DICOM PS3.17 2020a - Explanatory Information |
Page 893 |
NNNN Mapping of Visible Light Photography Related Attributes to EXIF and TIFF/EP Tags (Informative)
NNNN.1 Mapping
NNNN-1 describes a mapping of EXIF Tags as defined in [EXIF 2.31] and TIFF/EP Tags as defined in [ISO 12234-2] to Visible Light PhotographyrelatedAttributes.ThelistofEXIFandassociatedTIFFtagswasderivedfromthetableathttp://www.exiv2.org/tags.html, updated to remove redundancies and resolve differences between EXIF 2.3 and 2.31. The intent is that this mapping support the extraction of embedded EXIF information into DICOM Attributes for accessibility in DICOM-based systems. It is not intended to define a deterministic mapping supporting round-trip full-fidelity conversion from EXIF to DICOM and back to EXIF again. Not all EXIF Tags are mapped.
When mapping EXIF values to DICOM, some tags map to multiple DICOM Attributes and vice versa, as noted. In general, Ascii types are mapped to UT VR (except that dates and times and timezone offsets are mapped to the appropriate DA, TM, DT or SH VR), Short and Long are mapped to IS VR unless they are used for a list of defined terms, in which case US VR is used, Rational and SRational are mapped to DS (VR) (i.e., each numerator and denominator pair is mapped to a single decimal representation), and Undefined is mapped to OB or UT VR as appropriate. In some special cases, the structure within a single EXIF value is expected to be decoded, e.g., the tabular representation of the OECF and Spatial Frequency Response. An appropriate choice of DICOM Specific Character Set and any corresponding character set mapping encoding conversion is expected; special handling of the decoding of character sets may be necessary for some EXIF tags, such as 37510 UserComment.
Some unmapped EXIF tags may theoretically influence the way in which an image might be displayed differently than it is encoded. E.g.,Orientation(274)maycontainenumeratedvaluesdescribingthe"imageorientationviewed"intermsof"visualrowsandcolumns" related to encoded rows and columns, is not mapped. This may be a factor if the camera is physically rotated or if the user has edited the preferred orientation. In theory, such information could be mapped to a Presentation State, which is not defined by this Annex. There is no expectation that the mapping will involve re-encoding the pixel data in a different orientation, though this is not explicitly forbidden.
In NNNN-1, the word "tag" in the columns labelled "EXIF or TIFF Tag" and "EXIF or TIFF Tag description" refer to the tag in the EXIF or TIFF standards and not the DICOM Tag.
