Jump to content

  • Log in with Facebook Log in with Twitter Log in with Windows Live Log In with Google      Sign In   
  • Create Account

Submit your paper to J Biol Methods today!
Photo
- - - - -

Digital imager questions on correction

Digital Western blot Contrast Background Ethics

  • Please log in to reply
4 replies to this topic

#1 JDSBlueDevl

JDSBlueDevl

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 33 posts
0
Neutral

Posted 12 August 2015 - 06:47 PM

Our lab has started working with the Amersham AI600 imager, which has given some very sensitive blots, thus reducing the amount of antibody we use. However, I notice that the autocorrect function gives a lot of background and band saturation, and this is usually seen even at lower exposure times (semi-auto frequently tells me I have to expose for greater than 10 minutes, but I know that is almost never true). This, I have been finding myself adjusting the "contrast" (if that is what it is rather than black intensity), which brings out the contrasts between bands better and reduces the saturation of the highly intense bands (the bands would remain the same intensity according to ImageJ, whereas everything else is reduced). I was justifying this as merely doing "shorter" exposures, albeit in reverse. However, I keep wondering whether this might be considered image manipulation. Comparing the ImageJ quantification between the finished picture and the original TIFF (which, BTW, is not the same as the original autocorrected image) is giving me nightmares. Does anyone have a good read on whether what I'm doing is acceptable, or should I just do my quantification on the original, non-autocorrected TIFF?

 

EDIT: I am including an example of what I'm talking about.  The TIFF is the original provided by the imager, with image exposure of 6 minutes.TIFF as JPEG.jpg  

 

This is the original, "autocorrected" version that the imager spits out. patm.s1981br.sigdim 2015.08.06_14.08.43_Ch+Marker.jpg

 

This is the one on which I reduced the background over the entire image, as carried out on the imager itself. pATM (S1981) different DIM brands (3.0 uM) 2015.08.06_14.08.43_Ch+Marker.jpg

 

 

And here are the numbers I got:

"Autocorrected"

 Mock           7883.782

 IR                 23703.075

 BR (3.0 uM) 15997.095

 Sig (3.0 uM) 7774.782

 

From TIFF

 Mock            2514.154

 IR                 14845.631

 BR (3.0 uM)  5784.246

 Sig (3.0 uM)  2644.447

 

"Background-subtracted"

Mock             2854.79

IR                   22372.468

BR (3.0 uM)   11837.125

Sig (3.0 uM)   2924.962


Edited by JDSBlueDevl, 12 August 2015 - 08:06 PM.


#2 mdfenko

mdfenko

    an elder

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 3,057 posts
168
Excellent

Posted 13 August 2015 - 05:07 AM

the only place you may have a problem is where the band is saturated. in that case there is no way to know how much beyond the saturation point the band has gone.

 

the "autocorrected" and "background-subtracted" images both exhibit apparently saturated bands.

 

however, your "tiff" does not appear to saturate. you may be able to use that for further comparison.


talent does what it can
genius does what it must
i do what i get paid to do

#3 JDSBlueDevl

JDSBlueDevl

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 33 posts
0
Neutral

Posted 13 August 2015 - 05:49 AM

Well, the saturated band is a positive control, and I'm not entirely sure that the TIFF exposure would be considered dark enough by a reviewer.  There is still that question of the difference in fold value of BR/Mock between the "TIFF" and the "background-subtracted" pictures.  Which do you think is more reflective of a true comparison?  Would the "autocorrected" BR band still be saturating in your opinion?



#4 JDSBlueDevl

JDSBlueDevl

    Enthusiast

  • Active Members
  • PipPipPipPipPip
  • 33 posts
0
Neutral

Posted 13 August 2015 - 09:55 AM

OK, I think I see the problem.  The issue is with partially with ImageJ.  The quantifications are being done on an 8-bit scale, which is going to saturate at around 23000.  Even the TIFF image, which is 16-bit, is being quantified with 8-bit settings.  I'm trying to see how to set ImageJ to do quantifications on a 16-bit scale.  That might also tell me that when I do manual contrast, I might be accentuating differences to magnitudes that may not be true.  I was exploring this mainly because I was not seeing a difference in my second experiment (which I did not post here) when I looked at the "autocontrast" image, but I did see 2-fold when I did manual contrast.  After talking to my collaborator, we saw that the difference may have been in the amount of time the cells were plated prior to treatment (2 days for the posted images, 1 day for the experiment that didn't work).  If that's the case, then that may make a huge difference, especially when considering that this plating time was not even mentioned in the paper.



#5 mdfenko

mdfenko

    an elder

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 3,057 posts
168
Excellent

Posted 14 August 2015 - 03:18 AM

Well, the saturated band is a positive control, and I'm not entirely sure that the TIFF exposure would be considered dark enough by a reviewer.  There is still that question of the difference in fold value of BR/Mock between the "TIFF" and the "background-subtracted" pictures.  Which do you think is more reflective of a true comparison?  Would the "autocorrected" BR band still be saturating in your opinion?

the tiff could be contrast adjusted to a point less than saturated so that the ratio of the bands remain the same.

 

the autocorrected "br" band appears saturated, as does the background corrected band.


talent does what it can
genius does what it must
i do what i get paid to do





Home - About - Terms of Service - Privacy - Contact Us

©1999-2013 Protocol Online, All rights reserved.