Holes within particles

[Bill Mohler]

After considerable time puzzling it over, I'm concluding that Analyze
Particles with the Include Holes option unchecked actually *does not*
properly identify particles within particles (as portrayed in the wiki
lesson
http://imagejdocu.tudor.lu/doku.php?id=gui:analyze:analyze_particles).  
To be more specific and correct, it *can* find particles that reside
within holes inside of a larger particle.  However, it does not define
the boundary of the hole as an inner edge of the larger particle that
contains it.  In other words, a particle with a hole in it is not
portrayed as a hollow ROI, but rather as a solid ROI with no hint of the
inner hole.

Is this a known and unavoidable feature?  Is there a setting that I
haven't found yet that solves this problem?

I'm attaching a jpeg to demonstrate the problem.  ROI edges are in
cyan.  Look very closely at the Large ROIs that contain holes, and you
will notice that there is no cyan outline around the outer edge of a
hole: for example, look at the black region surrounding ROIs 9-11; it is
actually a hole withing the huge ROI 2, but the outer edges of the black
are not rimmed in cyan, because it has not been recognized.  But any
particles within a hole ( like 9-11, which are above my 300pix size
cutoff) are properly outlined.

I really need to work out a fix for this problem, because these
arrangements are very common in my images.

Thanks for any help,
Bill


--

William A. Mohler

Associate Professor
Dept. of Genetics and Developmental Biology
University of Connecticut Health Center

*Physical Address:***

Room R1159

Cell and Genome Sciences Building

400 Farmington Ave.

Farmington, CT

*Mail Address:*

MC-6403

263 Farmington Ave.

Farmington, CT06030-6403

[hidden email] <mailto:[hidden email]>
*Mobile: [phone]*

skype: wmohler
Office: [phone], room R1159
Lab: [phone], room R1265
Fax: [phone]

http://genetics.uchc.edu/Faculty/assoc_professors/mohler.html
PThink before you print

[Bill Mohler]

Gabriel-

Your Binary Reconstruction plugin does work beautifully well to rebuild
a giant topologically complex particle from the binary mask of all
particles!  And Edit>Selection>Create Selection followed by
Edit>Selection>Make Inverse appears to give me back a single
topologically complex ROI that bounds the outer and inner edges of this
huge holey blob correctly, and gives me area measurement identical to
the initially recognized particle (used for the seed) but a perimeter
that is much larger, presumably because it now counts the distances
around the inner holes.

This really does seem to be the solution of my problem.  Many many thanks!
Bill

On 12/3/10 11:02 AM, Bill Mohler wrote:

> Sounds like it definitely will!  I'll experiment right now and let you
> know.
>
>
> On 12/3/10 10:58 AM, Gabriel Landini wrote:
>> On Friday 03 Dec 2010 10:07:16 you wrote:
>>> Here's this binary tif file. Thanks for looking it over!
>>> I'm pasting in some discussion I had with Wayne last night...
>>> ...I think I see what you're telling me when I compare the Results table
>>> entries for huge area 2 to the area of the ROI 2 in the sample image that I
>>> sent you a screen shot of.� The table entry has a mean of 255 and an area
>>> of 612115, while the ROI has mean 214 and area 733701, clearly containing
>>> black patches.� What's different about the algorithm defining the measured
>>> area versus the ROI boundaries?� What's the obstacle(s) to using the
>>> measurement algorithm to define the ROI?� I'm naive, as usual, here.
>> Ah, yes, I see what you mean. I do not think there is a way of relating the
>> outside ROI to the inside ROIs you see with Create Selection because the ROIs
>> are just traces following a boundary. They are not logically related as to
>> which particle they belong to.
>>
>> The proper way to manipulate the individual objects, is to use the pixel
>> located at each XStart YStart coordinates in the results table and binary-
>> reconstruction. So you do a binary reconstruction (plugin is in the Morphology
>> collection in my page) of an empty image of the same size of the original
>> containing only the XStart YStart pixel (from the Results table) as the "seed"
>> and the original image as the "mask". That returns that  particle alone.
>>
>> Another simpler although not elegant way is to label all the regions with the
>> Binary_Label plugin (which will return a different greyscale value for each
>> particle and also in the morphology.zip) and threshold for that particular
>> value.
>>
>> If you need an example of this, see the KeepParticlesInRange.txt macro in the
>> morphology.zip
>>
>> I hope this helps!
>> Cheers
>>
>> Gabriel
>>
>> .
>>
>
> --
>
> William A. Mohler
>
> Associate Professor
> Dept. of Genetics and Developmental Biology
> University of Connecticut Health Center
>
> *Physical Address:***
>
> Room R1159
>
> Cell and Genome Sciences Building
>
> 400 Farmington Ave.
>
> Farmington, CT
>
> *Mail Address:*
>
> MC-6403
>
> 263 Farmington Ave.
>
> Farmington, CT06030-6403
>
> [hidden email] <mailto:[hidden email]>
> *Mobile: [phone]*
>
> skype: wmohler
> Office: [phone], room R1159
> Lab: [phone], room R1265
> Fax: [phone]
>
> http://genetics.uchc.edu/Faculty/assoc_professors/mohler.html
> PThink before you print
>

--

William A. Mohler

Associate Professor
Dept. of Genetics and Developmental Biology
University of Connecticut Health Center

*Physical Address:***

Room R1159

Cell and Genome Sciences Building

400 Farmington Ave.

Farmington, CT

*Mail Address:*

MC-6403

263 Farmington Ave.

Farmington, CT06030-6403

[hidden email] <mailto:[hidden email]>
*Mobile: [phone]*

skype: wmohler
Office: [phone], room R1159
Lab: [phone], room R1265
Fax: [phone]

http://genetics.uchc.edu/Faculty/assoc_professors/mohler.html
PThink before you print