Subject: | |
From: | |
Reply To: | |
Date: | Tue, 15 Jan 2002 21:51:45 +0100 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Aryeh,
what is the morphological opening operator?
Joachim
Aryeh M. Weiss wrote:
> Search the CONFOCAL archive at
> http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
>
> Wes Wallace wrote:
>
>>Search the CONFOCAL archive at
>>http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
>>
>>for the first time i am seriously trying to measure the size of a
>>fluorescent structure. i am realizing to my surprize that the size depends
>>on the contrast adjustment! for example if i set the contrast fairly
>>light i can see some structures well, but others are too faint to detect.
>>at a darker setting, the faint structures are detectable but now the
>>darker structures begin to grow larger and even blur together.
>>
>>i infer from this that not only fluorescence intensity measurements but
>>also size measurements depend on calibrating the fluorescence intensity!
>>
>>has anyone ever dealt with this issue, and if so, how do you calibrate the
>>contrast settings on different images so that measurements are comparable?
>>
>>i don't think that using identical contrast & brightness settings for each
>>image during acquisition is going to work, for two reasons: (1) the
>>fluorescent structures are quite variable in terms of absolute
>>brightness, due to variability in staining; (2) the confocal itself is
>>probably not giving exactly the same laser power on different days even if
>>the contrast and brightness settings are identical.
>>
>>any ideas???
>>
>>
> I see that there have been a few replies to this concerning calibration
> standards,
> beads, and such, but I think that is not the main problem here. You have
> not specified the nature of your structure, but I assume it is something
> that is localized (as apposed to diffuse) and sits on top of some
> diffuse background. The backround is probably not completely uniform,
> and it is hard to correct by subtraction alone if your aobjects are
> small "blips" on a large background.
> So playing with the threshold is not the solution.
>
> COnsider using a morhological filter. I find that the tophat filter is
> excellent
> for extracting peaks *without changing their relative intensities*. The
> filter should be set a big larger that you largest objects. Since many
> "canned" implementations
> of this filter use small kernels and also scale intensities, I suggest
> you make your own. The formula is:
>
> TH(image) = [image - opening(image))
>
> where opening is the morphological opening operator.
>
> I say all of this because you write that you are trying to measure the
> extent of something, and this should not require that you do intensity
> calibrations. As long as you do not saturate your detector (which will
> distort the intensity profile) you should be able to determine size
> based on the intensity profile.
>
> I have done this using Image Pro Plus, and I am sure you can do it in
> ImageJ also if they have a grey scale morphology plugin. If you send an
> image I will be happy to try it out and comment,
>
> --aryeh
> --
> Aryeh Weiss | email: [log in to unmask]
> Department of Electronics | URL:
> http://optics.jct.ac.il/~aryeh
> Jerusalem College of Technology | phone: 972-2-6751146
> POB 16031 | FAX: 972-2-6751275
> Jerusalem, Israel | ham radio: 4X1PB/KA1PB
>
--
------------------------------------------------------------------
Joachim Walter, Dipl. Phys.
Institut für Anthropologie und Humangenetik der LMU München
AG Cremer
Richard-Wagner-Straße 10/I
D-80333 München Tel. +49 - 89-2180-6713
Germany Fax +49 - 89-2180-6719
[log in to unmask]
------------------------------------------------------------------
|
|
|