Happy Scientific Image Processing

Februar 20, 2015 Academic 0 Kommentare

25 years of Photoshop: Scientists, watch your data images!

While celebrating 25 years of Photoshop, we send out a reminder to all the scientists who want to make the most out of their scientific images. While editing your image, don’t cross the fine line between beautification and fabrication. Photoshopping your data should happen in order to simplify your message and enhance the scientific content. But unmindful image manipulation can lead to loss of information and misinterpretation; at worst you deliberately fabricate data that was never been obtained experimentally.

Here are a few examples for the DOs and DON’Ts in scientific image manipulation:

Beautification

This is an example for a beautification attempt. The scientist apparently merged several gel runs into one picture, leading to the impression that the whole result was achieved in one run. This should at least be mentioned in the description of the image.

Beautification 1

Simple changes to your data image (e.g. microscopic image) are ok, if you apply these changes to the whole image (sample and control) and also mention the changes (e.g. increased brightness) in the publication.

Beautification 2
Beautification 3
You should not apply these changes only to the region of interest, e.g. positive signals in your image, because this would be fabrication:
Fraud 2

Fabrication

Unfortunately, image processing software like Photoshop makes it very easy to fabricate your image, e.g. copy and paste gel bands. This is definitely not allowed and will be handled as fraud:

Fraud 1

 

The journals which receive your publication have their methods in order to detect image manipulation and fraud attempts. For example via false coloring methods, revealing copy and paste cases:

 Fraud 3

Guidelines

Here are some rules for scientific image post-processing:

1     Scientific images are data that can be compromised

2     Always work on a copy of the original file

3     Simple adjustments to the whole image are acceptable

4     Cropping is acceptable

5     Images that will be compared have to be acquired/post-processed under the same conditions

6     If software filters are used, describe every parameter

7     Cloning or copying of objects into the same image is questionable

8     Avoid use of lossy compression (use PNG format instead of JPEG)

9     Use scale bars, not numerical magnification numbers
And last but not least, here’s to the next 25 years of great image processing!

http://video.tv.adobe.com/v/2102t1/?quality=6&autoplay=true