Compression is used to reduce image file size, increasing the number of images that can be stored in the image cache and archive. The primary goal of compression is to extend the capacity of the storage, increasing the amount of time that images are immediately accessible. By keeping studies in the cache for as long as possible:
Images are available for relevant comparison to new studies during primary diagnosis without having to access archive media, which takes more time.
The archive is accessed less frequently, which reduces wear.
The two basic types of compression are lossless compression and lossy compression.
Lossless compression storage options
IMPAX receives DICOM images from modalities such as CR, CT, MR, and others in a variety of formats. By default, the SCP performs lossless compression on almost all images as they are received and places them into the image cache, where they are available for primary diagnosis on an IMPAX Client. If an incoming file has already been compressed by the transmitting station into a format supported in IMPAX—for example, into MPEG—IMPAX retains that format.
The image and web caches have a finite size, whereas typically an archive is regularly expanded (for example, by adding removable media). Autopilot normally creates a STORE job for a study a short time after its arrival (hours or days) and the study is archived to a permanent location. After a study is archived, it remains in cache until it is deleted by Autopilot when disk space is needed for newer incoming images or their relevant prior studies.
Multiple Curator processes can run concurrently in IMPAX, processing images soon after they arrive in the IMPAX cluster and creating wavelet images (lossless compressed by default) to store in the web cache or caches. Wavelet images can be viewed at lower image quality with only partial file download, which is why they are created and used for Web distribution. Wavelet images can also be used for comparison during primary diagnosis.
Lossy compression storage options
Lossless compression can decrease file size and therefore increase the number of images immediately available, but it has limits. To extend the caches further, either additional storage hardware is required or the file sizes must be further reduced. To extend the capacity of the caches without additional storage hardware, lossy JPEG or lossy wavelet compression can be enabled, although this reduces the file size by permanently eliminating certain information, especially redundant information. When a lossy file is uncompressed, only a part of the original information still exists.
 | Note: Reduced file size also mean that it takes less time for the IMPAX Client to access images. Access time can be a problem with large images regardless of network bandwidth; for example, digitized film, CR, and mammography images can be 15 MB or more. |
Differences between lossy JPEG and wavelet compression include:
Wavelet images are configured in the Web Compression Manager, while lossy JPEG is configured in Network Management - Image and in Source Manager - Compressor (all in the Administration Tools).
Wavelet images are available soon after the study has arrived in IMPAX, whereas lossy JPEG compression is primarily intended to extend the image cache some time after primary diagnosis.
Wavelet compression has the ability to compress all data types, although some are exclusively lossy, whereas lossy JPEG has some data type limitations.
Using a mixture of lossy and lossless compression to extend storage space
A common system configuration is to partition the image caches using lossless JPEG compression (for example, 6 months worth) for primary diagnosis, initial film printing, and other DICOM needs, while enabling lossy wavelet compression to support Web distribution and comparison.
This significantly extends storage utilization (by 2 to 5 years or more) to support all the clinical workflows. This configuration does not preclude the use of lossy JPEG for extending the image cache, but there is no specific need for storing two lossy compressed version in cache, other than to support lossy archiving, especially when the lossless JPEG version is helpful for DICOM transfers. The rationale for lossy compression can be extended to the archive, termed lossy archiving, where hardware costs and retrieval time can be lowered by decreasing the file size.
 | CAUTION! Archiving of lossy JPEG compressed studies can be configured in IMPAX; however, if lossy archiving is enabled, when a study is lossy compressed, the lossless version is permanently removed from IMPAX. Therefore, Agfa does not recommend lossy archiving, particularly for mammography studies. |