Parallel transfers

This release contains a rewrite of the image transfer functionality. It is now completely expressed as Akka Streams graph and uses connection pools for HTTP requests. This enables sending files in parallel and transfers are now typically a factor 10 faster.

Updated version string in UI

This release is identical to 1.3.1 except for an update to the version string in the UI.

S3-upload With Retry

Users deploying slicebox to AWS with S3 as storage backend have reported problems with intermittently failing uploads making transfer and upload of large quantities of files difficult. This patch release addresses this issue and provides a series of minor updates and bugfixes.

Streaming DICOM

This release comes with a significant change in the technology used to ingest, parse, modify and transfer DICOM data. It is built on the AKKA Stream implementation of the Reactive Streams protocol, which means that DICOM data never fully buffered in memory. Instead, minimal fragments are read, processed and passed on to storage, another instance, or the client. This makes the application more stable and predictable in terms of system resources usage. Streams use back-pressure to communicate between network nodes the rate at which data should be produced and consumed. This avoids build-up of buffered data at any node and makes network utilization more even. These changes makes slicebox more stable and better performing under load.

Sensible sorting of transferred images

Fixed failing tests and implemented more sensible sorting of images in outgoing transactions.

Box transfer cleanup

This releases fixes a few bugs and nuisances regarding box-to-box transfer of images.

Rollback to stable mysql connector

This release rolls a dependency for using mysql databases back from milestone release 6 to stable release 5 because of trouble with specifying the use of ANSI quotes in the jdbc url.

Akka-HTTP

This release contains no known changes to API or functionality (hence the patch update) but comes with major changes in the libraries used. Previously, Spray was used to provide a REST API to slicebox internal API which is based on Akka actors. Spray has become part of the Akka project under the name Akka-HTTP and was recently made available in a first stable release. We have therefore upgraded from Spray to Akka-HTTP in this release which makes major future improvements such as streaming handling of DICOM data possible.

In this release we have also changed JSON libraries from spray-json to play-json since we have found play-json to be faster in demanding applications.

Cleaner anonymization records

This patch changes the anonymization service in the following two ways:

• Anonymization is not (doubly) carried out for anonymous datasets unless there are DICOM tag value mappings present
• Reverse anonymization is not attempted for not anonymous datasets as such datasets should not risk being altered

Polish and bugfixes

This patch fixes a few minor bugs and introduces a possibility to purge anonymization keys for which all corresponding images have been deleted. This is suitable in situations where you would like to keep a minimal amount of protected health information in your system.