>> I have recently run into some rather obnoxious comments from a reviewer
>> of one of my manuscripts, who implied that we had been less than honest
>> in our presentation of our images.  So I put the question to you all.
>> How much enhancement is too much?  What is acceptable?
>>
>> In this particular case, we had a double labelled FISH sample, using
>> propidium iodide and FITC.  The image was enhanced by setting all
>> green pixels below ca. 40 to 0 (to remove backgound FITC) and then enhancing
>> contrast by mapping the remaining pixel values over the entire 256 range for
>> each channel.  No other manuipulation was carried out.
>>
>> If we are out of line with this approach, I would appreciate some
>> guidance.  If the approach is appropriate, I would like to see your
>> comments. I hope to get back to the editor by weeks end.
 
I guess it all depends on what you're trying to prove.
 
Generally what I tell people when I'm training them is that an image like
yours is useful for making statements like the following:
 
"The probe seems to be localized to the cytoplasm and excluded from the
nucleus..."
 
"The feature of interest is 5 microns above/below/to the side of the
membrane..."
 
etc...  In other words, images which have been processed such that
intensities have been altered are useful for studies involving topology,
morphology, spatial realationships, etc.  However, if you are trying to say
something like:
 
"Since feature A is 15% brighter than feature X, we can see that our
protein of interest is being expressed at a 15% higher level in feature A."
 
This would be a mistake.  Quantitation from an image where you've messed
with the data is going to be a problem.  I tell people to make a copy of
any image they plan on doing enhancement on so that they have the raw data
to fall back on.  I also agree strongly with the comment that you should
outline everything you've done to your data in the paper itself.
 
Now... I won't even BEGIN to tackle the issue of whether you can
legitimately quantitate from even an UNALTERED confocal image... :-)  There
are tons of variables involved in any type of microscopy even when no
"image processing" is done to the raw image.  We're all making assumptions,
some of us just pretend we're not.
 
 
--
Charles Thomas                                  USPA Licence A-19548
Integrated Microscopy Resource
Univ. of Wisconsin-Madison
1675 Observatory Dr.
Madison, WI  53706
608-263-6288
[log in to unmask]
 
"I'd rather be skydiving..."