sct_deepseg_lesion

MS lesion Segmentation using convolutional networks. Reference: Gros C et al. Automatic segmentation of the spinal cord and intramedullary multiple sclerosis lesions with convolutional neural networks. Neuroimage. 2018 Oct 6;184:901-915.

usage: sct_deepseg_lesion -i <file> -c {t2,t2_ax,t2s} [-h]
                          [-centerline {svm,cnn,viewer,file}]
                          [-file_centerline <str>] [-brain {0,1}]
                          [-ofolder <str>] [-r {0,1}] [-v <int>]

MANDATORY ARGUMENTS

-i

Input image. Example: t2.nii.gz

-c

Possible choices: t2, t2_ax, t2s

Type of image contrast.

  • t2: T2w scan with isotropic or anisotropic resolution.

  • t2_ax: T2w scan with axial orientation and thick slices.

  • t2s: T2*w scan with axial orientation and thick slices.

OPTIONAL ARGUMENTS

-centerline

Possible choices: svm, cnn, viewer, file

Method used for extracting the centerline:

  • svm: Automatic detection using Support Vector Machine algorithm.

  • cnn: Automatic detection using Convolutional Neural Network.

  • viewer: Semi-automatic detection using manual selection of a few points with an interactive viewer followed by regularization.

  • file: Use an existing centerline (use with flag -file_centerline)

Default: “svm”

-file_centerline

Input centerline file (to use with flag -centerline file). Example: t2_centerline_manual.nii.gz

-brain

Possible choices: 0, 1

Indicate if the input image contains brain sections (to speed up segmentation). This flag is only effective with -centerline cnn.

Default: 1

-ofolder

Output folder. Example: My_Output_Folder

Default: “/home/docs/checkouts/readthedocs.org/user_builds/spinalcordtoolbox/checkouts/stable/documentation/source”

-r

Possible choices: 0, 1

Remove temporary files.

Default: 1

-v

Possible choices: 0, 1, 2

Verbosity. 0: Display only errors/warnings, 1: Errors/warnings + info messages, 2: Debug mode

Default: 1