Frequently asked questions about Compressor and Compressor Scheduler

This topic contains some of the common questions asked about the Compressor and Compressor Scheduler tools.

What is the compression ratio (Q Factor)?

Compression ratio is the ratio of the size or rate of the original data to the size or rate of the compressed data. For example, if a grayscale image was originally represented by 8 bits per pixel (bpp) and is compressed to 2 bpp, we say that the compression ratio is 4 to 1.

Compression ratio has no single value; it varies by image type. So we cannot provide definite figures, but some indicative values are:

• Lossless JPEG compression varies from 2:1 to 5:1 depending on the modality and vendor.

• Lossy JPEG compression depends on a user-configurable parameter that degrades image quality as compression increases; typically 6:1 to 10:1. To assist in selecting Q Factors, the application contains default values.

Top

What if both the modality and the station have a defined Q Factor?

The Q Factor used for lossy JPEG compression of a study is applied in the following order: station, modality, default. The logic used is:

• If a Q Factor is configured for a station, the station’s Q Factor is used.

• If no Q Factor exists for the station, but one exists for the modality type, the modality’s Q Factor is used.

• If no Q Factor exists for the station or the modality type, the default Q Factor is used.

The logic is applied to both SCP and SCU servers.

Top

Can the user display lossy and lossless versions of an image for compression?

Lossy compression provides sites with economy of cache and archive storage use. The design was not intended to provide simultaneous viewing of lossy and lossless versions of the same image. Retaining both versions of an image in cache increases the storage consumption, which contradicts the intended purpose of lossy compression.

Compressor does not provide for simultaneous viewing of lossy and lossless versions of an image; however, for demonstration or Q Factor comparison and configuration purposes, you can manually change the image UIDs and get copies of images with varying levels of compression in the same study.

Top

What happens if a site enables compression in both the cache and archive?

Lossy compression can be enabled in the following modes:

• Cache only—The study is lossless archived (if configured; for example, Telerad may not be), then lossy compressed. If it is diagnostically read, it is compressed when it reaches the age defined in Read Study Age. If it is not diagnostically read, it is compressed when it reaches the age defined in Study Age.

• Cache and archive—The study in cache is lossy compressed (following the compression parameters) and this lossy compressed version is archived. The IMPAX system no longer contains a lossless compressed version of the image.

  This means that archiving is now dependant on lossy archiving and, therefore, diagnosis. Specifically, before studies can be archived, they must be either diagnostically read and have reached the age defined in Read Study Age or, if they are not diagnostically read, they must have reached the age defined in Study Age. By default, this means the study stays in cache 14 to 21 days. This may not be desirable, so a trade-off between when to compress and when to archive is introduced by enabling lossy archiving.

Top

Is the quality of lossy images acceptable for diagnosis?

The medical community is still debating the issue. This debate has been going on for the past 20 years and is likely to continue for a long time. To avoid getting caught in medico-legal issues, we do not recommend the use of lossy compressed images for diagnosis. For this reason, lossy compression can be applied only after a diagnostic read and is conditional on study dictated or reported state. We recommend a reported state.

Top

With lossy compression, what is the image quality degradation level?

To objectively quantify image quality degradation is difficult. In other applications, line pairs per millimeter, modulation transfer function, bit depth, and many other parameters quantify specific aspects of image quality. None of these on its own is representative of overall image quality or what is left of it after compression. To complicate matters, all of these parameters have a different importance in different imaging applications. The best we can do is recommend the use of conservative default Q Factor settings for the different modality types and the periodic update of these defaults based on customer use trends.

Top

If routing rules indicate that images must be lossy compressed when IMPAX exports them to another system, are lossy compressed images compressed even more?

Lossy images are not further compressed.

Top

When another device does a DICOM retrieve from IMPAX, does IMPAX send the lossless or lossy version of the images? Can the other device specify whether it wants the lossy or the lossless version?

If another DICOM device requests a study, we send it in whatever form it is stored in the cache (depending on settings for that particular destination) and LEI (Little Endian Implicit). If it is not in cache, it is retrieved from the archive. The remote device cannot request a particular type of data, but it can control what it accepts from IMPAX and IMPAX can configure what types of data are sent to a particular destination. For instance, we can convert all images to LEI for a particular destination, or compress them in lossy JPEG or Mitra Wavelet form.

Top

When an IMPAX cluster contains multiple archives, can they be configured independently as lossy or lossless archive or do all archives have the same setting?

All archives in one cluster have the same setting.

Top

How can a user retrieve a lossless copy when lossy archiving is disabled?

Through the Client, a user can retrieve the lossless copy, if it is available in the archive, even if the lossy copy is in cache. The lossless copy can be retrieved through the Study Manager.

• If the study is lossy JPEG compressed in the cache and lossless archived:

  1. A retrieve job is created.

  2. Lossy copies of the study are removed from cache.

  3. After the retrieve job is created, the sop_instance_uid is set to its original value (as stored in the pre_compressed_uid) and the pre_compressed_uid is set to an empty string.

  4. The lossy_compressed_status for the study is set to F.

• If the lossy copy of a study exists in the cache and on an archive, the retrieve job is not created. No lossless copy is available.

• If the lossy compressed study is archived and no copy exists in cache, the retrieve job is created for the lossy compressed copy. No lossless copy is available.

Top

Can a user archive a lossless copy of a study even if lossy archiving is enabled?

Yes, the user can archive a lossless copy through CLUI by typing study store_lossless study_ref. However, the store job is created for a particular study only if the study has not been lossy JPEG compressed and lossy archiving is enabled.

Top

How does protecting a study affect the lossy/lossless state of the study?

If a study is protected, it is not lossy JPEG compressed. Protection also means that Autopilot does not remove the study from cache when the cache reaches its threshold for deletion. Therefore, the study must either be manually unprotected (so Autopilot can proceed normally) or be manually deleted from cache.

Top

How are 3D images lossy compressed?

3D images are typically transmitted as modality OT (Other). If no Q Factor definition for OT or Default exists, a Q Factor of zero is used, which results in unusable 3D images. Therefore, if for some reason the Default setting must be removed or is not conservative enough for 3D images, we recommend adding an entry for the OT modality in Network Management.

A typical value for the OT is Q Factor=90.

Top

Are studies lossy compressed based on cache volume (like Autopilot) or study_date?

The studies are lossy compressed based on study_date.

Top