Brain White Matter Lesion segmentation pipeline using SAMSEG in a docker
This pipeline uses the dockerized version of FreeSurfer (about 10 Gb). This pipeline provides the possibility to perform an automatic segmentation of white matter lesions using SAMSEG from FreeSurfer (SAMSEG).
The first figure below shows the window of this pipeline. This window contains the following information:
-
"MPRAGE" checkbox: used to indicate to the pipeline to use a MPRAGE image for the segmentation
-
"FLAIR" checkbox: used to indicate to the pipeline to use a FLAIR image for the segmentation
-
"Normalization" checkbox: used to indicate to the pipeline to preprocess the images with a normalization step.
-
"Select subjects" input: allows the user to script the automatic segmentation for subjects of the dataset by adding a list BIDS ID (without "sub-") separated by a comma. Possible values are: single BIDS ID (e.g. "001,002,006,013"), multiple folowing BIDS ID (e.g. "001-005" is the same as '001,002,003,004,005"), or all subjects ("all").
-
"Select sessions" input: allows the user to script the automatic segmentation for sessions of subjects of the dataset by adding a list session ID (without "ses-") separated by a comma. Possible values are: single session ID (e.g. "01,02,06,13"), multiple folowing session ID (e.g. "01-05" is the same as '01,02,03,04,05"), or all sessions ("all").
-
"Run Segmentation" button: launch the automatic segmentaion based on all information given by the user
Typically, a segmentation takes about 20 minutes
The user needs to specify the name of the MPRAGE sequence and the FLAIR sequence used in this dataset via the "add_info" dictionnary in the json file of the pipeline (cf. figure below). By default, the FLAIR sequence is "FLAIR" and MPRAGE is "acq-MPRAGE_T1w".
