The Digital Database for Screening Mammography (DDSM) image files are compressed using a lossless format, which they dubbed “lossless JPEG”. Unfortunately for mammography researchers, since this compression scheme was developed, a different lossless JPEG standard has emerged that is now widely supported (e.g., by Photoshop and Matlab). If one attempts to read the DDSM images using the built-in support for the new lossless JPEG standard, the resulting mammograms will be incorrect.
To read the DDSM mammograms, one must use the DDSM‘s ’jpeg’ software, which was originally written for a UNIX environment. It is possible to build the ‘jpeg’ program under Linux (by following the instructions given in the software package below), and a Windows executable is also provided (courtesy of Wang Weiwei, a masters degree student at the Oncology Informatics Laboratory, Institute for Biomedical Engineering and Instrumentation, Hangzhou Dianzi University, Hangzhou, China); the Windows executable must be run under the Cygwin Linux-like environment for Windows (the software was compiled under Cygwin version 1.5.25–15).
Whether on Windows or UNIX, the ‘jpeg’ program does not produce an image that is immediately useable, and the ‘ddsmraw2pnm’ program in the DDSM Software package below will allow you to convert to the PNM file format, which is widely supported and simple to read; one can also optionally use the other included software to convert to PNG format.
As an alternative to the DDSM, researchers may also be interested in the mini-MIAS database. This database has the advantage of storing its images in a format that is much easier to read than the DDSM images, but has fewer cases and the images were digitised at a relatively low resolution.
If you have any questions about the software or services that we developed, please contact me at .
The MIAS-Grid project developed several web-services for medical image analysis tasks, including a service which allowed users to search for and obtain DDSM data in a convenient file format. Unfortunately—after the project was completed—it has not been possible to maintain all of the servers that are required to run this service and this service is no longer live.
You may also download copies of the papers that describe the service: