not qualified, Do I need to generate
new images?

does someone of you know how to process or have some good program, software,
or plug-in that I can apply.

Best wishes
Hui Zhang



--
Sent from: https://nam12.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7C%7Ced10770afbb14ef57ee208d7bb7e0a0f%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184022309147093&amp;sdata=K5BkLym4uyIz%2BZb4az%2BNkb6GM4ZmkgIfZcaj5n1SLAU%3D&amp;reserved=0

--
ImageJ mailing list: https://nam12.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7C%7Ced10770afbb14ef57ee208d7bb7e0a0f%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184022309147093&amp;sdata=QMKYbkgwjT1WWRIhJaGsq5m%2FzXNkRFxUZ6B%2BU4XhBD0%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

Result of Result of fltr.png (4K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Michael Schmid
In reply to this post by Hui Zhang
Hi Hui,

I guess that it can be done with a bit of programming:

(1) High pass filter to get the background flat
   https://imagejdocu.tudor.lu/plugin/filter/highpass_filter/start

(2) Programming:
For each diagonal line (in upward direction), replace the pixel value
with the running sum over the pixels from the beginning to the current
point. Then subtract the average of that line.
Alternatives without diagonal lines:
Either have illumination from the left and do the same with the "usual"
horizontal image lines,
or do the operation for columns and lines, and sum/subtract these two
results.


Michael
________________________________________________________________
On 27.02.20 12:52, Hui Zhang wrote:

> <http://imagej.1557.x6.nabble.com/file/t382540/DIC-droplets.png> Hi
>
> I am a phD student in Germany. recently I am focusing on the phase
> separation property of a protein.
> As tested the GFP tag can enhance the ability of the protein to phase
> sepatate, thus forming bigger droplets.
> So I can only purified and utilize the untaged protein for assays and
> generate a series of two dimentional DIC images with round shapes.
> now the problem is that I need to quantify the number, size and roundess
> (aspect ratio), which requires a good program to segmentate.
> I attached an image also in case you want to have a look. you can see that
> the background is not even, maybe it is not qualified, Do I need to generate
> new images?
>
> does someone of you know how to process or have some good program, software,
> or plug-in that I can apply.
>
> Best wishes
> Hui Zhang

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Hui Zhang
In reply to this post by Robert Smith
Hi Robert Smith,

As we can see from the result you gave, the droplets are all moon-like
shapes, which is not the truth and not appliable to me.

here I attached another image with better background.
could you please try again and see if the result could be better.

Best,
Hui Zhang

<http://imagej.1557.x6.nabble.com/file/t382540/DIC-droplets-2.jpg>



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Hui Zhang
In reply to this post by Michael Schmid
Hi Michael,

I did try the Highpass filter, and indeed, the image looks better.
But I really have a lack of programming, cann't accomplish the (2) part.

Best,
Hui Zhang



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Jim Peterson
In reply to this post by Hui Zhang
Rather than trying to flatten the background, you can look at gradients.
This also takes care of the light and shadow of each droplet.  The following
script fragment will do this:

run("8-bit");
run("Find Edges");
run("Despeckle");
run("Gaussian Blur...", "sigma=2");
setAutoThreshold("Otsu dark no-reset");
setOption("BlackBackground", true);
run("Convert to Mask");

Or you can run the commands from the interface:
Image/Type/8-bit
Process/Find Edges
Process/Filter/Gaussian Blur...  (sigma=2)
Image/Adjust/Threshold... (I used Otsu - you can try other methods, or set
the threshold manually)
    Then hit Apply.

The result is here:
<http://imagej.1557.x6.nabble.com/file/t381520/mask.png>

- Jim




--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Hui Zhang
Hi Ji,
I tried your program, the result looks better, but still we can see that the
shadow-cast effect is not excluded (all bright and dark region are chosen,
which is not the truth)

do you have some idea to remove the shadow-cast effect of DIC images?



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Jim Peterson
I was a little sloppy with the Gaussian blur.  You may be able to reduce the
radius or remove it all together, and may get more accurate resultsget
better results.

As for removing the shadows, I would separately identify the shadows and
then subtract them from the results from the original segmentation.  A
simple thresholding using Otsu with light background seems to do a decent
job of finding the shadows.  In the upper-left and lower-right corners you
pick up a little noise in the background due uneven illumination, but it
shouldn't be a problem since it is in the background.

<http://imagej.1557.x6.nabble.com/file/t381520/Screen_Shot_2020-02-28_at_7.png>

So I would suggest then the following:
    1. Do the segmentation described previously, resulting in a binary mask
of the objects plus shadows.
    2. Segment the shadows using simple Otsu threshold, resulting in a
shadow mask.
    3. Subtract the shadow mask from the original mask.
    4. Run particle analysis on the result to generate object properties.

- Jim



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: Hello again,

Robert Smith
In reply to this post by Hui Zhang
Dr. Zhang,
 Yes Dr. I understand. As I mentioned the image was only an example of what may be possible with a better image, which you supplied. So now I will see how much the technique can be improved. As I noted to you it is the terminator point which separated the two sides of the droplet making the segmentation difficult. This image is much improved than the first, yet I still can notice a termination between sides, however it may not be severe enough to interfere with what you wish.  I will give it some more work and get back with you soon.​
Bob

________________________________
From: ImageJ Interest Group <[hidden email]> on behalf of Hui Zhang <[hidden email]>
Sent: Thursday, February 27, 2020 12:40 PM
To: [hidden email] <[hidden email]>
Subject: Re: HELP!!!-segmentation and quantification of DIC images

Hi Robert Smith,

As we can see from the result you gave, the droplets are all moon-like
shapes, which is not the truth and not appliable to me.

here I attached another image with better background.
could you please try again and see if the result could be better.

Best,
Hui Zhang

<https://eur04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382540%2FDIC-droplets-2.jpg&amp;data=02%7C01%7C%7C0c9c62ed5e3e4fcd03ea08d7bbac2c6e%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184220455196638&amp;sdata=tkGK5aYZYzN5sdUUsd57W6kTxdFr7h8p3sweot45RDY%3D&amp;reserved=0>



--
Sent from: https://eur04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7C%7C0c9c62ed5e3e4fcd03ea08d7bbac2c6e%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184220455206643&amp;sdata=jY%2FCed3s22uZ%2BspzCWD0tJ42wkWm1PxvIGVXwS6p5ds%3D&amp;reserved=0

--
ImageJ mailing list: https://eur04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7C%7C0c9c62ed5e3e4fcd03ea08d7bbac2c6e%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184220455206643&amp;sdata=6MwFq01xvapgWsxvgRNbQRG153GgYDHTCcj7PhyjrRw%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Robert Smith
In reply to this post by Robert Smith
Dr. Zhang,
This is what I was able to accomplish with the new image. It will take longer to describe how it was done than it took to do it so I will send you a detailed process explaination before my day is ended.​
Keep all images....
Later,​
Bob

________________________________
From: ImageJ Interest Group <[hidden email]> on behalf of Robert Smith <[hidden email]>
Sent: Thursday, February 27, 2020 10:22 AM
To: [hidden email] <[hidden email]>
Subject: Re: HELP!!!-segmentation and quantification of DIC images

Hello Dr. Zhang (you'll have to get used to it sooner or later),
The main problem is the angle of incidence of the light source being too steep. If illumination could be more normal to the surface and more evenly spread, it would be much better
ImageJ/Fiji is as good a software as any you will find to accomplish your goal.
Back to the illumination, because of the angle the droplets are highly illuminated on one side, and dark on the other leaving a terminator boundry between the two sides making a split . If you improve one side you decrease the other side and still maintain the split. The best option is to retake the images, diffuse the light and bring it more to the normal of the plane.
That being said, it can be done much better than I have here. I have included it only as an example as to what can be done.
Let us know if you will attempt again or just want to work with what you have.
Good Luck,
Bob


________________________________
From: Hui Zhang <[hidden email]>
Sent: Thursday, February 27, 2020 6:52 AM
To: [hidden email] <[hidden email]>
Subject: HELP!!!-segmentation and quantification of DIC images

<https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382540%2FDIC-droplets.png&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206004254&amp;sdata=C4ziZ97xc%2BdaGDS0hMYD4wl4Qf5dqfUPTRI6gPQE2%2FE%3D&amp;reserved=0> Hi

I am a phD student in Germany. recently I am focusing on the phase
separation property of a protein.
As tested the GFP tag can enhance the ability of the protein to phase
sepatate, thus forming bigger droplets.
So I can only purified and utilize the untaged protein for assays and
generate a series of two dimentional DIC images with round shapes.
now the problem is that I need to quantify the number, size and roundess
(aspect ratio), which requires a good program to segmentate.
I attached an image also in case you want to have a look. you can see that
the background is not even, maybe it is not qualified, Do I need to generate
new images?

does someone of you know how to process or have some good program, software,
or plug-in that I can apply.

Best wishes
Hui Zhang



--
Sent from: https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206004254&amp;sdata=fDuxvh90OF0%2FkyqwKt975uIgKoSrefUgyATtcmM0FLQ%3D&amp;reserved=0

--
ImageJ mailing list: https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206014245&amp;sdata=5kiBCMpgvxIE%2BdlfWPrTP70rge1zdVQVnys0nG53%2B%2B0%3D&amp;reserved=0

--
ImageJ mailing list: https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206014245&amp;sdata=5kiBCMpgvxIE%2BdlfWPrTP70rge1zdVQVnys0nG53%2B%2B0%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

Drops 2result.png (4K) Download Attachment
drops 2 filter.png (1K) Download Attachment
2020-02-28 at 11-26-08.png (8K) Download Attachment
2020-02-28 at 11-29-17.png (10K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Michael Schmid
In reply to this post by Hui Zhang
Hi Hui,

as I might also need something like this in the future, I have quickly
programmed an ImageJ plugin for differential interference contrast
microscopy (DIC) images of sparse objects on a uniform background.

It is far from ideal, but should do the basic job to some degree, see

https://imagejdocu.tudor.lu/plugin/filter/dic_processor/start


Michael
________________________________________________________________
On 27.02.20 18:59, Hui Zhang wrote:
> Hi Michael,
>
> I did try the Highpass filter, and indeed, the image looks better.
> But I really have a lack of programming, cann't accomplish the (2) part.
>
> Best,
> Hui Zhang

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Kenneth Sloan-2
When I last had to deal with DIC images (mid 1980s), my most successful approach was to separately identify the light and dark halves (sometimes using, and sometimes deriving, the directionality) and then merging the two sets of objects into matched pairs.

Alas, this was long before FIJI existed, so I can't offer a plugin (I'm not sure I still have the code).  

Note that this required considerable processing one level above pure "image processing".

An alternative idea might be template matching (at a range of scales matching the expected range of sizes of your particles).

Finally...the boundaries separating the light halves from the dark halves are often the strongest edges in the image.  If you combine that with information on the directionality (or, perhaps estimate it), it might suffice to find and characterize (by length) these boundaries.  This may depend on how variable the objects are in their shape.  If you are able to assume they are all circular, then the separating boundary completely characterizes each object.

I'm very interested in Michael's new plugin - is it possible to get a quick gloss on the methods used?  My interest these days (actually, a potential future interest) is in "split-detector" adaptive optics images (which look like DIC to me, but the people who do this sort of thing don't seem to call it that).

--
Kenneth Sloan
[hidden email]
Vision is the art of seeing what is invisible to others.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Joel Sheffield
In reply to this post by Hui Zhang
The image arises from a DIC optical system.  DIC images of spherical
objects show the shadow like appearance that you see in each of the
particles.  It is often very difficult to get a uniform background in these
images as well.  So, the problem that you face is twofold.  1. smooth the
background, and 2. measure the particles, given that they are distorted by
the shadows.  ImageJ has some routines that might be helpful in this.  As
has already been suggested, you can use the FFT>bandpass filter to smooth
the background, since the background variation can be thought of as a long
wavelength curve of intensity, and so you are removing the low frequency
component of the image.  Measuring the particles is trickier.  However, it
seems to me that you can use the Process>filters>variance routine to make
the shadows equivalent from both sides.  The problem with this is that it
elongates the image.  On the other hand, since you would now have a
population of ellipses, you could threshold them and then use the Particle
Analyzer to look for the short axis of each ellipse.    These are just
ideas, since I haven't actually tried them out.


Best of luck,
Joel


Joel B. Sheffield, Ph.D
Department of Biology
Temple University
Philadelphia, PA 19122
Voice: 215 204 8839
e-mail: [hidden email]
Office:  Biolife 311
URL:  *https://bio.cst.temple.edu/~jbs/ <https://bio.cst.temple.edu/~jbs/>
<http://tinyurl.com/khbouft>*


On Thu, Feb 27, 2020 at 7:10 AM Hui Zhang <[hidden email]> wrote:

> <http://imagej.1557.x6.nabble.com/file/t382540/DIC-droplets.png> Hi
>
> I am a phD student in Germany. recently I am focusing on the phase
> separation property of a protein.
> As tested the GFP tag can enhance the ability of the protein to phase
> sepatate, thus forming bigger droplets.
> So I can only purified and utilize the untaged protein for assays and
> generate a series of two dimentional DIC images with round shapes.
> now the problem is that I need to quantify the number, size and roundess
> (aspect ratio), which requires a good program to segmentate.
> I attached an image also in case you want to have a look. you can see that
> the background is not even, maybe it is not qualified, Do I need to
> generate
> new images?
>
> does someone of you know how to process or have some good program,
> software,
> or plug-in that I can apply.
>
> Best wishes
> Hui Zhang
>
>
>
> --
> Sent from: http://imagej.1557.x6.nabble.com/
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Cammer, Michael-3
To flatten the background, I used
run("Bandpass Filter...", "filter_large=512 filter_small=0 suppress=None tolerance=5");
Then to find features, duplicated the results and did a low values threshold image on one and a high values threshold on the other, OR combined the two results, dilated and eroded.

However, real flatfield correction (image divided by out of focus or no features image) is an excellent (better?) way to clean up Nomarski images at the image collection stage.  

Cheers-

Michael Cammer, Sr Research Scientist, DART Microscopy Laboratory
NYU Langone Health, 540 First Avenue, SK2 Microscopy Suite, New York, NY  10016
Office: 646-501-0567 Cell: 914-309-3270  [hidden email]  
http://nyulmc.org/micros  http://microscopynotes.com/
 




-----Original Message-----
From: Joel Sheffield <[hidden email]>
Sent: Friday, February 28, 2020 12:36 PM
To: [hidden email]
Subject: Re: HELP!!!-segmentation and quantification of DIC images

[EXTERNAL]

The image arises from a DIC optical system.  DIC images of spherical objects show the shadow like appearance that you see in each of the particles.  It is often very difficult to get a uniform background in these images as well.  So, the problem that you face is twofold.  1. smooth the background, and 2. measure the particles, given that they are distorted by the shadows.  ImageJ has some routines that might be helpful in this.  As has already been suggested, you can use the FFT>bandpass filter to smooth the background, since the background variation can be thought of as a long wavelength curve of intensity, and so you are removing the low frequency component of the image.  Measuring the particles is trickier.  However, it seems to me that you can use the Process>filters>variance routine to make the shadows equivalent from both sides.  The problem with this is that it elongates the image.  On the other hand, since you would now have a population of ellipses, you could threshold them and then use the Particle
Analyzer to look for the short axis of each ellipse.    These are just
ideas, since I haven't actually tried them out.


Best of luck,
Joel


Joel B. Sheffield, Ph.D
Department of Biology
Temple University
Philadelphia, PA 19122
Voice: 215 204 8839
e-mail: [hidden email]
Office:  Biolife 311
URL:  *https://urldefense.proofpoint.com/v2/url?u=https-3A__bio.cst.temple.edu_-7Ejbs_&d=DwIBaQ&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=hUBj2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNnM&m=xWRj63kWDSxpj6udTH1_XBO099bBfuBK-g3VfBCutBc&s=i-OAu5WEs8lnoVRm-GPQ5aXTGdeTA2lnyPudb7mLotw&e=  <https://urldefense.proofpoint.com/v2/url?u=https-3A__bio.cst.temple.edu_-7Ejbs_&d=DwIBaQ&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=hUBj2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNnM&m=xWRj63kWDSxpj6udTH1_XBO099bBfuBK-g3VfBCutBc&s=i-OAu5WEs8lnoVRm-GPQ5aXTGdeTA2lnyPudb7mLotw&e= > <https://urldefense.proofpoint.com/v2/url?u=http-3A__tinyurl.com_khbouft&d=DwIBaQ&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=hUBj2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNnM&m=xWRj63kWDSxpj6udTH1_XBO099bBfuBK-g3VfBCutBc&s=E5bywrDcvns2JA0F2cZOqTs2Hw4K7RpWLqoRpZrYxYg&e= >*


On Thu, Feb 27, 2020 at 7:10 AM Hui Zhang <[hidden email]> wrote:

> <https://urldefense.proofpoint.com/v2/url?u=http-3A__imagej.1557.x6.na
> bble.com_file_t382540_DIC-2Ddroplets.png&d=DwIBaQ&c=j5oPpO0eBH1iio48Dt
> sedeElZfc04rx3ExJHeIIZuCs&r=hUBj2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNn
> M&m=xWRj63kWDSxpj6udTH1_XBO099bBfuBK-g3VfBCutBc&s=zeAa0NBgJIgxeWwUmDh_
> dLRYOR01qn8OM-mPR6kAsXc&e= > Hi
>
> I am a phD student in Germany. recently I am focusing on the phase
> separation property of a protein.
> As tested the GFP tag can enhance the ability of the protein to phase
> sepatate, thus forming bigger droplets.
> So I can only purified and utilize the untaged protein for assays and
> generate a series of two dimentional DIC images with round shapes.
> now the problem is that I need to quantify the number, size and
> roundess (aspect ratio), which requires a good program to segmentate.
> I attached an image also in case you want to have a look. you can see
> that the background is not even, maybe it is not qualified, Do I need
> to generate new images?
>
> does someone of you know how to process or have some good program,
> software, or plug-in that I can apply.
>
> Best wishes
> Hui Zhang
>
>
>
> --
> Sent from:
> https://urldefense.proofpoint.com/v2/url?u=http-3A__imagej.1557.x6.nab
> ble.com_&d=DwIBaQ&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=hUBj
> 2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNnM&m=xWRj63kWDSxpj6udTH1_XBO099bB
> fuBK-g3VfBCutBc&s=UfHaeI5QmjGbtwjgyghnIyFpcBSDk-iOPly14DEmGUU&e=
>
> --
> ImageJ mailing list:
> https://urldefense.proofpoint.com/v2/url?u=http-3A__imagej.nih.gov_ij_
> list.html&d=DwIBaQ&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=hUB
> j2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNnM&m=xWRj63kWDSxpj6udTH1_XBO099b
> BfuBK-g3VfBCutBc&s=Vu7nhMAzYdSDu1RLywtaiQ1kYam-2pcnt-bTINkWnlc&e=
>

--
ImageJ mailing list: https://urldefense.proofpoint.com/v2/url?u=http-3A__imagej.nih.gov_ij_list.html&d=DwIBaQ&c=j5oPpO0eBH1iio48DtsedeElZfc04rx3ExJHeIIZuCs&r=hUBj2D5n6oKThx2L01qn8IORZb5f-ruLVXPmQ1zQNnM&m=xWRj63kWDSxpj6udTH1_XBO099bBfuBK-g3VfBCutBc&s=Vu7nhMAzYdSDu1RLywtaiQ1kYam-2pcnt-bTINkWnlc&e=

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: Hello Again Dr.

Robert Smith
In reply to this post by Robert Smith
Open your image, go to Image >Type > 8bit, go to Plugins > Polynomial Shading Corrector- set to 4-4-2 in the dialog box click run,

Open File > New > Image- set to 8bit, same size run.​

Use rectangle tool > pick an area with no bubbles and obtain section > Edit > Copy, then paste onto the new image....The size of the sample does not matter. An example of how to pick is included.​

Go to Process > Filters > Gaussian blur- set to 4.0.  Get color picker (eye dropper) from tool bar and pick within the image. To prevent unintentional accidents, switch to the rectangle tool again and select the area you have just made in the new image > Edit > Clear > Invert​

Pick the Flood fill tool, (pouring bucket) and click into the new image. You have just made a filter.​

Go to Process > Image Calculator-- Set as shown in example, Click O.K.​

Go to Image > Adjust > Threshold-- Set as shown in example, Click O.K.​

Go to Process > Binary > Options-- Set as shown in example, Click O.K.​

You are now done!​
Good Luck,​
Bob

________________________________
From: ImageJ Interest Group <[hidden email]> on behalf of Robert Smith <[hidden email]>
Sent: Thursday, February 27, 2020 10:22 AM
To: [hidden email] <[hidden email]>
Subject: Re: HELP!!!-segmentation and quantification of DIC images

Hello Dr. Zhang (you'll have to get used to it sooner or later),
The main problem is the angle of incidence of the light source being too steep. If illumination could be more normal to the surface and more evenly spread, it would be much better
ImageJ/Fiji is as good a software as any you will find to accomplish your goal.
Back to the illumination, because of the angle the droplets are highly illuminated on one side, and dark on the other leaving a terminator boundry between the two sides making a split . If you improve one side you decrease the other side and still maintain the split. The best option is to retake the images, diffuse the light and bring it more to the normal of the plane.
That being said, it can be done much better than I have here. I have included it only as an example as to what can be done.
Let us know if you will attempt again or just want to work with what you have.
Good Luck,
Bob


________________________________
From: Hui Zhang <[hidden email]>
Sent: Thursday, February 27, 2020 6:52 AM
To: [hidden email] <[hidden email]>
Subject: HELP!!!-segmentation and quantification of DIC images

<https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382540%2FDIC-droplets.png&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206004254&amp;sdata=C4ziZ97xc%2BdaGDS0hMYD4wl4Qf5dqfUPTRI6gPQE2%2FE%3D&amp;reserved=0> Hi

I am a phD student in Germany. recently I am focusing on the phase
separation property of a protein.
As tested the GFP tag can enhance the ability of the protein to phase
sepatate, thus forming bigger droplets.
So I can only purified and utilize the untaged protein for assays and
generate a series of two dimentional DIC images with round shapes.
now the problem is that I need to quantify the number, size and roundess
(aspect ratio), which requires a good program to segmentate.
I attached an image also in case you want to have a look. you can see that
the background is not even, maybe it is not qualified, Do I need to generate
new images?

does someone of you know how to process or have some good program, software,
or plug-in that I can apply.

Best wishes
Hui Zhang



--
Sent from: https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206004254&amp;sdata=fDuxvh90OF0%2FkyqwKt975uIgKoSrefUgyATtcmM0FLQ%3D&amp;reserved=0

--
ImageJ mailing list: https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206014245&amp;sdata=5kiBCMpgvxIE%2BdlfWPrTP70rge1zdVQVnys0nG53%2B%2B0%3D&amp;reserved=0

--
ImageJ mailing list: https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7C%7Cb2cb599532d0446bf4f508d7bb9941ba%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637184139206014245&amp;sdata=5kiBCMpgvxIE%2BdlfWPrTP70rge1zdVQVnys0nG53%2B%2B0%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

2020-02-28 at 16-58-49.png (9K) Download Attachment
Example area.png (305K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Till Bretschneider-2
In reply to this post by Hui Zhang
You can try our QuimP plugin for ImageJ which has a DIC preprocessing plugin. I tried it on your example and it works reasonably well (Decay factor 0.2, all other parameters unchanged).

Using QuimP you can also measure the size of your droplets.
The main question is why not use the fluorescence channel if your droplets are GFP tagged.
Also, it looks like the quality of your DIC image can be improved.
Best wishes,
Till

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Joel Sheffield
Do you have a link to this plugin?


Joel B. Sheffield, Ph.D
Department of Biology
Temple University
Philadelphia, PA 19122
Voice: 215 204 8839
e-mail: [hidden email]
Office:  Biolife 311
URL:  *https://bio.cst.temple.edu/~jbs/ <https://bio.cst.temple.edu/~jbs/>
<http://tinyurl.com/khbouft>*


On Sat, Feb 29, 2020 at 9:08 AM Till Bretschneider <
[hidden email]> wrote:

> You can try our QuimP plugin for ImageJ which has a DIC preprocessing
> plugin. I tried it on your example and it works reasonably well (Decay
> factor 0.2, all other parameters unchanged).
>
> Using QuimP you can also measure the size of your droplets.
> The main question is why not use the fluorescence channel if your droplets
> are GFP tagged.
> Also, it looks like the quality of your DIC image can be improved.
> Best wishes,
> Till
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Till Bretschneider-2
In reply to this post by Hui Zhang
Hi Joel,

QuimP is available through the Fiji updater which is the easiest way to install it.

Help->Update
Click "Manage Update Sites" and tick "QuimP" in the list.
Apply changes and restart Fiji.

Other ways to install it are described on the homepage: https://warwick.ac.uk/fac/sci/dcs/people/till_bretschneider/quimp

In Fiji you find the DIC filter installed under Plugins->QuimP-19.08.01->DIC

The method used is described in Zvi Kam. Microscopic differential interference contrast image processing by line integration (LID) and deconvolution. Bioimaging, 6(4):166–176, 1998.

Some rudimentary help is here: https://pilip.lnx.warwick.ac.uk/docs/master/QuimP_Guide.html#x1-250001

In terms of segmentation there are a number of options:
For images like the vesicle example some simple form of thresholding might do, or you can produce a mask based on thresholding which feeds into a random-walker method for segmentation (Plugins->QuimP-19.08.01->RandomWalk). If clusters of vesicles need to be resolved you would need to manually provide seeds for each particle in the cluster, or use a watershed method afterwards. The Boa plugin gives you full control over segmentation with the possibility to manually correct. QuimP contains a number of tools for detailed time-series analyses of shape changes.

We have tested quite a few different approaches to processing DIC images. Although, there are more sophisticated ones than the one by Kam, we found that in practice the latter is very reliable and essentially has only one free parameter that needs tuning.
Most people will record DIC images alongside fluorescence because usually they come for free. However, very few take care and have the skills to actually optimise their DIC setup. Therefore, ,ost DIC images I got to see are of very poor quality. If possible, detailed shape analyses should be based on fluorescence and 3D optical sectioning.

Best wishes,
Till

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: HELP!!!-segmentation and quantification of DIC images

Hui Zhang
In reply to this post by Till Bretschneider-2
Hi Till,

As I tested in my case, the GFP tag could effect the size of droplets
compared with untagged protein. So I plan latter to do all such experiments
using the wild type and mutated proteins with no tags (maybe there are some
other tags for choice, but really time comsumming to test).
as for the quality of DIC images, I have no clear idea which kind is the
best, what I could image is merely clear droplets with even background,
which I am trying to improve recently.

Best,
Hui



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html