Surface Area Confusion

[victoria.gonzalez]

Hello Fellow ImageJ-ers!
I am pretty new to ImageJ, and I've been combing the thread looking for some help. I am working with photographs of children's faces attempting to measure the surface areas of their freckles and moles as previously defined and delineated by a super special UV camera my department purchased.  The problem I seem to  be running into is that when I try to get ImageJ to calculate the surface areas of these pre-delineated "spots", ImageJ often includes "normal" skin or unwanted areas.  How can I avoid this?/Can I avoid this?  I'd really appreciate any help anyone has to offer in terms of suggestions or plugins or other objects I might want to beat my head against (the wall has gotten old).

Thanks!
Victoria Gonzalez

[e-mail]

[Gabriel Landini]

On Wednesday 08 Dec 2010 14:45:13 you wrote:
> problem I seem to  be running into is that when I try to get ImageJ to
> calculate the surface areas of these pre-delineated "spots", ImageJ often
> includes "normal" skin or unwanted areas.  How can I avoid this?

Can you post some images, so we can see what you mean?

Cheers

Gabriel

[victoria.gonzalez]

A couple of people have requested a sample image, so here it is.
This is the UV channel and the camera/software delineates certain spots in yellow that it deems UV damage.  We would like to determine the surface area of only these particular spots.



If anyone has trouble with the image above, shoot me an email and I can email you a larger image.

What I have tried doing so far is converting the image to 8-bit, and then making it binary and then running "Analyze particles".

I also tried using Adobe Photoshop's magic wand tool and using their histogram tool to give me a pixel count of the delineated spots.

Cheers,
Victoria

[e-mail]

[Gabriel Landini]

On Thursday 09 December 2010 09:44:00 you wrote:
> A couple of people have requested a sample image, so here it is.
> This is the UV channel and the camera/software delineates certain spots in
> yellow that it deems UV damage.  We would like to determine the surface area
> of only these particular spots.

> What I have tried doing so far is converting the image to 8-bit, and then
> making it binary and then running "Analyze particles".

Using jpegs for imaging is not a good idea as they have lots of nasty
artifacts. That will prevent from segmenting the spots with certainty.

Also, it is not clear if the yellow areas were added by you or they are yellow
in the original image
It the yellow is originally there, you could try the colour threshold in
HSB mode, but the jpeg posted has lots of artifacts, so it is difficult to
know if that would work.

Cheers
Gabriel

[karo03]

Hi, for the yellowish regions it could be helpful to look into the hue h of Lch. h=atan(a,b) or h=atan(b/a) from Lab:
run("Lab Stack"); /* a plugin from ImageJ website */
run("Stack to Images");
imageCalculator("Divide create 32-bit", "b","a");
selectWindow("Result of b");
run("Macro...", "code=v=atan(v)");

Of course Gabriel's remark is valid, as you will see!
Regards
Kasten

Am 09.12.2010 um 18:44 schrieb victoria.gonzalez:

> A couple of people have requested a sample image, so here it is.
> This is the UV channel and the camera/software delineates certain spots in
> yellow that it deems UV damage.  We would like to determine the surface area
> of only these particular spots.
>
> http://imagej.588099.n2.nabble.com/file/n5819009/UV_Overlay.jpg 
>
> If anyone has trouble with the image above, shoot me an email and I can
> email you a larger image.
>
> What I have tried doing so far is converting the image to 8-bit, and then
> making it binary and then running "Analyze particles".
>
> I also tried using Adobe Photoshop's magic wand tool and using their
> histogram tool to give me a pixel count of the delineated spots.
>
> Cheers,
> Victoria
>
> [hidden email]
>
> --
> View this message in context: http://imagej.588099.n2.nabble.com/Surface-Area-Confusion-tp5817116p5819009.html
> Sent from the ImageJ mailing list archive at Nabble.com.

Karsten
[hidden email]