Processing and visualising cortical surfaces¶

Cortical surface processing¶

Preprocessed cortical data are provided, but participants are also encouraged to develop/utilise other cortical metrics for prediction. e.g. cortical surface area. However, their use should be clearly motivated.

At this stage, we ask participants to derive cortical metrics from structural data only. We aim to introduce further metrics (e.g. functional/structural connectivity, resting-state fMRI spatial maps, and subcortical metrics) in future SLCN challenges.

A useful software for deriving cortical metrics is Workbench Command, available as part of Connectome Workbench, which is freely available to download here. A full list of Workbench Command functions can be found here.

Cortical surface visualisation¶

Visualisations are an essential component of interpreting cortical surface data and any deep learning model applied to the cortical surface. As such, interpretability will be an essential component of the evaluation for any teams wishing to submit a full paper on their method to the MLCN workshop (more details under the Evaluation page).

A suitable tool for visualising cortical surfaces is wb_view, available as part of Connectome Workbench (graphical user interface shown below). A tutorial for wb_view is here, and a Guide to Workbench Annotations is here. This is only a suggestion, and participants are welcome to use whichever data visualisation tool they are familiar with.