Integrating with Applications

Developing medical imaging applications for clinical use is an undertaking. Ideally, developers should be allowed to focus on the application code itself, and worry less about imaging hardware, protocols, communication and patient integrity. In reality, such obstacles can consume a majority of project resources. By letting a Slicebox instance take care of the communication with image sources as well as storage and management of image data, the path to a running imaging application is shortened considerably.

Most imaging applications are created with a certain type of image in mind. For instance, you may be developing a program that calculates the dynamic uptake of a radioactive isotope in the kidneys (a renography). When your program starts you want it to contact the connected Slicebox instance and ask for all images of this type. But how can you distinguish such images from others? The vital information resides in the attributes of the DICOM files. Let's assume that by inspecting the attributes of the images you have received from the hospital you work at or collaborate with, you notice that these images share the following unique characteristic.

• SeriesDescription = RENOGRAPHY

To make sure that you get the right type of renography image, you may decide to be a little more specific:

• SeriesDescription = RENOGRAPHY
• Modality = NM
• NumberOfFrames = 80

Now, the NuberOfFrames tag is not indexed by Slicebox so you cannot write this rule as a query to Slicebox. Instead, Slicebox introduces the concept of series types which are a set of labels associated with each series of the DICOM hierarchy. Administrator users of Slicebox can define any number of series types by just coming up with a name for each of them. In this case we create the series type RENO through the series type view in the Slicebox GUI. A series type can be associated with one or more rules of the type defined above. We can for instance associate RENO with the rule of three attributes above. Series containing DICOM attributes that exactly match all three attribute equalities above are assigned to the RENO series type. If a second rule is defined for RENO, say

• SeriesDescription = RENOGRAPHY
• Modality = NM
• NumberOfFrames = 40

then series mathing one or both rules will be labelled RENO. Series can have zero, one or many series types assigned to them.

The maker of the renography application can now ask Slicebox to list patients, studies and series associated with the RENO series type. When the application is moved to another hospital or starts receiving images from another camera, this may mean that image attributes look different. It is then simple to define new rules for the RENO series type that make sure these new images are also labelled correctly.