Anatomy of DICOM datasets

Whew! That might have been a lot to take in. Don’t worry if some of this is not quite coming together just yet. We will keep returning to these concepts as digging through DICOM files is an indispensable part of analyzing medical imaging datasets. For now, it is important to remember that these concepts exist, and know where to find them. You might want to return to this section and re-watch the lecture once you start working with DICOM.

One of the important things to capture here is that per the DICOM standard, 3D medical images are stored as files on a file system where each file represents an instance of Image DICOM Information Entity. In the case of 3D medical images, each such instance shares context with other instances that belong to the same series and same study. Thus, each DICOM file stores metadata that describes attributes of study and series that the respective instance is a part of, and this metadata is replicated across other instances that belong to the same study. DICOM files are usually stored with .dcm extension and are usually grouped in directories (but they don’t have to be) to represent data from series, studies and patients. Relationships between individual .dcm files are defined by the metadata stored within them.

New Vocabulary

SOP - Service-Object Pair. DICOM standard defines the concept of an Information Object, which is the representation of a real-world persistent object, such as an MRI image (DICOM Information Objects consist of Information Entities). The standard also defines the concept of Services that could be performed on Information Objects. One such service is the Storage service (we will touch on others later in the course), and a DICOM image stored as a file on a file system is an instance of Storage service performed on an Image Information Object. Such Service-Object Pairs have unique identifiers that help unambiguously define what type of data we are dealing with. A list of SOP Classes can be found in Part 4 of the Standard. This list is a useful reference for all possible data types that could be stored per the DICOM standard.

Data Element - a DICOM metadata “field”, which is uniquely identified by a tuple of integer numbers called group id and element id. The convention is to write the element identifier as group id followed by the element id in parentheses like so: (0008,0020) - this one is the DICOM Element for Study Date. DICOM data elements are usually called “tags”. You can find the list of all possible DICOM tags in Part 6, Chapter 6 of the standard.

VR - Value Representation. This is the data type of a DICOM data element. DICOM standard imposes some restrictions on what form the data can take. There are short strings, long strings, integers, floats, datetime types, and more. You can find the reference for DICOM data types in Part 5, Section 6 of the standard

Data Element Type - identifiers that are used by Information Object Definitions to specify if Data Elements are mandatory, conditional or optional. Data Element Type reference can be found in Part 5, Section 7 of the standard

DICOM Information Object - representation of a real-world object (such as an MRI scan) per DICOM standard.

IOD - Information Object Definition. Information Object Definition specifies what metadata fields have to be in place for a DICOM Information Object to be valid. Scanner manufacturers follow the relevant parts of the DICOM standard when saving the digital data acquired by the scanner. When parsing DICOM data, it is often useful to reference the relevant IODs to see what data elements could be expected in the particular class of information objects, and what they mean. For example, in Part 3 of the standard, you can find MR Image IOD and CT Image IOD which we will use in this course quite a bit. You might have noticed that the table with all DICOM data elements does not really provide any description of what these elements mean. The reason for that is that elements may mean slightly different things depending on what Information Object Definition uses them, therefore, to find the real meaning of the element you need to look them up in the respective IOD.