This tool aims to automatically model the topological folding structure of the human hippocampus. It is currently set up to use sub-millimetric T2w MRI data, but may be adapted for other data types. This can then be used to apply the hippocampal unfolding methods presented in this paper, and ex-vivo subfield boundaries can be topologically applied from this paper.
The overall workflow can be summarized in the following steps:
-
Resampling to a 0.3mm isotropic, coronal oblique, cropped hippocampal block
-
Automatic segmentation of hippocampal tissues and surrounding structures via deep convolutional neural network U-net (Li et al., 2017) OR Manual segmentation of hippocampal tissues and surrounding structures using this protocol
-
Post-processing via fluid label-label registration to a high resolution, topoligically correct averaged template
-
Imposing of coordinates across the anterior-posterior, proximal-distal, and laminar dimensions of hippocampal grey matter via solving the Laplace equation
-
Extraction of a grey matter mid-surface and morpholigical features (thickness, curvature, gyrification index, and, if available, quantitative MRI values sampled along the mid-surface)
-
Quality assurance via inspection of Laplace gradients, grey matter mid-surface, and flatmapped features
-
Application of subfield boundaries according to predifined topological coordinates
under development
Currently, Matlab code is provided for steps 2-6, with these dependencies
Coming soon: fully containerized BIDSapp with MCR compiled Matlab code and all dependencies.
Simple example Matlab scripts are provided showing batching of subjects that are already resampled, or running a new subject starting from a whole-brain T2w image.
If you are using ComputeCanada, you can adapt one of the example scripts for your file names & directories and then submit it using:
regularSubmit matlab -r example_batchScript_manualSeg
Note: you must have neuroglia-core, neuroglia-helpers, and matlab (license + module loaded). See the Khan lab ComputeCanada wiki (or request access) at https://osf.io/4u5jr.
under development
This section breaks down the step 1) above into more detail. Note that all fully-automated segmentation and unfolding should be inspected prior to drawing conclusions! Useful tools are included for visualization.