How should key images be presented?

Opinions will vary but, in general, when a study that has one or more key images is displayed, only the key images should be presented, by default. This may vary for different roles though—for example, it may be appropriate to show General Practitioners only the key images (and the report) by default, while it would be better to provide Surgeons and Radiologists with the full set of images, by default.

If only the key images are displayed by default, often an option to show the full set of images is provided (this option is often controlled by a permission assigned to a role).

Another common feature is to provide a visual indicator—such as a ‘key’ or star icon—on the key images, when viewing the full set of images, so that the user can differentiate the key mages from the rest.

How are key images shared?

Which images are key, and for what purpose, may be managed within the system where they are created, or may be created in a standard and interoperable format.

If the users are on the same system, the user experience between author and consumer is normally quite seamless. If they are on different systems—such as a PACS and an enterprise viewer—the key images may not be communicated from one system to the other. See below.

What are KIN and KOS?

These are two different things, but I find many people refer to them as if they are the same.

KIN, or Key Image Notes, is an IHE integration profile that defines how to create, manage, and display key images.

The data object that KIN specifies to be used is the DICOM Key Object Selection, or KOS, SOP Class. It is worth noting that the DICOM KOS SOP Class is used for other purposes than just identifying key images, including acting as a manifest in the IHE XDS-I integration profile.

The IHE KIN integration profile includes instructions on where to define the purpose of the key images (Document Title), the actual ‘note’ (Key Image Description), and the person that created the key image note (Observation Context), along with a reference to all the relevant images.