What is DICOM?

DICOM is the standard that lets scanners, archives and viewers from different makers exchange medical images. If you were handed a CD or a download from a hospital and found files ending in .dcm, this is the format they use.

Last reviewed 18 October 2026. Written by the dicompeek editors; sources are listed at the end.

The short answer

DICOM stands for Digital Imaging and Communications in Medicine. It is both a file format and a set of rules for sending images over a network. It is developed by the DICOM Standards Committee, with NEMA (the National Electrical Manufacturers Association) serving as its secretariat, and it is also published by ISO as ISO 12052. Its predecessor was the ACR/NEMA standard “Digital Imaging and Communications”, from 1988. The full text is free to read at dicom.nema.org. “DICOM” is a registered trademark of NEMA.

CT, MR, X-ray, ultrasound, nuclear medicine and many other devices write DICOM. So do the systems that store and show their output (the archives hospitals call PACS), which is why one viewer can usually open images from a scanner of any brand.

What is inside a .dcm file

A DICOM file on disk, called a “Part 10 file” after the part of the standard that defines it, is built like this:

  1. A 128-byte preamble, normally empty, followed by the four letters DICM. A file without them is probably not a Part 10 file at all.
  2. File meta information: a few elements that say what kind of object follows and, importantly, how it is encoded (the transfer syntax; see transfer syntaxes explained).
  3. The data set: a list of data elements. Each has a tag, a value representation (VR, the data type), a length and a value.

A tag is a pair of numbers written like (0010,0010), which is Patient's Name. (0008,0060) is Modality,(0028,0010) is Rows, and (7FE0,0010) is Pixel Data: the picture itself is just one element among many. Tags in odd-numbered groups are private: a manufacturer can put anything there. Elements can also hold sequences of items, which is how DICOM stores nested information such as a list of referenced images. The DICOM tag viewer shows all of them for a file.

More than a picture

A DICOM image keeps the information needed to use the pixels properly, not only to look at them:

Patient, study, series, image

The standard models the data as a hierarchy. A patient has studies (one visit or order, for example “CT head”). A study has series (one acquisition or reconstruction, for example “axial, 5 mm”). A series has instances, the individual images or other objects. Each study, series and instance carries a globally unique identifier (a UID, a string of numbers separated by dots). A CT scan of 300 slices is therefore usually 300 files with the same Study and Series UID and different SOP Instance UIDs and Instance Numbers. A viewer groups files by Series Instance UID and sorts them by Instance Number, which is how the dicompeek viewer builds a stack.

Why the files look odd

The parts of the standard

The standard is published as numbered parts. Four are useful to know when you meet a file:

PartCovers
PS3.3Information object definitions: what a CT image, an MR image or a report must contain
PS3.5Data structures and encoding, including transfer syntaxes
PS3.6The data dictionary: every standard tag, its name and VR
PS3.10Media storage and the file format

Other parts cover the network protocol, web services (DICOMweb), conformance, security profiles and more.

Handling DICOM files safely

Treat every file as containing a patient's name, ID and birth date, because most do. Only open or share files you have the right to. The dicompeek tools read files in your browser and do not upload them. To remove identifying data before sharing, seeanonymize DICOM, and read its limits first.

Sources

Not for clinical diagnosis. See the medical disclaimer.