11.1 The DICOM file format

For a complete description of the DICOM format, see the specification page at

http://medical.nema.org/standard.html

which provides a brief description. Another excellent resource is the blog by Roni Zaharia:

http://dicomiseasy.blogspot.ca/

Each DICOM file contains a number of concatenated attributes (a.k.a. elements), one of which defines the embedded binary image pixel data. The other attributes act as meta-data, which can contain identity information of the subject, equipment settings when the image was acquired, spatial and temporal properties of the acquisition, voxel spacings, etc.... The image data typically represents one or more 2D images, concatenated, representing slices (or ‘frames’) of a 3D volume whose locations are described by 13.5 the meta-data. This image data can be a set of raw pixel values, or can be encoded using almost any image-encoding scheme (e.g. JPEG, TIFF, PNG). For medical applications, the image data is typically either raw or compressed using a lossless encoding technique. Complete DICOM acquisitions are typically separated into multiple files, each defining one or few frames. The frames can then be assembled into 3D image ‘stacks’ based on the meta-information, and converted into a form appropriate for display.

Each DICOM attribute is composed of:

• •

  a standardized unique integer tag in the format (XXXX,XXXX) that defines the group and element of the attribute

• •

  a value representation (VR) that describes the data type and format of the attribute’s value (see Table 11.1)

• •

  a value length that defines the length in bytes of the attribute’s value to follow

• •

  a value field that contains the attribute’s value

This layout is depicted in Figure 11.2. A list of important attributes are provided in Table 11.2.