Quantitative Cell Biology and ethics:
One of the historical problems with data by microscopic imaging is the issue of investigator bias and ethics when reporting
imaging data. In any published microscopy image, some common questions have to be asked of the qualitative data:
1. Is the image presented reproducible and representative? Or is this just an uncommon cell found on the dish?
2. Does this image represent the average result from a cell biological experiment?
3. How did the author chose to present this particular image?
4. How was this image manipulated prior to press submission?
For any given cell biology experiment, there is the conscious or unconscious temptation to find the cell image that fits the
hypothesis, rather than objectively obtain the data an access the hypothesis. In addition, in the digital imaging era, there is a
considerable amount of power in software to alter imaging, and some alteration can affect the actual data.
The solution to this is to quantify the imaging, with at least three replicated experiments, and visualization of hundreds of
cells. Captured cells need to be blinded to the investigator -this can be done by using a co-transfected fluorescent protein in
another color channel, and only observing this color at time of multi-channel image acquisition. Additionally, experiments can
be blinded to the investigator by another investigator. After quantification, only the image representing the average result
is presented, along with graphed data, with means, standard errors, and suitable statistics with probability values. For more
on this, see publications from the Day lab in Biotechniques.
For rules on how to manipulate images, we refer to the guidelines published by the Journal of Cell Biology. Additionally, see
the website of imaging ethics at the University of Arizona.
For digital data integrity, all imaging data in the lab is captured to mirroring (RAID) arrays, the backed up to a third hard
drive, and each investigator must burn all data to a read-only CD-ROM or DVD-ROM. Notebooks should be able to trace all
raw images to locations on discs.
All data quantification, and statistics should be saved to ROM media as well.
For some common data mistakes in imaging, please see the following page.