Compressor is the program responsible for compressing study images into lossy JPEG form. Each Compressor in the cluster monitors its own queue. When a job is ready, Compressor changes the job status from New to Progress and starts processing it by creating the list of all images in that study.
Each Compressor in the cluster requires cache space to store both lossy and lossless copies of a study while it is being processed so that the lossless image can be decompressed into a LEI format and then lossy compressed. The requested amount of space is configurable; the default value is 500 MB. Autopilot is responsible for managing image cache; therefore, the cache may run out of space for lossy compression if Autopilot configuration and performance is not managed properly.
After all of the images in the study are successfully processed, the lossy_compressed_status is updated in the dosr_study table and the lossless copy in the cache is marked for deletion. Deletion of the lossless copies is done for two reasons:
To maximize the cache space and allow for additional studies to be stored in the cache.
Because the DICOM header cannot be used to distinguish between the lossy and lossless copy and there is no valid transfer syntax in the database.
Autopilot then removes the marked studies from the cache.
Optimal compression performance
Optimal compression performance, from a compression workflow perspective, depends on device management (which SCP stations are receiving images), cache configuration (how the majority of the cache is distributed on the cluster, whether they are on Network Gateways or Archive Servers), and lossy compression mode (whether lossy archiving is enabled).
These factors, along with system hardware capacity (CPU, RAM, and so on), must be considered when configuring where Compressors are enabled in the cluster and where specific studies should be compressed.
How does Compressor handle errors?
If an error occurs while processing an image, an exception is sent. The job status is not changed to Retry because the images most likely cannot be processed. To maintain consistency in the IMPAX cluster, if one image in the study cannot be processed, the whole study is not lossy JPEG compressed. If no cache space is available, an MCF exception is sent every hour. The required amount of available cache space is configurable; the default value is 500 MB.
What images cannot be lossy JPEG compressed?
Studies with images that cannot be lossy JPEG compressed include:
Images where the number of bits stored is greater than 12
The current algorithms do not support compression of studies where the number of bits stored is greater than 12, as this causes a loss of data. The number of bits allocated is defined by the bits_stored tag in the DICOM file.
Images with embedded overlay data in the pixels
Since lossy JPEG compression can remove embedded overlay data, the overlay_bits_allocated tag is checked to determine if the image is considered to have an embedded overlay. If the value is 1, the image can be compressed unless the overlay_bits_allocation value is equal to the bits_allocated value of the image, in which case the data is considered to be embedded overlay and the image cannot be compressed.
Images where photometric interpretation must be considered
If the photometric interpretation tag in the DICOM file is set to Palette Color or RGB, the study is not lossy JPEG compressed.
 | Note: If Compressor runs after presentation states have been created in the IMPAX Client, none of the applied presentation states are displayed on the compressed images. Therefore, avoid using Compressor on images that require presentation states. For more presentation state information, refer to “Publishing softcopy presentation states” (topic number 8555) in the IMPAX 6.5.1 Client Knowledge Base: Extended. |
Lossy compressed status values
If a study has been successfully lossy JPEG compressed, the lossy_compressed status value in the dosr_study table is set to T (True) or C (Cache), depending on the mode of compression, whether or not lossy archiving is enabled. In addition, when lossy archiving is enabled:
If a study was not archived before it was lossy compressed and lossy archiving was enabled, the lossy_compressed_status is set to T (True).
If the study was fully archived before lossy JPEG compression was performed, the lossy_compressed_status is set to C (Cache).
For each lossy compressed image, the original sop_instance_uid for each image is moved to the pre_compressed_uid column in the dosr_object table and the new sop_instance_uid is created.
If a study cannot be lossy JPEG compressed, the lossy_compressed_status is set to E (Error). No changes are made to the sop_instance_uid value.
If a study is lossy JPEG compressed in the cache and the lossless copy is retrieved from the archive, the lossy_compressed_status of the study is set to F (False). The original sop_instance_uid for each image in the study is moved from the pre_compressed_uid to the sop_instance_uid column in the dosr_object table, and the pre_compressed_uid is set to null. The lossy copy is marked for deletion and is deleted by Autopilot.