Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy

Tonghe Wang; Joseph Harms; Yang Lei; Beth Ghavidel; William Stokes; Tian Liu; Walter J. Curran; Mark McDonald; Jun Zhou; Xiaofeng Yang

doi:10.1117/12.2549275

Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy

Tonghe Wang
, Joseph Harms
, Yang Lei
, Beth Ghavidel
, William Stokes
, Tian Liu
, Walter J. Curran
, Mark McDonald
, Jun Zhou
, Xiaofeng Yang

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

Proton radiation therapy has shown highly conformal distribution of prescribed dose in target with outstanding normal tissue sparing stemming from its steep dose gradient at the distal end of the beam. However, the uncertainty in everyday patient setup can lead to a discrepancy between treatment dose distribution and the planning dose distribution. Cone-beam CT (CBCT) can be acquired daily before treatment to evaluate such inter-fraction setup error, while a further evaluation on resulted dose distribution error is currently not available. In this study, we developed a novel deep-learning based method to predict the relative stopping power maps from daily CBCT images to allow for online dose calculation in a step towards adaptive proton radiation therapy. 20 head-and-neck patients with CT and CBCT images are included for training and testing. Our CBCT RSP results were evaluated with RSP maps created from CT images as the ground truth. Among all the 20 patients, the averaged mean absolute error between CT-based and CBCT-based RSP was 0.04±0.02, the averaged mean error was -0.01±0.03 and the averaged normalized correlation coefficient was 0.97±0.01. The proposed method provides sufficiently accurate RSP map generation from CBCT images, possibly allowing for CBCT-guided adaptive treatment planning for proton radiation therapy.

Original language	English
Title of host publication	Medical Imaging 2020
Subtitle of host publication	Physics of Medical Imaging
Editors	Guang-Hong Chen, Hilde Bosmans
Publisher	SPIE
ISBN (Electronic)	9781510633919
DOIs	https://doi.org/10.1117/12.2549275
State	Published - 2020
Event	Medical Imaging 2020: Physics of Medical Imaging - Houston, United States Duration: Feb 16 2020 → Feb 19 2020

Publication series

Name	Progress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume	11312
ISSN (Print)	1605-7422

Conference

Conference	Medical Imaging 2020: Physics of Medical Imaging
Country/Territory	United States
City	Houston
Period	02/16/20 → 02/19/20

Keywords

CBCT
Machine learning
Proton
Stopping power

Access to Document

10.1117/12.2549275

Cite this

Wang, T., Harms, J., Lei, Y., Ghavidel, B., Stokes, W., Liu, T., Curran, W. J., McDonald, M., Zhou, J., & Yang, X. (2020). Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy. In G.-H. Chen, & H. Bosmans (Eds.), Medical Imaging 2020: Physics of Medical Imaging Article 113124L (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 11312). SPIE. https://doi.org/10.1117/12.2549275

@inproceedings{e828ccde3bad4338bb41e1f7e56da1ca,

title = "Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy",

abstract = "Proton radiation therapy has shown highly conformal distribution of prescribed dose in target with outstanding normal tissue sparing stemming from its steep dose gradient at the distal end of the beam. However, the uncertainty in everyday patient setup can lead to a discrepancy between treatment dose distribution and the planning dose distribution. Cone-beam CT (CBCT) can be acquired daily before treatment to evaluate such inter-fraction setup error, while a further evaluation on resulted dose distribution error is currently not available. In this study, we developed a novel deep-learning based method to predict the relative stopping power maps from daily CBCT images to allow for online dose calculation in a step towards adaptive proton radiation therapy. 20 head-and-neck patients with CT and CBCT images are included for training and testing. Our CBCT RSP results were evaluated with RSP maps created from CT images as the ground truth. Among all the 20 patients, the averaged mean absolute error between CT-based and CBCT-based RSP was 0.04±0.02, the averaged mean error was -0.01±0.03 and the averaged normalized correlation coefficient was 0.97±0.01. The proposed method provides sufficiently accurate RSP map generation from CBCT images, possibly allowing for CBCT-guided adaptive treatment planning for proton radiation therapy.",

keywords = "CBCT, Machine learning, Proton, Stopping power",

author = "Tonghe Wang and Joseph Harms and Yang Lei and Beth Ghavidel and William Stokes and Tian Liu and Curran, \{Walter J.\} and Mark McDonald and Jun Zhou and Xiaofeng Yang",

note = "Publisher Copyright: {\textcopyright} 2020 SPIE; Medical Imaging 2020: Physics of Medical Imaging ; Conference date: 16-02-2020 Through 19-02-2020",

year = "2020",

doi = "10.1117/12.2549275",

language = "English",

series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

publisher = "SPIE",

editor = "Guang-Hong Chen and Hilde Bosmans",

booktitle = "Medical Imaging 2020",

}

Wang, T, Harms, J, Lei, Y, Ghavidel, B, Stokes, W, Liu, T, Curran, WJ, McDonald, M, Zhou, J & Yang, X 2020, Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy. in G-H Chen & H Bosmans (eds), Medical Imaging 2020: Physics of Medical Imaging., 113124L, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 11312, SPIE, Medical Imaging 2020: Physics of Medical Imaging, Houston, United States, 02/16/20. https://doi.org/10.1117/12.2549275

Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy. / Wang, Tonghe; Harms, Joseph; Lei, Yang et al.
Medical Imaging 2020: Physics of Medical Imaging. ed. / Guang-Hong Chen; Hilde Bosmans. SPIE, 2020. 113124L (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 11312).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy

AU - Wang, Tonghe

AU - Harms, Joseph

AU - Lei, Yang

AU - Ghavidel, Beth

AU - Stokes, William

AU - Liu, Tian

AU - Curran, Walter J.

AU - McDonald, Mark

AU - Zhou, Jun

AU - Yang, Xiaofeng

PY - 2020

Y1 - 2020

N2 - Proton radiation therapy has shown highly conformal distribution of prescribed dose in target with outstanding normal tissue sparing stemming from its steep dose gradient at the distal end of the beam. However, the uncertainty in everyday patient setup can lead to a discrepancy between treatment dose distribution and the planning dose distribution. Cone-beam CT (CBCT) can be acquired daily before treatment to evaluate such inter-fraction setup error, while a further evaluation on resulted dose distribution error is currently not available. In this study, we developed a novel deep-learning based method to predict the relative stopping power maps from daily CBCT images to allow for online dose calculation in a step towards adaptive proton radiation therapy. 20 head-and-neck patients with CT and CBCT images are included for training and testing. Our CBCT RSP results were evaluated with RSP maps created from CT images as the ground truth. Among all the 20 patients, the averaged mean absolute error between CT-based and CBCT-based RSP was 0.04±0.02, the averaged mean error was -0.01±0.03 and the averaged normalized correlation coefficient was 0.97±0.01. The proposed method provides sufficiently accurate RSP map generation from CBCT images, possibly allowing for CBCT-guided adaptive treatment planning for proton radiation therapy.

AB - Proton radiation therapy has shown highly conformal distribution of prescribed dose in target with outstanding normal tissue sparing stemming from its steep dose gradient at the distal end of the beam. However, the uncertainty in everyday patient setup can lead to a discrepancy between treatment dose distribution and the planning dose distribution. Cone-beam CT (CBCT) can be acquired daily before treatment to evaluate such inter-fraction setup error, while a further evaluation on resulted dose distribution error is currently not available. In this study, we developed a novel deep-learning based method to predict the relative stopping power maps from daily CBCT images to allow for online dose calculation in a step towards adaptive proton radiation therapy. 20 head-and-neck patients with CT and CBCT images are included for training and testing. Our CBCT RSP results were evaluated with RSP maps created from CT images as the ground truth. Among all the 20 patients, the averaged mean absolute error between CT-based and CBCT-based RSP was 0.04±0.02, the averaged mean error was -0.01±0.03 and the averaged normalized correlation coefficient was 0.97±0.01. The proposed method provides sufficiently accurate RSP map generation from CBCT images, possibly allowing for CBCT-guided adaptive treatment planning for proton radiation therapy.

KW - CBCT

KW - Machine learning

KW - Proton

KW - Stopping power

UR - https://www.scopus.com/pages/publications/85086722216

U2 - 10.1117/12.2549275

DO - 10.1117/12.2549275

M3 - Conference contribution

AN - SCOPUS:85086722216

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2020

A2 - Chen, Guang-Hong

A2 - Bosmans, Hilde

PB - SPIE

T2 - Medical Imaging 2020: Physics of Medical Imaging

Y2 - 16 February 2020 through 19 February 2020

ER -

Deep learning-based relative stopping power mapping generation with cone-beam CT in proton radiation therapy

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this