Medical Images in DICOM

Abstract

Let me assume, my dear reader, that the time we used to wander in the land of VRs and DICOM objects has been well spent, and that you have a much better idea of how DICOM deals with medical data. This has prepared us for the next important step in our voyage: looking at how DICOM works with medical images. Certainly, images possess some well-known properties (width, height, bits per pixel), which can be found in the DICOM Data Dictionary, and which DICOM encodes with explicit or implicit VRs, as we have already learned. But the most interesting image attribute is the image itself, the sequence of image pixel values, that DICOM stores in the standard (7FE0, 0010) “Pixel Data” attribute, using either OB (for 1-byte pixel samples) or OW (for 2-byte pixel samples) encoding. DICOM supports a wide range of image formats for storing these (7FE0, 0010) pixels. The formats can be loosely broken into two main groups:

1. DICOM-specific: the formats that are used by DICOM only. They are typically the oldest ones, introduced at the dawn of the computer era before better image formats had been developed. They resemble raw BMPs with varying ways of packing the pixel bytes.

2. Independent standard formats accepted by DICOM. These include such well-known formats as JPEG, RLE (run-length encoding), ZIP, and the lessknown (but becoming more and more popular) JPEG2000, and JPEG-LS. All of these standards are also associated with various image-compression techniques, both reversible and irreversible, which makes them particularly useful in medical imaging (reducing image data size is important).