Dictionary-based medical image I/O.

A general-purpose image I/O library is presented, whose properties derive from externally defined data dictionaries. The library has primarily been developed for image I/O in a medical image processing environment. It hence addresses all issues found in this field such as a variety of pixel types, architecture independence, automatic type conversion, and an unlimited number of image parameters. The data access routines in the library automatically convert the data types when needed. By redefining the data dictionaries, the I/O library can be tuned for other application areas. Explicit support for image compression is provided. The library is written in C++, and is available for interested parties.