Usage

Image2Radiomics

Image2Radiomics operates with a PIPELINE file, which contains all the instructions required to process images. This file defines a sequence of modules that are executed sequentially for each image in the folder to be processed. Examples are available in the PIPELINES folder.

Additionally, a CONFIGS file can be used to define configurations for radiomics feature extraction (see Radiomics Configuration File). An example is available in the RADIOMICS_CONFIGS folder. It is also possible to directly use a pyradiomics configuration file with the RADIOMICS module (see RADIOMICS module)

Image2Radiomics can be executed using the i2r.py program to compute radiomics features for all images in a folder, following the instructions specified in a pipeline configuration file.

`i2r.py` can be executed with the following options:

Required:

  • -c <PIPELINE_FILE>: Specifies the pipeline configuration file to use (see Pipeline File).

Optional:

  • -v: Enables verbose output.

  • --log <file.log>: Saves the output to <file.log>.

  • --new_log: Overwrites the previous log file.

  • -i <input folder>: If the <PIPELINE_FILE> does not specify an input folder (e.g., inputFolder: .), the -i option is used to define the input folder.

It is recommended to use a script to submit a job or run i2r, as this allows you to define the computational resources, environment variables, and the PIPELINE file to execute.

Examples of scripts are available in the SGE folder (for SGE clusters), the SLURM folder (for SLURM clusters), and the NoJobScheduler folder (for Linux systems without a job scheduler).

  • qsub_img2radiomics.sh (SGE cluster), sbatch_img2radiomics.sh (SLURM cluster), and img2radiomics.sh (No Job Scheduler) are examples for running img2radiomics on a dataset.

  • qsub_img2radiomics_auto.sh (SGE cluster), sbatch_img2radiomics_auto.sh (SLURM cluster), and img2radiomics_auto.sh (No Job Scheduler) are examples for deploying the model (see next section).

In these scripts, ensure that the variables I2R, CONDA, and C3D are correctly set to the paths of Imgage2Radiomics, Anaconda/Miniconda, and Convert3D, respectively.

With SGE Job Scheduler

You can submit the job using the following command:

qsub SGE/qsub_img2radiomics.sh

With SLURM Job Scheduler

You can submit the job using the following command:

sbatch SLURM/sbatch_img2radiomics.sh

Without a Job Scheduler

(Typically used on a personal computer) You can run the script directly using the following command:

sh NoJobScheduler/sbatch_img2radiomics.sh

Init Auto Folder

init_auto_folder.py can be used to create a folder where data is automatically processed when new files are copied into it.

Example with SGE Job Scheduler:

./init_auto_folder.py -m 15 --job_scheduler SGE --cdelay 5 --tdelay 2 -i $HOME/data/newfiles_SGE -o $HOME/data/tmp/ -r --job_name SGE/qsub_img2radiomics_auto.sh --log $HOME/logs/crontab_img2radiomics_sge.log

Example with SLURM Job Scheduler:

./init_auto_folder.py -m 15 --job_scheduler SLURM --cdelay 5 --tdelay 2 -i $HOME/data/newfiles_SLURM -o $HOME/data/tmp/ -r --job_name SLURM/sbatch_img2radiomics_auto.sh --log $HOME/logs/crontab_img2radiomics_slurm.log

Example without a Job Scheduler

./init_auto_folder.py -m 15 --job_scheduler None --cdelay 5 --tdelay 2 -i $HOME/data/newfiles -o $HOME/data/tmp/ -r --job_name NoJobScheduler/img2radiomics_auto.sh --log $HOME/logs/crontab_img2radiomics.log

Create a Pipeline File

Pipeline files for the training set need to be created manually based on the instructions in the Pipeline File section. You can also edit the example pipeline files included in PIPELINES folder of imgage2radiomics.

To generate a pipeline file for a testing set that matches the instructions from the training set, use the gen_testing_pipeline.py tool. Pass the path to the folder containing the PIPELINE files and radiomics files from the training set using the -r flag. For details on the required contents of the reference folder, refer to the Additional Information about Modules in the Pipeline File section.

gen_testing_pipeline.py can also be used with the -s flag to select the strategy for defining new input and output paths for the testing set:

  • Manual (default): Prompts you for all paths.

  • Suffix: Prompts you for the first path and a path to store new results; all other paths will be generated automatically.

  • Auto: Only prompts for the path to store new results. When using auto mode, run img2radiomics with the -i flag to specify the input folder. This mode is best suited for setting up a folder where new data is automatically processed (see Init Auto Folder for details).

If the model folder also contains a model built with scikit-learn saved as a .pkl file, you can use the -p flag to add a PREDICT module to the testing pipeline file (see PREDICT module).

Example of using `gen_testing_pipeline.py`:

./gen_testing_pipeline_file.py -r /path/to/reference/folder -s auto -p