SU-E-J-137: Image Registration Tool for Patient Setup in Korea Heavy Ion Medical Accelerator Center

Authors

  • Kim M,

    1. Department of Biomedical Engineering, Research Institute of Biomedical Engineering, The Catholic University of Korea, Seoul
    2. Borame Medical Center, Seoul National University Hospital, Seoul, Seoul
    3. Korea Institute of Radiological & Medical Sciences, Seoul, Seoul
    Search for more papers by this author
  • Cho W,

    1. Department of Biomedical Engineering, Research Institute of Biomedical Engineering, The Catholic University of Korea, Seoul
    2. Borame Medical Center, Seoul National University Hospital, Seoul, Seoul
    3. Korea Institute of Radiological & Medical Sciences, Seoul, Seoul
    Search for more papers by this author
  • Jung W,

    1. Department of Biomedical Engineering, Research Institute of Biomedical Engineering, The Catholic University of Korea, Seoul
    2. Borame Medical Center, Seoul National University Hospital, Seoul, Seoul
    3. Korea Institute of Radiological & Medical Sciences, Seoul, Seoul
    Search for more papers by this author
  • Suh T

    1. Department of Biomedical Engineering, Research Institute of Biomedical Engineering, The Catholic University of Korea, Seoul
    2. Borame Medical Center, Seoul National University Hospital, Seoul, Seoul
    3. Korea Institute of Radiological & Medical Sciences, Seoul, Seoul
    Search for more papers by this author

Abstract

Purpose:

A potential validation tool for compensating patient positioning error was developed using 2D/3D and 3D/3D image registration.

Methods:

For 2D/3D registration, digitally reconstructed radiography (DRR) and three-dimensional computed tomography (3D-CT) images were applied. The ray-casting algorithm is the most straightforward method for generating DRR. We adopted the traditional ray-casting method, which finds the intersections of a ray with all objects, voxels of the 3D-CT volume in the scene. The similarity between the extracted DRR and orthogonal image was measured by using a normalized mutual information method. Two orthogonal images were acquired from a Cyber-Knife system from the anterior-posterior (AP) and right lateral (RL) views. The 3D-CT and two orthogonal images of an anthropomorphic phantom and head and neck cancer patient were used in this study. For 3D/3D registration, planning CT and in-room CT image were applied. After registration, the translation and rotation factors were calculated to position a couch to be movable in six dimensions.

Results:

Registration accuracies and average errors of 2.12 mm ± 0.50 mm for transformations and 1.23° ± 0.40° for rotations were acquired by 2D/3D registration using an anthropomorphic Alderson-Rando phantom. In addition, registration accuracies and average errors of 0.90 mm ± 0.30 mm for transformations and 1.00° ± 0.2° for rotations were acquired using CT image sets.

Conclusion:

We demonstrated that this validation tool could compensate for patient positioning error. In addition, this research could be the fundamental step for compensating patient positioning error at the first Korea heavy-ion medical accelerator treatment center.

Ancillary