How to quantify a regular arrangement on my images?

[lechristophe]

Dear ImageJists,

I have images from immunocytochemistry (proteins labeled with an antibody
in a cell culture), and I'd like to quantify a striking feature: one of the
proteins is periodically organized (see Image1.tif), whereas the other
seems to be more randomly clusterized (see Image2.tif). The exemple images
are a crop of the whole image, but basically this is the pattern for the
whole labeled structure.

I'd like to analyze and quantify how "band-like" is the pattern, i.e. a
number that would be a measure of the labeling disorder, from completely
periodic to completely random. I don't know how to do it. I tried the
OrientationJ and Directionality plugins, but they seem to find orientations
in both cases, and I'm more interested in the intensity of the organization
than in the directions of the pattern.

If you have any advice on how to do this, that would be great!

Thanks for your help,

Christophe

P.S. In case the attachments are stripped, you can find the images here:
Image1.tif: http://www.cleterrier.net/up/Image1.tif
Image2.tif: http://www.cleterrier.net/up/Image2.tif

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

[Herbie-3]

Christophe.

that's a nice one for Fourier-Analysis!

Apply "FFT" to your images which gives you the log of the Power Spectra
as 8bit images. You may now extract features from these.

One of such features could be the integral along straight lines through
the spectral origin as a function of the angle. Another could be the
dual function, i.e. the integral on circles around the spectral origin
as a function of their radii. There are PlugIns for both features. More
precision is obtained if you use 32bit spectra.

HTH

Herbie

::::::::::::::::::::::::::::::::::::::::::::::
On 05.02.14 17:11, Christophe Leterrier wrote:

> Dear ImageJists,
>
> I have images from immunocytochemistry (proteins labeled with an antibody
> in a cell culture), and I'd like to quantify a striking feature: one of the
> proteins is periodically organized (see Image1.tif), whereas the other
> seems to be more randomly clusterized (see Image2.tif). The exemple images
> are a crop of the whole image, but basically this is the pattern for the
> whole labeled structure.
>
> I'd like to analyze and quantify how "band-like" is the pattern, i.e. a
> number that would be a measure of the labeling disorder, from completely
> periodic to completely random. I don't know how to do it. I tried the
> OrientationJ and Directionality plugins, but they seem to find orientations
> in both cases, and I'm more interested in the intensity of the organization
> than in the directions of the pattern.
>
> If you have any advice on how to do this, that would be great!
>
> Thanks for your help,
>
> Christophe
>
> P.S. In case the attachments are stripped, you can find the images here:
> Image1.tif: http://www.cleterrier.net/up/Image1.tif
> Image2.tif: http://www.cleterrier.net/up/Image2.tif
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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

[Leon Espinosa-3]

Dear Christophe, I have tried an approach based on the eucledian distance and skeleton analysis. The idea is that the "band pattern" should generate bands structures in an eucledian image map and will show longer branchs in the skeleton structure:

- make intensity mask
- make the exact eucledian distance map (in fiji plugin ; process/signed eucledian exact diastance)
- make a mask with a distance threshold (the same for both conditions)
- skeletonize
- G. Landini plugin Morphologie/Binary connectivity
- Quantify the ratio branch/total (= threshold(3,3) / threshold(1,255) )

In my hands with the exemple images I have found


image1
1113/1524 = 0.73

image2
534/834 = 0.64

delta=13%

Other measures of the branch objects (Binary connectivity, threshold(3,3)):

size
image1 = 5.95 (sd = 5.23)
image2 = 4.76 (sd = 4.27)
delta=20%

AR (espect ratio)
image1 = 3,36 (sd 1,978)
image2 = 2,86 (sd 1,639)
delta=15%


Hope it helps...

Leon





Leon Espinosa
[hidden email]

Plateforme de Biophotonique Appliquée à la Microbiologie

LCB CNRS UMR7283
31, Chemin Joseph AIGUIER
13009 MARSEILLE
Tel : [phone]
Cell : 06 79 25 97 40
Tel sécr.: 04 91 16 40 77
Fax : [phone]


Le 5 févr. 2014 à 17:11, Christophe Leterrier a écrit :

> Dear ImageJists,
>
> I have images from immunocytochemistry (proteins labeled with an antibody
> in a cell culture), and I'd like to quantify a striking feature: one of the
> proteins is periodically organized (see Image1.tif), whereas the other
> seems to be more randomly clusterized (see Image2.tif). The exemple images
> are a crop of the whole image, but basically this is the pattern for the
> whole labeled structure.
>
> I'd like to analyze and quantify how "band-like" is the pattern, i.e. a
> number that would be a measure of the labeling disorder, from completely
> periodic to completely random. I don't know how to do it. I tried the
> OrientationJ and Directionality plugins, but they seem to find orientations
> in both cases, and I'm more interested in the intensity of the organization
> than in the directions of the pattern.
>
> If you have any advice on how to do this, that would be great!
>
> Thanks for your help,
>
> Christophe
>
> P.S. In case the attachments are stripped, you can find the images here:
> Image1.tif: http://www.cleterrier.net/up/Image1.tif
> Image2.tif: http://www.cleterrier.net/up/Image2.tif
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
> <Image1.tif><Image2.tif>

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