deepsulci.sulci_labeling.capsul.labeling.SulciDeepLabeling

SulciDeepLabeling

Process to label a new graph using a 3D U-Net convolutional neural network.

The process can work using a GPU or on CPU. It requires a fair amount of RAM memory (about 4-5 GB). If not enough memory can be allocated, the process will abort with an error (thus will not hang the whole machine).

Note

• Type ‘SulciDeepLabeling.help()’ for a full description of this process parameters.

• Type ‘<SulciDeepLabeling>.get_input_spec()’ for a full description of this process input trait types.

• Type ‘<SulciDeepLabeling>.get_output_spec()’ for a full description of this process output trait types.

Inputs

[Mandatory]

graph: a string or os.PathLike object ([‘File’] - mandatory)

input graph to segment

roots: a string or os.PathLike object ([‘File’] - mandatory)

root file corresponding to the input graph

model_file: a string or os.PathLike object ([‘File’] - mandatory)

file (.mdsm) storing neural network parameters

param_file: a string or os.PathLike object ([‘File’] - mandatory)

file (.json) storing the hyperparameters (cutting threshold)

rebuild_attributes: a boolean ([‘Bool’] - mandatory)

No description.

skeleton: a string or os.PathLike object ([‘File’] - mandatory)

skeleton file corresponding to the input graph

allow_multithreading: a boolean ([‘Bool’] - mandatory)

No description.

cuda: an integer ([‘Int’] - mandatory)

device on which to run the training(-1 for cpu, i>=0 for the i-th gpu)

fix_random_seed: a boolean ([‘Bool’] - mandatory)

Use same random sequence

Outputs

labeled_graph: a string or os.PathLike object ([‘File (filename: input)’] - mandatory)

output labeled graph