Skip to content

Outer to inner surface mesh deformation using radial distance map for human fetal MRI

License

Notifications You must be signed in to change notification settings

FNNDSC/pl-gifit

Repository files navigation

Inner Subplate Surface Fit

Version MIT License ci

pl-gifit is a ChRIS ds plugin which performs mesh deformation. Different deformation parameters are used depending on the gyrification index of the input surface. Similar to the CIVET function expand_from_white, pl-gifit is a wrapper around surface_fit.

Models

The primary goal of pl-gifit is to deform inwards mesh deformation from the outer subplate surface to the inner subplate surface for in-vivo human brain MRI.

A "model" for pl-gifit refers to a CSV file containing parameters for surface_fit. A model CSV file optimized for inner subplate fitting is provided as the default.

One "schedule" of parameters is run per brain hemisphere. A "schedule" is one or more runs of surface_fit. Sometimes, it is ideal to run surface_fit multiple times with different paremeters. Each run with different parameters is called a "stage." One row of a model CSV file corresponds to one "stage" of a "schedule."

Usage

pl-gifit is a ChRIS plugin, meaning it can run from either within ChRIS or the command-line.

The easiest way to run pl-gifit is using this pipeline which accepts subplate segmentations as inputs and extracts both inner and outer subplate surfaces. https://github.com/FNNDSC/Fetal_Brain_MRI_Surface_Extraction_Pipeline/blob/main/Snakefile

To run pl-gifit manually, you must prepare its input directory according to its expected convention. The input directory should contain subdirectories where each subdirectory contains a brain. A subdirectory should contain either one or both brain hemispheres. Files for left and right hemisphere should be prefixed with lh. and rh., respectively. Each hemisphere should have a laplacian map (.mnc) and a starting surface (.obj). To obtain surfaces of the inner subplate, provide a radial distance map generated from the inner subplate segmentation and the outer subplate (white matter) surface. Use pl-bichamfer to create the distance map and pl-fetal-surface-extract to extract the starting surface from white matter segmentation.

For example:

incoming
├── BCH_0062_s1
│   ├── lh.spinner.mnc
│   ├── lh.wm_81920.obj
│   ├── rh.spinner.mnc
│   └── rh.wm_81920.obj
└── BCH_0063_s1
    ├── lh.spinner.mnc
    ├── lh.wm_81920.obj
    ├── rh.spinner.mnc
    └── rh.wm_81920.obj

With all your data organized, simply run

apptainer run docker://ghcr.io/fnndsc/pl-gifit incoming/ outgoing/

About

Outer to inner surface mesh deformation using radial distance map for human fetal MRI

Resources

License

Stars

Watchers

Forks

Packages