An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy

G.J.J. Warmerdam, P. Steininger, M. Neuner, G. Sharp, B. Winey

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

Purpose: To determine the robustness and accuracy of an open source 2D/3D GPU accelerated image registration algorithm in the context of cranial image guided radiotherapy. Methods: The open source 2D/3D image registration algorithm, Reg23, has been released under the GNU license. The algorithm utilizes an iterative digitally reconstructed radiograph (DRR) approach to the image registration problem. The DRR generator is accelerated on a GPU to rapidly iterate the optimization process. Multiple cost functions are supported and were analyzed. Robustness was determined by comparing a baseline set of orthogonal kV images of a cranial phantom with a predetermined isocenter to the planned isocenter in the CT image set and introducing more than 6000 combinations of rotations and translation to the position of the isocenter in the CT. Accuracy and time efficiency of various cost function were analyzed for the virtual patient shifts. Furthermore, a set of 43 experimental orthogonal images were acquired with a linac mounted kV imaging system of predetermined physical shifts which were compared to the results of the Reg23 algorithm. Results: The Reg23 algorithm was found to be accurate to 0.04±0.02mm for the virtual isocenter shifts and 0.23±0.40mm for real images compared to the CBCT registration results. Time to solution could be reduced from >70 s to <40 s without a significant change in the algorithm accuracy depending upon the cost function employed. Conclusions: The Reg23 algorithm is robust and sensitive to sub-mm variations of virtual shifts of the isocenter position. The Normalized Cross Correlation (NCC) cost function was determined to be most accurate and fastest for cranial image registration. For real experimental data, the Gradent Difference (GD) cost function was most accurate and both GD and NCC delivered results accurate to within 0.5 mm and 0.4° when compared to CBCT/CT registrations. © 2012 American Association of Physicists in Medicine
Original languageEnglish
Title of host publicationFifty‐fourth annual meeting of the American Association of physicists in medicine
Pages4001-4001
DOIs
Publication statusPublished - 2012

Publication series

NameMedical Physics
Volume39
ISSN (Print)0094-2405

Fingerprint

Image-Guided Radiotherapy
Costs and Cost Analysis
Licensure
Medicine

Cite this

Warmerdam, G. J. J., Steininger, P., Neuner, M., Sharp, G., & Winey, B. (2012). An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy. In Fifty‐fourth annual meeting of the American Association of physicists in medicine (pp. 4001-4001). (Medical Physics; Vol. 39). https://doi.org/10.1118/1.4736326
Warmerdam, G.J.J. ; Steininger, P. ; Neuner, M. ; Sharp, G. ; Winey, B. / An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy. Fifty‐fourth annual meeting of the American Association of physicists in medicine. 2012. pp. 4001-4001 (Medical Physics).
@inproceedings{3e24bbac37c54eb6ab0f659eb2668108,
title = "An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy",
abstract = "Purpose: To determine the robustness and accuracy of an open source 2D/3D GPU accelerated image registration algorithm in the context of cranial image guided radiotherapy. Methods: The open source 2D/3D image registration algorithm, Reg23, has been released under the GNU license. The algorithm utilizes an iterative digitally reconstructed radiograph (DRR) approach to the image registration problem. The DRR generator is accelerated on a GPU to rapidly iterate the optimization process. Multiple cost functions are supported and were analyzed. Robustness was determined by comparing a baseline set of orthogonal kV images of a cranial phantom with a predetermined isocenter to the planned isocenter in the CT image set and introducing more than 6000 combinations of rotations and translation to the position of the isocenter in the CT. Accuracy and time efficiency of various cost function were analyzed for the virtual patient shifts. Furthermore, a set of 43 experimental orthogonal images were acquired with a linac mounted kV imaging system of predetermined physical shifts which were compared to the results of the Reg23 algorithm. Results: The Reg23 algorithm was found to be accurate to 0.04±0.02mm for the virtual isocenter shifts and 0.23±0.40mm for real images compared to the CBCT registration results. Time to solution could be reduced from >70 s to <40 s without a significant change in the algorithm accuracy depending upon the cost function employed. Conclusions: The Reg23 algorithm is robust and sensitive to sub-mm variations of virtual shifts of the isocenter position. The Normalized Cross Correlation (NCC) cost function was determined to be most accurate and fastest for cranial image registration. For real experimental data, the Gradent Difference (GD) cost function was most accurate and both GD and NCC delivered results accurate to within 0.5 mm and 0.4° when compared to CBCT/CT registrations. {\circledC} 2012 American Association of Physicists in Medicine",
author = "G.J.J. Warmerdam and P. Steininger and M. Neuner and G. Sharp and B. Winey",
year = "2012",
doi = "10.1118/1.4736326",
language = "English",
series = "Medical Physics",
pages = "4001--4001",
booktitle = "Fifty‐fourth annual meeting of the American Association of physicists in medicine",

}

Warmerdam, GJJ, Steininger, P, Neuner, M, Sharp, G & Winey, B 2012, An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy. in Fifty‐fourth annual meeting of the American Association of physicists in medicine. Medical Physics, vol. 39, pp. 4001-4001. https://doi.org/10.1118/1.4736326

An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy. / Warmerdam, G.J.J.; Steininger, P.; Neuner, M.; Sharp, G.; Winey, B.

Fifty‐fourth annual meeting of the American Association of physicists in medicine. 2012. p. 4001-4001 (Medical Physics; Vol. 39).

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy

AU - Warmerdam, G.J.J.

AU - Steininger, P.

AU - Neuner, M.

AU - Sharp, G.

AU - Winey, B.

PY - 2012

Y1 - 2012

N2 - Purpose: To determine the robustness and accuracy of an open source 2D/3D GPU accelerated image registration algorithm in the context of cranial image guided radiotherapy. Methods: The open source 2D/3D image registration algorithm, Reg23, has been released under the GNU license. The algorithm utilizes an iterative digitally reconstructed radiograph (DRR) approach to the image registration problem. The DRR generator is accelerated on a GPU to rapidly iterate the optimization process. Multiple cost functions are supported and were analyzed. Robustness was determined by comparing a baseline set of orthogonal kV images of a cranial phantom with a predetermined isocenter to the planned isocenter in the CT image set and introducing more than 6000 combinations of rotations and translation to the position of the isocenter in the CT. Accuracy and time efficiency of various cost function were analyzed for the virtual patient shifts. Furthermore, a set of 43 experimental orthogonal images were acquired with a linac mounted kV imaging system of predetermined physical shifts which were compared to the results of the Reg23 algorithm. Results: The Reg23 algorithm was found to be accurate to 0.04±0.02mm for the virtual isocenter shifts and 0.23±0.40mm for real images compared to the CBCT registration results. Time to solution could be reduced from >70 s to <40 s without a significant change in the algorithm accuracy depending upon the cost function employed. Conclusions: The Reg23 algorithm is robust and sensitive to sub-mm variations of virtual shifts of the isocenter position. The Normalized Cross Correlation (NCC) cost function was determined to be most accurate and fastest for cranial image registration. For real experimental data, the Gradent Difference (GD) cost function was most accurate and both GD and NCC delivered results accurate to within 0.5 mm and 0.4° when compared to CBCT/CT registrations. © 2012 American Association of Physicists in Medicine

AB - Purpose: To determine the robustness and accuracy of an open source 2D/3D GPU accelerated image registration algorithm in the context of cranial image guided radiotherapy. Methods: The open source 2D/3D image registration algorithm, Reg23, has been released under the GNU license. The algorithm utilizes an iterative digitally reconstructed radiograph (DRR) approach to the image registration problem. The DRR generator is accelerated on a GPU to rapidly iterate the optimization process. Multiple cost functions are supported and were analyzed. Robustness was determined by comparing a baseline set of orthogonal kV images of a cranial phantom with a predetermined isocenter to the planned isocenter in the CT image set and introducing more than 6000 combinations of rotations and translation to the position of the isocenter in the CT. Accuracy and time efficiency of various cost function were analyzed for the virtual patient shifts. Furthermore, a set of 43 experimental orthogonal images were acquired with a linac mounted kV imaging system of predetermined physical shifts which were compared to the results of the Reg23 algorithm. Results: The Reg23 algorithm was found to be accurate to 0.04±0.02mm for the virtual isocenter shifts and 0.23±0.40mm for real images compared to the CBCT registration results. Time to solution could be reduced from >70 s to <40 s without a significant change in the algorithm accuracy depending upon the cost function employed. Conclusions: The Reg23 algorithm is robust and sensitive to sub-mm variations of virtual shifts of the isocenter position. The Normalized Cross Correlation (NCC) cost function was determined to be most accurate and fastest for cranial image registration. For real experimental data, the Gradent Difference (GD) cost function was most accurate and both GD and NCC delivered results accurate to within 0.5 mm and 0.4° when compared to CBCT/CT registrations. © 2012 American Association of Physicists in Medicine

U2 - 10.1118/1.4736326

DO - 10.1118/1.4736326

M3 - Conference contribution

T3 - Medical Physics

SP - 4001

EP - 4001

BT - Fifty‐fourth annual meeting of the American Association of physicists in medicine

ER -

Warmerdam GJJ, Steininger P, Neuner M, Sharp G, Winey B. An open-source 2D/3D-image-registration algorithm : cranial image guided radiotherapy. In Fifty‐fourth annual meeting of the American Association of physicists in medicine. 2012. p. 4001-4001. (Medical Physics). https://doi.org/10.1118/1.4736326