Image processing is necessary for us to present picture-format data in publication quality figures, but inappropriate manipulation of images can lead to paper rejection and doubts on research credibility. Even though intentional fraud is relatively rare, inappropriate manipulation of images due to ignorance is often seen. The Journal of Cellular Biology is among the first journals to rigorously examine images and screen for image manipulation. The guidelines1 developed by The Rockefeller University Press, the publisher of JCB,have been widely used since its publication.Other journals and organizations have also put out their guidelines2-4. Below we summarize the major points of what are allowed and what should be avoided when you prepare your images for publication.
1. Brightness and contrast change.These changes are considered linear (e.g., all pixels become brighter) and are generally allowed when you apply themto the entire image. Don’t only manipulate a local area in a picture to brighten or dim some features.
Keep in mind that, even though these changes are allowed, the changes should be moderate and should not cause changes in the interpretation of the results. For example, if you increase the brightness too much, you might make too many pixels to have the maximum brightness and therefore lose the subtle differences among these pixels. Avoid using an image program’s auto contrast, auto levels, and auto color tools, because these may overprocess your image. Adjusting the gamma value is considered a non-linear change and should be avoided. When gamma value adjustments and pseudo-coloring are necessary, they should be disclosed in the paper. Another point to keep in mind is that if you have a group of images to be shown together for comparison (e.g., a control vs several different treatments), all the pictures should be taken using the same setting and manipulated the same way. Yes, this means you should have a record of the setting used for photographing and the way you manipulated the images.
2. Cropping. We often don’t need to show a whole picture as we get from the camera, so it is allowed to crop an image to remove unnecessary information or empty space around the edges. However, don’t cut some image pieces and then stitch them together as if it were one picture. If you have to stitch some pieces together, the pieces should be outlined to clearly show that they are from different sources. For example, if you have a scan of a western blot with the lanes in the order of A, B, C, D, but to fit your story better, you want to show them in the order of A, D, B, C. You can cut lane D out and insert between A and B, but you have to draw lines around D to disclose the fact that it has been moved.
A few things to keep in mind when you crop images: First, cropping should not introduce bias, i.e., the area shown should still be representative of the sample. Second, sometimes empty space is necessary. This is particularly important in the case of western blots, where it is customary to show only the protein band of interest. Cropping too close to the band may make readers suspect that you are trying to cover up the fact that there isanother band nearby.Some journals require thatcertain areaof the blot above and below the band must be shown in the figure. Third, the group of images to be shown together for comparison should be cropped to the same size.
3. Re-sizing.An image published in a journal article is almost never the same size as the original picture. Re-sizing is a necessary step to set your figure to a journal-desired size. Generally it is acceptable to make an image smaller (i.e., decreasing the number of pixels). The computer software can combine multiple pixels into one pixel by averaging their values, and the human eye will not detect a difference. Don’t try to increase the number of pixels of an image, because the computer software will have to create extra pixels that are not already there. This causes misinterpretation of your data and doesn’t make a low-resolution image looking better anyway.
Before re-sizing, a scale bar should be added to the image so that you have a correctly sized scale bar for the final figure. Again, the group of images to be shown together for comparison should be re-sized the same way.
The above three types of changes are the most basic, and pretty much the only manipulations you are allowed to do to make a picture look better. More complicated processing would all need to be described in the paper. The following tips can help you minimize the amount of manipulations you have to do to the images.
- 1. Plan ahead. If you have planned your western blot experiments well, you are less likely to find it necessary to rearrange the lanes in your figure.
- 2. Take the best pictures. Get familiar with the instrument and software you use for image acquisition, so that the pictures you take are close to the way they should appear in your paper. Take pictures at the highest resolution possible and save in the correct format (TIFF, not JPEG). You should also save files in the native file format of the image acquisition software, because such files may contain metadata of instrument setting such as the magnification information.
- 3. Always save the original image files. Save another copy when you start processing the image so that you always have the original file. One, if you overprocess or make some other mistakes, you can always start over. Two, journal editors may ask to see your original images to make sure that images have been properly handled. Three, a digital image contains information beyond what we can see with our eyes. An 8 bit grayscale image can have 256 shades of gray, while humans are only able to distinguish about 30 shades of gray. The processing described above aims for the best visual effects for human readers, so in a processed image the image data have been altered and some information may have been lost. If you need to do quantitative measurements, you should use the original images, not the processed ones.
1. Rossner M, Yamada K. What’s in a picture: the temptation of image manipulation. J Cell Biol. 2004;166:11-15.Doi: 10.1083/jcb.200406019
2. Nature.Image integrity and standards. http://www.nature.com/authors/editorial_policies/image.html(Accessed October 3, 2019).
3. Office of Research Integrity.G.uidelines for best practices in image processing. https://ori.hhs.gov/education/products/RIandImages/guidelines/list.html (Accessed October 3, 2019)
4. Cromey DW. Avoiding twisted pixels: ethical guidelines for the appropriate use and manipulation of scientific digital images. Science and Engineering Ethics. 2010;16:639-667. https://doi.org/10.1007/s11948-010-9201-y