Compression and decompression

[stefanoostwegel]

Hi all,

We have created a tool that can retrieve massive amounts of DICOM studies from an archive, to store it to a predefined destination. This helps with migration cases and research cases.
We use a c-move to retrieve the data to our local SCP, and then use a c-store to drop it off at the desired destination.
For clinical cases, we would love to use compressed file formats as it saves a lot of space and time.
Unfortunately, the research cases require uncompressed files because the software and workflows using this data can't handle the compressed files.

When doing this with large amount of studies, it takes a long time when doing this with decompressed files.
Is there a easy way to retrieve the files always compressed, then decompress before sending? Or is the overhead of decompression to big that it doesnt provide any time saving?

Maybe there is easier ways to send compressed files as decompressed in a c-sotre, or do you always need to run pydicom.decompress() before transferring it as decompressed?

Thank you

[scaramallion]

Whether or not you get compressed or uncompressed data depends on the transfer syntaxes offered by the SCP. If you have your Move Destination's storage SCP only support compressed transfer syntaxes then that will be what's sent. This depends on the storage SCU being able to send compressed data, of course.

[stefanoostwegel]

Thanks for the reply. So may-be i explained this a bit the wrong way.

1. I request our archive to store compressed data to our SCP.
2. We have stored all dicom files of this series as JPEG lossless (1.2.840.10008.1.2.4.70)
3. When sending the files from disk, i loop over the folder and send them within the association.
4. If i force this association to only send decompressed (1.2.840.10008.1.2), would i still need to decompress() the files before sending? Or does the SCU take care of that while sending?

Because if i first have to loop over all the files and try to decompress() them, this could be timeconsuming and i might not be better off in comparison of requesting in uncompressed to begin with (step 1 and 2).
I know i have some options to perform this, and i can make it work however i like, but im looking for the more 'common sense' solution that fitst the pynetdicom and pydicom options the best.

[scaramallion]

Yes, you'll need to decompress manually beforehand.