Confocal Microscopy List

Re: percentile calculation, ImageJ or otherwise?

Posted by Shalin Mehta on
URL: http://confocal-microscopy-list.275.s1.nabble.com/Live-cell-Imaging-with-fluoview-1000-tp2242987p2249248.html

Dear Steffen and list,

 Unless I am missing a point here, I think the simple and best strategy is just to compute the average intensity of each plane to examine how signal decays with depth.

Averaging  p = mxn  pixels will take time to the order of 'p' operations (in algorithm complexity parlance averaging is an O(n) operation).
Obtaining percentile involves sorting of all pixels which is an O(nlogn) operation in general. i.e., sorting p=mxn pixels will take approximately p*log(p) operations. Computing histogram is therefore slower than computing average, but that's the intuitive way if one really needs to compute percentile.

best regards
shalin
On Sat, Jan 31, 2009 at 3:26 AM, Jeff Reece <[hidden email]> wrote:
 
Dear Steffen and List,
 
ImageJ does something very much like Julio's option 2: choose Image->Adjust->Brightness/Contrast, and every click on the "Auto" button clips a certain % of pixels from the high end of the display contrast (and showing the pixel value where that occurs).  I don't know what the % is but wouldn't be hard to figure out.
 
Cheers,
Jeff
--------------------
Jeff M. Reece
Reecent Technologies, LLC
"Honing the edge of quantitative microscopy"
919-672-4681
 
----- Original Message -----
Sent: Friday, January 30, 2009 12:20 PM
Subject: Re: percentile calculation, ImageJ or otherwise?

Hi Steffen, 

there are probably more elegant and/or faster ways, but you could:

1) with ImageJ, plot the Histogram (Analyze > Histogram ) and then get the values as a list. You can paste the list in Excel and use the Excel "Sum" function to determine where the top 5% intensities are. Not elegant, bit will give you an answer.

2) with Photoshop, open your image (in greyscale mode) and use the curves tool. In the Curves window, use the "Options" button for the Auto adjustment to set the minimum at zero and the maximum clipping at 5%. Apply to your image and then look in the curves window where Photoshop set the thresholds.


--
Julio Vazquez, PhD
Director of Scientific Imaging
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.,  mailstop DE-512
Seattle, WA 98109-1024

Tel: Office: 206-667-1215/ Lab: 206-667-4205
FAX: 206-667-6845


--------------------------------------------------
DISCLAIMER:

This message is confidential, intended only for the named recipient(s) and may contain information that is privileged or exempt from disclosure under applicable law.  If you are not the intended recipient(s), you are notified that the dissemination, distribution or copying of this information is strictly prohibited.  If you received this message in error, please notify the sender then delete this message.



 



On Jan 30, 2009, at 2:25 AM, Steffen Dietzel wrote:

Dear all,

I have a stack of images (16bit) for which I need some measurement of the average signal intensity, to describe signal loss with increasing depth. Since the particular signal is sort of repetitive (Collagen matrix), I thought of just going for the 5-percentile (5% of all pixels in an image plane have the grayvalue X or higher).

ImageJ readily calculates the median (50-percentile) but I didn't find an option for other percentiles in the program or on the plug-ins web page. It is probably easy to program it if one knows how and I am sure others have done it before but programming isn't one of my capabilities.

Other suggestions to solve this problems also would be welcome

Best regards

Steffen

-- 
---------------------------------------------------------------------------------------------------
Steffen Dietzel, PD Dr. rer. nat
Ludwig-Maximilians-Universität München
Walter-Brendel-Zentrum für experimentelle Medizin (WBex)
Head of light microscopy

Mail room (for letters etc.):
Marchioninistr. 15, D-81377 München

Building location and address for courier, parcel services etc:
Marchioninistr. 27, D-81377 München (Großhadern)

Phone: +49/89/2180-76509
Fax-to-email:   +49/89/2180-9976509
skype: steffendietzel


Free forum by Nabble Edit this page