Image preparation

Last revised by Frank Gaillard on 14 Apr 2023

Images are clearly a very important part of and we aim for high quality and uniformity across the site. As always patient anonymity is essential, and any images with patient details included will be removed, and cases will usually be deleted.

Radiology is all about image quality and as such it is crucial for the site as a whole and for your online contributions to be of high quality. We are not all blessed with state-of-the-art hardware, but that is no excuse for poorly optimized images.

Before uploading an image please ensure that: 

  1. the images have appropriate window/level/contrast/brightness

  2. are of sufficient resolution (see below)

  3. are appropriate for teaching (slice thickness, views, etc.) 

In the new uploader (available as an opt-in beta feature in the user settings) you can crop/pan/zoom images prior to uploading. 

  • JPG: small size but degraded by compression; use quality of 80 or above to minimize image degradation 

  • PNG: non-lossy, reasonable size, the preferred format for diagrams, as edges and text will be crisp

  • DICOM: lossless, an original image format that is the standard for storing and transmitting medical images; Uploading DICOM images to Radiopaedia directly (with full client-side anonymisation) is available as an opt-in beta feature in the user settings accepts (essentially) any size and then resizes it to serve up to viewers. Thus uploading a larger size, particularly native modality resolution is recommended.

Never (except for plain films and mammography) upload images that are smaller than the size acquired: 

  • CT: 512 x 512 pixels (px)

  • MRI: sequence-specific ranging from 128 to 1024 px

  • x-ray: variable but generally larger than 1600 x 1600 px

  • ultrasound: 512 x 512 px

  • mammography: 2048 x 2048 px

  • angiography (DSA): 1600 x 1600 px

If in doubt upload a larger image. We keep all the original images and optimize the images for display on browsers. 

Most images on the site are displayed in a square 1:1 container. Any space not taken up by the image will have a black background. Therefore, for any diagrams or images with non-black background please ensure the image is square so that it displays well. 

In the new uploader (available as an option in your user settings) you can crop, pan and zoom images and stacks. 

The ability to scroll through a stack of images is the cornerstone of cross-sectional imaging and is now expected for all cases containing CT or MRI. 

Uploading stacks is easy but depends on the image format. 

  • JPG and PNG: For cross-sectional images (CT and MRI) or temporal sequences (DSA run), sequential images from a series can be uploaded as a .zip file containing sequentially numbered images. See stacks for more information.

  • DICOM: sequential images are automatically separated into stacks when uploaded using this format; the uploader is available as an opt-in beta feature in the user settings

PLEASE avoid uploading every single thin image you have. A 650 x 0.6 mm stack is useless and takes forever to load. It is very unusual for a single series to require more than 100 images. 

Make sure you create the best possible learning experience (which rarely requires thins that you may need for diagnosis). 

It is, however, good to include more than just the area of abnormality as search-and-find is an important skill to develop. 

In the new uploader (available as an option in your user settings) you can trim unnecessary images from the start and end of stacks. 

No. In general please have only one image/slice/projection per image. This is especially the case for cross-sectional imaging.

The one exception (sometimes) is long/thin x-rays (e.g. limbs) which can be better presented as side by side views. 

No. With the exception of diagrams or illustrations you have created, images should not include your name or any other attribution. If you need to attribute the images to someone (e.g. a colleague) then please do so in the findings or discussion section.

No. We want images to be a clean and easy to use as possible. That is the point of Radiopaedia. Adding watermarks or logos etc... is not acceptable.

A number of tools are useful to make sure your images are the best they can be (for and for publication).

  • Adobe Photoshop

    • not-free... not even close

    • the 'gold standard' of photo editing software

    • (Windows and Mac)

  • Krita

    • opensource Photoshop-like

    • (Windows, Linux and Mac) 

  • Gimp

    • opensource alternative to photoshop

    • (Windows, Linux and Mac) 

  • Irfan viewer

    • free

    • powerful image converted (great for batch processing images)

    • (Windows only)

  • Horos

    • Free, open source DICOM viewer

    • Can be used to create and organize your own teaching library

    • Powerful DICOM editing tools and many export options 

    • (Mac)

If you do not want to install software on the computer you are using you can use online browser-based programs of which there are dozens.


    • free

    • lets you crop and convert to greyscale


Taking photos of printed film is tricky, and generally discouraged. Some historical films are worth preserving and with a little practice, reasonable results can be obtained.

General tips

  • camera settings

    • black/white (you can convert to greyscale later too)

    • macro

    • high-quality jpg (90 or 100)

      • you can convert .raw or .tiff image types to .png, but this is of limited utility as jpg image degradation is not the quality limiting step

  • tripod ideal

Common problems

  • glare from next to the film is horrible so you need to mask it out

    • for CT/MR which are the standard size a hole cut in card works fine

    • for odd shapes need to be more inventive

  • reflections may not be apparent until you look at the image, so a pitch-black room except for the lightbox is best, and even then you may see a reflection of yourself

  • autofocus often does not work (fluffy images) so focusing on a sheet of paper with lines drawn on it placed in front of the film works well; alternatively set your focus to the central dot, and aim initially at the patient demographics (remember to exclude these or edit them out later)

  • it can be really hard to replicate a nice flat image when there is dense bone (when the film is completely transparent you get a lot of glare and can't see the rest)

ADVERTISEMENT: Supporters see fewer/no ads

Cases and figures

  • Example 1
    Drag here to reorder.
  • Example 2
    Drag here to reorder.
  • Example 3
    Drag here to reorder.
  • Example 4: poor quality photo of printed film
    Drag here to reorder.
  • Updating… Please wait.

     Unable to process the form. Check for errors and try again.

     Thank you for updating your details.