Releases · HealthBioscienceIDEAS/Medical-Image-Registration-Short-Course

Documentation

Modernisation of educational practices for Medical-Image-Registration-Short-Course, adopting sandpaper R package and including:

Medical Image Registration led by @mxochicale, including:
- Discussions of tools for visualisation of volumetric data and the selection of ITK-SNAP due to its well maintained development in source-forge.
- Visualisation of volumetric data with ITK-SNAP
  - adds _dependencies/requirements.txt with conda or python virtual environment
  - use vector file figures to illustrate itk-snap application interaction and its datasets visualisation,
  - Usage of sandpaper features such as spoilers, callouts and code sections for bash and python scripts
  - Updates sections Handling multiple DICOM images Colour overlay, Layer inspector tools, using CT-PET-VI-02/inhale_BH_CT and CT-PET-VI-02/exhale_BH_CT dicoms.
- NiBabel section #14, including:
  - Replicated code in PDF images to python scripts in bash boxes
  - Migrate @jamie-mcclelland`s material adding updates where ITK-SNAP is used
  - adds text/figs changes and clarifications when using inhale_BH_CT and exhale_BH_CT for CT-PET-VI-02 aa5c90d
  - Various final updates to fix imports, no need files and brief details on coordinate system 9d85afb
DemonsReg led by @zsenousy, including
- Step by step explanation for Demons image Registration code using Sandpaper features along with code snippets and description for the whole code.
- Cleaning and modifying the code for live display and navigation between different plots.
- Initial display of 5 subplots has been created (Source, target, warped, deformation field and update field) and then removed later as requested by Course leaders.
- Live update function has been created to plot a figure with 3 subplots (warped image, deformation field, and update field) and a top message to show text information (level, iteration and MSD). This to show live update of registration process without the need to pop up multiple figures during each level and without the need to close/open new figure for each level.
- An interactive final display figure of 3 subplots with key event handler has been created (1- navigation between source, target, warped images, 2- navigation between deformation field and Jacobian and 3- difference image). This final display is only shown after registration levels end.
- Added imaging data for the exercise in 'episodes/data'
- Added the up-to-date code files to 'src/mirc/utils'
Medical Image datasets.
- Discussed and agreement on public datasets from TCIA for this work that includes DICOM for CT and PET data.
- Adds explicit description on the creation of datasets from a public datasets, avoiding sharing links of prepossessed datasets (see future work).
Improving reproducibility of the repo #15 (to add relevant hashes by MX)
- INSTRUCTIONS for installing R, Carpentries, and previewing Sandpaper for multi OS (Linux, Windows and Mac)
- Previewing episodes in any OS platform
- update CONTRIBUTIONS guidelines with better workflows on PRs with multiple branches, cleaning merged branches

Project management

GitHub project management using dashboards, milestones and the common issue tracking and PR review workflow to flesh out deliverables by @thompson318
Creation of initial scalfolding carpentry project with the help of @milanmlft and @davecash75 https://github.com/HealthBioscienceIDEAS/Medical-Image-Registration-Short-Course
Adds details on about section: description, website and topics to improve find-ability of the GitHub repository
Transfer Episode list and learning objectives #10 issue to private repo image-registration-notes for internal discussions and fleshing out public material.
HealthBioscienceIDEAS/image-registration-notes#10
private discussions at image-registration-notes:

New features

Adds the current release summary for version control
Adds ISSUES_TEMPLATE and pull_request_template
Adds discussion forum to engage with audiences

Future work

Data links with DOIs for exercises can potentially be hosted in zenodo, where zenodo host upto 50GB.
Orientation instructions were added but require further review
Add linting process to improve code quality for src/mirc and global imports working in windows, Linux and mac-os.
Split episode 4 into (a) medical image data visualisation and (b) nibabel section
Fix spelling errors in itksnap.Rmd
Realised that comments in long review PRs collapse, which make it is difficult to find it and reopened.
So, we might like to think of better ways to request reviews with more specific feedback to avoid long reviews process.