Remove Dura

This algorithm removes some dura present after skull stripping.

Input Types

This algorithm requires the original image (i.e. image with skull) as well as its white matter and gray matter memberships and a filled whiter matter mask. TOADS2008 can be used to get the segmentations. The JIST layout explains the usage.

The dura is removed using the following procedure.

1) Add WM and GM memberships and run topology correction starting from the WM Mask.
2) Correct the topology of the WM membership starting from the WM Mask.
3) Run TGDM on the pial surface estimate with an additional force that regularizes cortical thickness.
4) Create mask volume using the pial surface level set.
5) Dilate the brain mask and apply it to the original data.

Input and output

Original Volume	The raw MR volume with skull.
WM Volume	Filled white matter membership of the MR image. The membership values has the range of [0,255]. TOADS2008 can be used to get the mebership.
GM Volume	Gray matter membership of the MR image. The membership values has the range of [0,255].
WM Mask	White matter mask, foreground has the value 255 and background 0.

Parameters

Curvature force	Controls the smoothness of the brain mask. Increasing the parameter increases the smoothness
Level Set Alignment force
Pressure force	Controls the effect of the pressure force in TGDM.
Target Membership	The iso level for the GM membership function, defining the boundary of brain.
Mask Translation	Amount to dilate the final volumetric brain mask.
Level Set Translation	Amount to translate the brain mask level set surface.
Maximum Iteration	TGDM iterations.
WM and GM regularization amount	Amount of regularization enforced on the WM membership topology correction. Smoothness increases with the increase of regularization.

Example Usage

Example JIST layout
Example input image with skull.
Input WM membership from TOADS2008.
Input GM membership from TOADS2008.
Input WM mask from TOADS2008.
Output stripped volume.
Output mask.

---