App for the manual evaluation of surfaces reconstructed using the MIRTK recon-neonatal-cortex command.
The tools subdirectory contains utility scripts for the import of the data files from the
dHCP project, the execution of the surface reconstruction, and the preparation of the SQLite
database including a script to take screenshots that are presented by the graphical App to
the expert for manual scoring and comparison to the reference method.
To import data files from the dHCP project, use the bin/import command and provide either a
CSV file with columns "SubjectId,SessionId" for each subject session data files to be imported,
or list individual sessions as "SubjectId-SessionId", e.g., "CC00050XX01-7201".
To run the MIRTK surface reconstruction, use the bin/recon command and provide either a
CSV file with columns "SubjectId,SessionId" for each subject session for which to run the
cortical surface reconstruction, or list individual sessions as "SubjectId-SessionId",
e.g., "CC00050XX01-7201".
With all data files available in the images, meshes, and labels subdirectories,
the SQLite database file with corresponding offline rendered slices of automatically
determined ROIs saved to PNG image files can be created using the bin/eval-db command.
This command executes the following steps:
init: Write new SQLite database file and create tables if database file missing.select-rois: Select ROIs and best orthogonal viewing directions if none found in database.take-screenshots: Save screenshots of selected ROIs to PNG image files if files and/or database entry missing.add: Perform all of the above steps.
To add the information of an expert rater, i.e., email address and "password" (stored plain text!),
use the tools/add-rater.py script.
This application was developed with Node.js 7.2. Earlier versions may work as well,
including the version available through the official Ubuntu 14.04 nodejs package.
However, nwbuild to create a package failed with this version.
Therefore, use nvm to install a more recent version with the commands below or
see here
for alternative installation methods.
sudo apt-get install libssl-dev
curl -o- https://raw.githubusercontent.com/creationix/nvm/v0.31.0/install.sh | bash
nvm install 7.2
brew install node npm
This application was developed with NW.js 0.19.0.
npm install -g nw nw-gyp
Required dependencies are
- jQuery
- Tether (required by Bootstrap 4)
- Bootstrap 4
- node-sqlite3
You can install these easily with the following npm commands:
cd app
npm install jquery tether bootstrap@4.0.0-alpha.5
## https://www.npmjs.com/package/sqlite3#building-for-node-webkit
NODE_WEBKIT_VERSION=0.19.0 # see latest version at https://github.com/rogerwang/node-webkit#downloads
npm install sqlite3 --build-from-source --runtime=node-webkit --target_arch=x64 --target=$NODE_WEBKIT_VERSION
An App release package can be build using nw-builder. Additionally, appdmg is used
to create a .dmg file for macOS users. These build dependencies can be installed
with the following npm commands:
npm install -g nw-builder
npm install -g appdmg
To create a tar archive for Linux users, run:
./bin/build-app linux64
To create a dmg file for macOS users, run:
./bin/build-app osx64
When no platform argument is given, the release package for the host platform is build.
Finally, run the evaluation App and open the database file to perform the manual evaluation
of the cortical surface reconstruction results. The assigned scores and comparison choices
are stored in the SQLite database for consecutive analysis. Figures to visualize the results
can be created using the tools/plot-results.py script.