view TODO @ 2106:5144d59d0fb2 deep-learning

integration mainline->deep-learning
author Sebastien Jodogne <s.jodogne@gmail.com>
date Wed, 22 Nov 2023 07:44:11 +0100
parents fdb012c86a75 5653b6335642
children de049fd88697
line wrap: on
line source

Update OrthancFramework to 1.12.1 to benefit from https://orthanc.uclouvain.be/hg/orthanc/rev/57c0fba5e5c7

======================
== Stone Web viewer ==
======================

------------
Improvements
------------

* Save/load annotations as attachments of their parent series
  (including detection of collisions thanks to revisions).
  https://orthanc.uclouvain.be/book/faq/features.html#metadata-attachments
  
* Internationalization and translations (i18n). Some possible libraries:
  https://phrase.com/blog/posts/the-best-javascript-i18n-libraries/

* Configurable keyboard shortcuts. See Osimis Web viewer:
  https://bitbucket.org/osimis/osimis-webviewer-plugin/src/master/doc/default-configuration.json

* Handle mobile gestures.

* Display GSPS layers (Grayscale Softcopy Presentation State).

* Use "ImagerPixelSpacing" (0018,1164) tag if "PixelSpacing" is missing

* Override which tags are displayed in the image information in the
  top-left, top-right and bottom-right pannels.

  For reference, in the case of MRI, "Echo Time" (0018,0081),
  "Repetition Time" (0018,0080), "Magnetic Field Strength (0018,0087)
  and "Slice Thickness" (0018,0050) would be useful to have.

  https://groups.google.com/g/orthanc-users/c/_kDp_ieYTgI/m/KHBxpSSOCQAJ
  
* Add a button (and/or a keyboard shortcut) to select the next series
  in the main viewport. In Osimis viewer, this was implemented by
  up/down arrow keys (prev/next series).
  https://groups.google.com/g/orthanc-users/c/u_lH9aqKsdw/m/KQ7U9CkiAAAJ.

* Minor: Rotate the anchors of the text after rotation of the display.


------------
Known issues
------------

* Compatibility with Google Chrome browser on iPad Air 4:
  https://groups.google.com/g/orthanc-users/c/7SgedbIiA2k/
  https://groups.google.com/g/orthanc-users/c/RfQJFgkOUYY/m/za7rkcLNBQAJ

-----------------
Code refactorings
-----------------

* Replace "ParsedDicomCache" by "Orthanc::ParsedDicomCache" (new in
  Orthanc framework 1.9.0)

* Replace calls to /studies/.../series/.../metadata by calls to
  /instances?StudyInstanceUID=...&SeriesInstanceUID=...&includefield=...&includefield=...
  to benefit from ExtraMainDicomTags stored in DB and avoid reading from disk when not necessary.
  This way, we request only the tags that the viewer is going to use.
  The /metadata requests all tags and therefore will read from the storage unless you have
  specified "DicomWeb.SeriesMetadata": "MainDicomTags"; this is an acceptable workaround
  for now but the first approach is cleaner and removes dependencies between the DicomWeb config
  and stone.  It might also be more efficient on other backends as well.

--------
Wishlist
--------

* Display video files even if the Orthanc REST API is not available
  (using pure DICOMweb). This could possible be done using the
  DICOMweb Bulk Data URI, and/or a dedicated JavaScript video player.

* "Start the Viewer with a specific Dicom file and display the
  specific images. For example I start the Viewer with dicomFile and
  open slide1, slide4, slide8, slide12 ?"
  https://groups.google.com/g/orthanc-users/c/6KSe0WEEfco/m/nOIl57EzAAAJ

* Configuration option related to the "DicomSource" class in order to
  systematically request server-side transcoding.

* Fully implement "Basic Image Review" profile from IHE:
  https://wiki.ihe.net/index.php/Basic_Image_Review

* Support hanging protocols (automatically adapt the layout given the
  type of modality to be displayed):
  https://groups.google.com/g/orthanc-users/c/bGtK3q9ZUmg/m/gr8kCcVWCAAJ

* Create secondary capture DICOM images with the annotations burned in:
  https://groups.google.com/g/orthanc-users/c/T3h-6dIjvww/m/sVtHxwT_AQAJ

* Configure the default layout (e.g. 1x2 instead of 1x1) and the
  default view of the left toolbox:
  https://groups.google.com/g/orthanc-users/c/y2B_RIzegNc/m/H9MhY9y2AgAJ

* Display the PatientName & ID in the page title such that it appears 
  in the browser tab name
  https://groups.google.com/g/orthanc-users/c/F7cK8axehfA/m/ku-CPsvbAQAJ

* Configuration option (plus GET parameter in the URL) to set where to
  start in the series by default (plus possibly a dictionary mapping
  SOP Class UIDs for per-modality setting):
  https://groups.google.com/g/orthanc-users/c/tHPMBHe-Z-A/m/XXlRBdhoAgAJ

* Sort the studies by date in the study picker:
  https://discourse.orthanc-server.org/t/the-study-pickers-list-does-not-sort-properly/3489

--------
Won't do
--------

* Debian packaging, which requires source-level build dependencies:
  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=975069#22

  => A solution for AMD64 architectures is to use the standalone
  Debian repository that is provided by Sebastien Jodogne:
  https://orthanc.uclouvain.be/book/users/cookbook.html#obtaining-binaries



==================================
== Stone of Orthanc C++ library ==
==================================

-------
General
-------

* Documentation

-------------
Optimizations
-------------

* Speedup RT-STRUCT projection in sagittal + coronal views
* Speedup RT-STRUCT loading
* "ParseDicomFromOrthanc": new command

-----------------
Platform-specific
-----------------

* WebAssembly: Possible issue in "emscripten_set_timeout()" if the
  Oracle is destroyed while a sleep event is still being processed
  (the callback payload would be invalid). This is not a problem as
  long as the Oracle singleton is never destroyed. A cleaner solution
  would be similar to "WebAssemblyViewport::CreateObjectCookie()".

* Add precompiled headers for Microsoft Visual Studio.