Phase contrast image segmentation for counting fibroblasts

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

Phase contrast image segmentation for counting fibroblasts

sfreema1
Good morning,

I am trying to find a workflow for counting cells in a phase contrast image.
These are fibroblasts that I grew in a 96-well plate. Several images were
stitched together.

As I look through other similar questions here and elsewhere, I find a lot
of the examples are geared towards more circular cells. I have tried using
the common workflow (thresholding/binarizing -> analyze particles) but I
have not been content with my results.

Because the images were taken in the 96-well plate, all my phase contrast
images have a feature where the centers of the wells are darker and have
more contrast, while as I move away from center, the features brighten and
lose contrast.

I found that Sharpen->Find Edges was helpful in outlining the cells and
apparently removing the uneven illumination, but then I have not been able
to proceed from there.

I feel another issue with these cells is that they crowd very closely
together, and I think it is hard to distinguish two separate cells even by
eye sometimes.

I would appreciate insight anyone has to give. I have attached a sample
image below from my dataset. A12_-1_1_1_Stitched[Phase_Contrast]_001.tif
<http://imagej.1557.x6.nabble.com/file/t382700/A12_-1_1_1_Stitched%5BPhase_Contrast%5D_001.tif>  



--
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: Phase contrast image segmentation for counting fibroblasts

Krs5
Dear sfreema1,

To improve your image you can use a Gaussian Blur filter (Process>Filters>Gaussian Blur...) of 8 and divide your image with the filtered image using the image calculator (Process>Image Calculator...). The macro code below will do the same on an open image

title = getTitle();
run("Duplicate...", "title=filter");
run("Gaussian Blur...", "sigma=8");
imageCalculator("Divide create 32-bit", title,"filter");
selectWindow("Result of A12_-1_1_1_Stitched[Phase_Contrast]_001.tif");

You can play around with the type of filter or filter settings to see if you can get a better result.
I don't think you really will be able to count the number of cells (others might have better ideas) but it might be possible to collect information about the area covered by cells using the Labkit plugin - https://imagej.net/Labkit

Best wishes

Kees


Dr Ir K.R. Straatman

Advanced Imaging Facility

University of Leicester
www.le.ac.uk/advanced-imaging-facility<http://www.le.ac.uk/advanced-imaging-facility>

________________________________
From: sfreema1 <[hidden email]>
Sent: 28 July 2020 13:47
Subject: Phase contrast image segmentation for counting fibroblasts

Good morning,

I am trying to find a workflow for counting cells in a phase contrast image.
These are fibroblasts that I grew in a 96-well plate. Several images were
stitched together.

As I look through other similar questions here and elsewhere, I find a lot
of the examples are geared towards more circular cells. I have tried using
the common workflow (thresholding/binarizing -> analyze particles) but I
have not been content with my results.

Because the images were taken in the 96-well plate, all my phase contrast
images have a feature where the centers of the wells are darker and have
more contrast, while as I move away from center, the features brighten and
lose contrast.

I found that Sharpen->Find Edges was helpful in outlining the cells and
apparently removing the uneven illumination, but then I have not been able
to proceed from there.

I feel another issue with these cells is that they crowd very closely
together, and I think it is hard to distinguish two separate cells even by
eye sometimes.

I would appreciate insight anyone has to give. I have attached a sample
image below from my dataset. A12_-1_1_1_Stitched[Phase_Contrast]_001.tif
<https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382700%2FA12_-1_1_1_Stitched%255BPhase_Contrast%255D_001.tif&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C08c703362faf4bf6278a08d832f68dcc%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C637315381818990765&amp;sdata=yfNiHKkzoHm%2BKIwrBAAd9ZyLs50UeaqT2SwuUax9O00%3D&amp;reserved=0>



--
Sent from: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C08c703362faf4bf6278a08d832f68dcc%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C637315381818990765&amp;sdata=rxvSj3RksGjDaC3lqblkrTwSrrVbecKSjOR8ZU%2FERNM%3D&amp;reserved=0

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C08c703362faf4bf6278a08d832f68dcc%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C637315381818990765&amp;sdata=tZBsub4Fa3JuW6zw9OWL24cRDpuBIST6vrzaS3Bz8AI%3D&amp;reserved=0

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

Re: Phase contrast image segmentation for counting fibroblasts

foxtango
Dear sfreema1,

I do not understand what you mean by stitched. It looks as if the cellular
structures run continuously across the vertical and horizontal shadows in
the image. The darkening is in the center only indicating that the image is
single exposure.

Best, Thomas

Am Di., 28. Juli 2020 um 18:09 Uhr schrieb Straatman, Kees (Dr.) <
[hidden email]>:

> Dear sfreema1,
>
> To improve your image you can use a Gaussian Blur filter
> (Process>Filters>Gaussian Blur...) of 8 and divide your image with the
> filtered image using the image calculator (Process>Image Calculator...).
> The macro code below will do the same on an open image
>
> title = getTitle();
> run("Duplicate...", "title=filter");
> run("Gaussian Blur...", "sigma=8");
> imageCalculator("Divide create 32-bit", title,"filter");
> selectWindow("Result of A12_-1_1_1_Stitched[Phase_Contrast]_001.tif");
>
> You can play around with the type of filter or filter settings to see if
> you can get a better result.
> I don't think you really will be able to count the number of cells (others
> might have better ideas) but it might be possible to collect information
> about the area covered by cells using the Labkit plugin -
> https://imagej.net/Labkit
>
> Best wishes
>
> Kees
>
>
> Dr Ir K.R. Straatman
>
> Advanced Imaging Facility
>
> University of Leicester
> www.le.ac.uk/advanced-imaging-facility<
> http://www.le.ac.uk/advanced-imaging-facility>
>
> ________________________________
> From: sfreema1 <[hidden email]>
> Sent: 28 July 2020 13:47
> Subject: Phase contrast image segmentation for counting fibroblasts
>
> Good morning,
>
> I am trying to find a workflow for counting cells in a phase contrast
> image.
> These are fibroblasts that I grew in a 96-well plate. Several images were
> stitched together.
>
> As I look through other similar questions here and elsewhere, I find a lot
> of the examples are geared towards more circular cells. I have tried using
> the common workflow (thresholding/binarizing -> analyze particles) but I
> have not been content with my results.
>
> Because the images were taken in the 96-well plate, all my phase contrast
> images have a feature where the centers of the wells are darker and have
> more contrast, while as I move away from center, the features brighten and
> lose contrast.
>
> I found that Sharpen->Find Edges was helpful in outlining the cells and
> apparently removing the uneven illumination, but then I have not been able
> to proceed from there.
>
> I feel another issue with these cells is that they crowd very closely
> together, and I think it is hard to distinguish two separate cells even by
> eye sometimes.
>
> I would appreciate insight anyone has to give. I have attached a sample
> image below from my dataset. A12_-1_1_1_Stitched[Phase_Contrast]_001.tif
> <
> https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2Ffile%2Ft382700%2FA12_-1_1_1_Stitched%255BPhase_Contrast%255D_001.tif&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C08c703362faf4bf6278a08d832f68dcc%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C637315381818990765&amp;sdata=yfNiHKkzoHm%2BKIwrBAAd9ZyLs50UeaqT2SwuUax9O00%3D&amp;reserved=0
> >
>
>
>
> --
> Sent from:
> https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C08c703362faf4bf6278a08d832f68dcc%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C637315381818990765&amp;sdata=rxvSj3RksGjDaC3lqblkrTwSrrVbecKSjOR8ZU%2FERNM%3D&amp;reserved=0
>
> --
> ImageJ mailing list:
> https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C08c703362faf4bf6278a08d832f68dcc%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C637315381818990765&amp;sdata=tZBsub4Fa3JuW6zw9OWL24cRDpuBIST6vrzaS3Bz8AI%3D&amp;reserved=0
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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

Re: Phase contrast image segmentation for counting fibroblasts

sfreema1
Thank you for the response. By stitched, I mean that system took several
images and then combined arranged them together to create a larger image. In
this case, the objective was 10X and the system took 16 images and arranged
them in a grid to create an images of an overall larger area than a single
image.

I am told that the darkness in the center is a common observance when taking
phase contrast images in a 96-well plate.



--
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: Phase contrast image segmentation for counting fibroblasts

Gabriel Landini
On Wednesday, 29 July 2020 02:54:40 BST sfreema1 wrote:
> I am told that the darkness in the center is a common observance when taking
> phase contrast images in a 96-well plate.
~
The image quality needs improving. I would start by learning how to adjust the
phase rings and how to do background illumination correction.
Without those two issues resolved you will struggle to get any data out.

To answer the original question, here is a method we worked on to count cells
in phase contrast images:
https://onlinelibrary.wiley.com/doi/pdf/10.1111/jmi.12726

Cheers

Gabriel

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

Re: Phase contrast image segmentation for counting fibroblasts

Aryeh Weiss
On 29/07/2020 15:40, Gabriel Landini wrote:

> On Wednesday, 29 July 2020 02:54:40 BST sfreema1 wrote:
>> I am told that the darkness in the center is a common observance when taking
>> phase contrast images in a 96-well plate.
> ~
> The image quality needs improving. I would start by learning how to adjust the
> phase rings and how to do background illumination correction.
> Without those two issues resolved you will struggle to get any data out.
>
> To answer the original question, here is a method we worked on to count cells
> in phase contrast images:
> https://onlinelibrary.wiley.com/doi/pdf/10.1111/jmi.12726
>
> Cheers
>
> Gabriel
>
> --

I will add that in 96 well plates, the meniscus displaces the image of
the phase ring. That is why the center of your image has better phase
contrast, and the edges are more like brightfield images. The Biovoxxel
Psuedo Flat Field Correction does a nice job on correcting the
background (and there are other methods that have been suggested a work
well). If you run Fiji you can add the Biovoxxel update site in the updater.

If you want to do phase contrast, I suggest that you restrict yourself
to the center of the well, and if possible also reduce the cell density
if possible.

--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: 972-3-7384051

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