10.35. Registration Operator

10.35.1. Overview

Registration operator in Clara Deploy SDK is based on ITK.

10.35.2. Data Input

Input1: Folder with Input images Input2: Target image from Input1 Input3: Output folder name (to write registered images)

Supports DICOM or Nifti data ingestion.

10.35.3. Data Output

Registered output can be saved as nifti or nifti_series

10.35.4. Parameters

Registration Framework supports:

Transform: Affine Metric: Mattes Mutual Information Optimizer: Regular step Gradient Descent Interpolation: Linear Registrered Image Pixel Type: Unsigned short (16 bits)

Parameters for Metric and Optimizer can be tuned. Registration pipeline provides the mechanism to ingest several values of those parameters into the pipeline.

10.35.5. Execution steps.

For local execution, following command must be executed in sequence on a command prompt.

  • Execute ‘docker images’ and locate the ‘TAG’ of clara/python-base image.

  • Execute following to tag python-base image, use value of ‘TAG’ as identified in above step:

    docker tag clara/python-base:'TAG' clara/python-base:latest
    
  • Locate the folder containing code for registration operator (app_registration) in the sdk. Execute ‘make’ inside the folder.

  • Locate the input folder. Keep relevant datasets in the folder. Store the location of input folder as ‘input_data_folder’

  • Locate the output folder. Store the location of output folder as ‘output_data_folder’

  • Locate the target image. Store the target image as ‘target_image’. If input datasets are in DICOM format, ‘target_image’ will be a folder otherise it will be the filename.

  • If input datasets are in DICOM format: .. code-block:: bash

    ./register.sh input_data_folder output_data_folder target_image

  • If input datasets are in NIFTI format: .. code-block:: bash

    ./register_nifti.sh input_data_folder output_data_folder target_image